@inproceedings{pmlr-v227-behrendt24a, author = {F. Behrendt and D. Bhattacharya and J. Krüger and R. Opfer and A. Schlaefer}, title = {Patched Diffusion Models for Unsupervised Anomaly Detection in Brain MRI.}, year = {2024}, volume = {227.}, pages = {1019-1032}, month = {10-12 Jul}, editor = {In Oguz, Ipek and Noble, Jack and Li, Xiaoxiao and Styner, Martin and Baumgartner, Christian and Rusu, Mirabela and Heinmann, Tobias and Kontos, Despina and Landman, Bennett and Dawant, Benoit (Eds.)}, publisher = {PMLR:}, series = {Proceedings of Machine Learning Research}, booktitle = {Medical Imaging with Deep Learning}, url = {https://proceedings.mlr.press/v227/behrendt24a.html}, abstract = {The use of supervised deep learning techniques to detect pathologies in brain MRI scans can be challenging due to the diversity of brain anatomy and the need for annotated data sets. An alternative approach is to use unsupervised anomaly detection, which only requires sample-level labels of healthy brains to create a reference representation. This reference representation can then be compared to unhealthy brain anatomy in a pixel-wise manner to identify abnormalities. To accomplish this, generative models are needed to create anatomically consistent MRI scans of healthy brains. While recent diffusion models have shown promise in this task, accurately generating the complex structure of the human brain remains a challenge. In this paper, we propose a method that reformulates the generation task of diffusion models as a patch-based estimation of healthy brain anatomy, using spatial context to guide and improve reconstruction. We evaluate our approach on data of tumors and multiple sclerosis lesions and demonstrate a relative improvement of 25.1% compared to existing baselines.} } @inproceedings{Behrendt.2024Isbi, author = {F. Behrendt and D. Bhattacharya and L. Maack and J. Krüger and R. Opfer and R. Mieling and A. Schlaefer}, title = {Diffusion models with ensembled structure-based anomaly scoring for unsupervised anomaly detection.}, year = {2024}, pages = {Accepted}, booktitle = {IEEE International Symposion on Biomedical Imaging (ISBI)}, organization = {IEEE} } @inproceedings{behrendt2024combining, author = {F. Behrendt and D. Bhattacharya and L. Maack and J. Krüger and R. Opfer and A. Schlaefer}, title = {Combining Reconstruction-based Unsupervised Anomaly Detection with Supervised Segmentation for Brain {MRI}s.}, year = {2024}, pages = {Accepted}, note = {under review}, booktitle = {Submitted to Medical Imaging with Deep Learning}, url = {https://openreview.net/forum?id=iWfUcg4FrD}, abstract = {In contrast to supervised deep learning approaches, unsupervised anomaly detection (UAD) methods can be trained with healthy data only and do not require pixel-level annotations, enabling the identification of unseen pathologies. While this is promising for clinical screening tasks, reconstruction-based UAD methods fall short in segmentation accuracy compared to supervised models. Therefore, self-supervised UAD approaches have been proposed to improve segmentation accuracy. Typically, synthetic anomalies are used to train a segmentation network in a supervised fashion. However, this approach does not effectively generalize to real pathologies. We propose a framework combining reconstruction-based and self-supervised UAD methods to improve both segmentation performance for known anomalies and generalization to unknown pathologies. The framework includes an unsupervised diffusion model trained on healthy data to produce pseudo-healthy reconstructions and a supervised Unet trained to delineate anomalies from deviations between input- reconstruction pairs. Besides the effective use of synthetic training data, this framework allows for weakly-supervised training with small annotated data sets, generalizing to unseen pathologies. Our results show that with our approach, utilizing annotated data sets during training can substantially improve the segmentation performance for in-domain data while maintaining the generalizability of reconstruction-based approaches to pathologies unseen during training.} } @conference{, author = {F. Behrendt and S. Sonawane and D. Bhattacharya and L. Maack and J. Krüger and R. Opfer and A. Schlaefer}, title = {Quantitative evaluation of activation maps for weakly-supervised lung nodule segmentation.}, journal = {Medical Imaging 2024: Computer-Aided Diagnosis SPIE.}, year = {2024}, pages = {Accepted}, booktitle = {Medical Imaging 2024: Computer-Aided Diagnosis SPIE} } @inproceedings{neidhardt20240507, author = {M. Neidhardt and R. Mieling and S. Latus and M. Fischer and T. Maurer and A. Schlaefer}, title = {A Modified da Vinci Surgical Instrument for OCE based Elasticity Estimation with Deep Learning.}, year = {2024}, pages = {Accepted for presentation}, booktitle = {Proceedings of the 10th IEEE RAS/EMBS International Conference on Biomedical Robotics and Biomechatronics (BioRob 1024)}, doi = {10.48550/arXiv.2403.09256}, url = {https://doi.org/10.48550/arXiv.2403.09256}, abstract = {Robot-assisted surgery has advantages compared to conventional laparoscopic procedures, e.g., precise movement of the surgical instruments, improved dexterity, and high-resolution visualization of the surgical field. However, mechanical tissue properties may provide additional information, e.g., on the location of lesions or vessels. While elastographic imaging has been proposed, it is not readily available as an online modality during robot-assisted surgery. We propose modifying a da~Vinci surgical instrument to realize optical coherence elastography (OCE) for quantitative elasticity estimation. The modified da~Vinci instrument is equipped with piezoelectric elements for shear wave excitation and we employ fast optical coherence tomography (OCT) imaging to track propagating wave fields, which are directly related to biomechanical tissue properties. All high-voltage components are mounted at the proximal end outside the patient. We demonstrate that external excitation at the instrument shaft can effectively stimulate shear waves, even when considering damping. Comparing conventional and deep learning-based signal processing, resulting in mean absolute errors of 19.27 kPa and 6.29 kPa, respectively. These results illustrate that precise quantitative elasticity estimates can be obtained. We also demonstrate quantitative elasticity estimation on ex-vivo tissue samples of heart, liver and stomach, and show that the measurements can be used to distinguish soft and stiff tissue types.} } @inproceedings{neidhardt2023vrhtr, author = {M. Neidhardt and S. Latus and L. Maack and S. Gerlach and F. von Brackel and B. Busse and A. Schlaefer}, title = {VR-based Body Tracking for Homecare Training.}, year = {2024}, pages = {Accepted for presentation}, booktitle = {Proceedings of the Smart Healthy Environments Converence International Conference 2024} } @article{https://doi.org/10.1002/mp.16804, author = {S. Gerlach and F.-A. Siebert and A. Schlaefer}, title = {Robust stochastic optimization of needle configurations for robotic HDR prostate brachytherapy.}, journal = {Medical Physics.}, year = {2024}, volume = {51.}, number = {(1),}, pages = {464-475}, doi = {https://doi.org/10.1002/mp.16804}, url = {https://aapm.onlinelibrary.wiley.com/doi/abs/10.1002/mp.16804}, keywords = {inverse optimization, robotic brachytherapy, robust optimization}, abstract = {Abstract Background Ideally, inverse planning for HDR brachytherapy (BT) should include the pose of the needles which define the trajectory of the source. This would be particularly interesting when considering the additional freedom and accuracy in needle pose which robotic needle placement enables. However, needle insertion typically leads to tissue deformation, resulting in uncertainty regarding the actual pose of the needles with respect to the tissue. Purpose To efficiently address uncertainty during inverse planning for HDR BT in order to robustly optimize the pose of the needles before insertion, that is, to facilitate path planning for robotic needle placement. Methods We use a form of stochastic linear programming to model the inverse treatment planning problem. To account for uncertainty, we consider random tissue displacements at the needle tip to simulate tissue deformation. Conventionally for stochastic linear programming, each simulated deformation is reflected by an addition to the linear programming problem which increases problem size and computational complexity substantially and leads to impractical runtime. We propose two efficient approaches for stochastic linear programming. First, we consider averaging dose coefficients to reduce the problem size. Second, we study weighting of the slack variables of an adjusted linear problem to approximate the full stochastic linear program. We compare different approaches to optimize the needle configurations and evaluate their robustness with respect to different amounts of tissue deformation. Results Our results illustrate that stochastic planning can improve the robustness of the treatment with respect to deformation. The proposed approaches approximating stochastic linear programming better conform to the tissue deformation compared to conventional linear programming. They show good correlation with the plans computed after deformation while reducing the runtime by two orders of magnitude compared to the complete stochastic linear program. Robust optimization of needle configurations takes on average 59.42 s. Skew needle configurations lead to mean coverage improvements compared to parallel needles from 0.39 to 2.94 percentage points, when 8 mm tissue deformation is considered. Considering tissue deformations from 4  to 10 mm during planning with weighted stochastic optimization and skew needles generally results in improved mean coverage from 1.77 to 4.21 percentage points. Conclusions We show that efficient stochastic optimization allows selecting needle configurations which are more robust with respect to potentially negative effects of target deformation and displacement on the achievable prescription dose coverage. The approach facilitates robust path planning for robotic needle placement.} } @conference{, author = {S. Latus and M. Kulas and J. Sprenger and D. Bhattacharya and P. C. Breda and L. Wittig and T. Eixmann and G. Hütmann and L. Maack and D. Eggert and C. Betz and A. Schlaefer}, title = {Motion-compensated OCT imaging of laryngeal tissue..}, year = {2024}, pages = {Accepted}, booktitle = {Medical Imaging 2024: Image-Guided Procedures, Robotic Interventions, and Modeling SPIE.}, abstract = {The increasing incidence of laryngeal carcinomas requires approaches for early diagnosis and treatment. In clinical practice, white light endoscopy of the laryngeal region is typically followed by biopsy under general anesthesia. Thus, image based diagnosis using optical coherence tomography (OCT) has been proposed to study sub-surface tissue layers at high resolution. However, accessing the region of interest requires robust miniature OCT probes that can be forwarded through the working channel of a laryngoscope. Typically, such probes generate A-scans, i.e., single column depth images, which are rather difficult to interpret. We propose a novel approach using the endoscopic camera images to spatially align these A-scans. Given the natural tissue motion and movements of the laryngoscope, the resulting OCT images show a three-dimensional representation of the sub-surface structures, which is simpler to interpret. We present the overall imaging setup and the motion tracking method. Moreover, we describe an experimental setup to assess the precision of the spatial alignment. We study different tracking templates and report root-mean-squared errors of 0.08 mm and 0.18 mm for sinusoidal and freehand motion, respectively. Furthermore, we also demonstrate the in-vivo application of the approach, illustrating the benefit of spatially meaningful alignment of the A-scans to study laryngeal tissue.} } @phdthesis{Bengs_2023, author = {Bengs, Marcel}, title = {Spatio-temporal deep learning for medical image sequences.}, year = {2023}, doi = {10.15480/882.8891}, url = {https://hdl.handle.net/11420/44429}, type = {phdthesis}, abstract = {In dieser Arbeit untersuchen und präsentieren wir räumlich-zeitliche tiefe Lernverfahren für die Analyse medizinischer Bildsequenzen. Wir konzentrieren uns auf zwei Anwendungsszenarien, die Bewegungsanalyse und die dynamische Elastographie, unter Verwendung der optischen Kohärenztomographie und des Ultraschalls als Bildgebungsmodalitäten. Unsere Ergebnisse zeigen, dass Deep Learning für die End-to-End-Verarbeitung von Sequenzen medizinischer Bilddaten, einschließlich Sequenzen volumetrischer Bilder, effektiv genutzt werden kann.} } @article{https://doi.org/10.1002/mp.16436, author = {C. Stapper and S. Gerlach and T. Hofmann and C. Füweger and A. Schlaefer}, title = {Automated isocenter optimization approach for treatment planning for gyroscopic radiosurgery.}, journal = {Medical Physics.}, year = {2023}, volume = {50.}, number = {(8),}, pages = {5212-5221}, doi = {https://doi.org/10.1002/mp.16436}, url = {https://aapm.onlinelibrary.wiley.com/doi/abs/10.1002/mp.16436}, keywords = {gyroscopic radiosurgery, radiation therapy, treatment planning, ZAP-X}, abstract = {Abstract Background Radiosurgery is a well-established treatment for various intracranial tumors. In contrast to other established radiosurgery platforms, the new ZAP-X® allows for self-shielding gyroscopic radiosurgery. Here, treatment beams with variable beam-on times are targeted towards a small number of isocenters. The existing planning framework relies on a heuristic based on random selection or manual selection of isocenters, which often leads to a higher plan quality in clinical practice. Purpose The purpose of this work is to study an improved approach for radiosurgery treatment planning, which automatically selects the isocenter locations for the treatment of brain tumors and diseases in the head and neck area using the new system ZAP-X®. Methods We propose a new method to automatically obtain the locations of the isocenters, which are essential in gyroscopic radiosurgery treatment planning. First, an optimal treatment plan is created based on a randomly selected nonisocentric candidate beam set. The intersections of the resulting subset of weighted beams are then clustered to find isocenters. This approach is compared to sphere-packing, random selection, and selection by an expert planner for generating isocenters. We retrospectively evaluate plan quality on 10 acoustic neuroma cases. Results Isocenters acquired by the method of clustering result in clinically viable plans for all 10 test cases. When using the same number of isocenters, the clustering approach improves coverage on average by 31 percentage points compared to random selection, 15 percentage points compared to sphere packing and 2 percentage points compared to the coverage achieved with the expert selected isocenters. The automatic determination of location and number of isocenters leads, on average, to a coverage of 97 ± 3\% with a conformity index of 1.22 ± 0.22, while using 2.46 ± 3.60 fewer isocenters than manually selected. In terms of algorithm performance, all plans were calculated in less than 2 min with an average runtime of 75 ± 25 s. Conclusions This study demonstrates the feasibility of an automatic isocenter selection by clustering in the treatment planning process with the ZAP-X® system. Even in complex cases where the existing approaches fail to produce feasible plans, the clustering method generates plans that are comparable to those produced by expert selected isocenters. Therefore, our approach can help reduce the effort and time required for treatment planning in gyroscopic radiosurgery.} } @article{, author = {D. Bhattacharya and F. Behrendt and B. T. Becker and D. Beyersdorff and E. Petersen and M. Petersen and B. Cheng and D. Eggert and C. Betz and A. S. Hoffmann and A. Schlaefer}, title = {Multiple instance ensembling for paranasal anomaly classification in the maxillary sinus.}, journal = {International Journal of Computer Assisted Radiology and Surgery.}, year = {2023}, doi = {10.1007/s11548-023-02990-3}, url = {https://doi.org/10.1007/s11548-023-02990-3}, abstract = {Paranasal anomalies are commonly discovered during routine radiological screenings and can present with a wide range of morphological features. This diversity can make it difficult for convolutional neural networks (CNNs) to accurately classify these anomalies, especially when working with limited datasets. Additionally, current approaches to paranasal anomaly classification are constrained to identifying a single anomaly at a time. These challenges necessitate the need for further research and development in this area.} } @inproceedings{10.1117/12.2651525, author = {D. Bhattacharya and F. Behrendt and B. T. Becker and D. Beyersdorff and E. Petersen and M. Petersen and B. Cheng and D. Eggert and C. Betz and A. S. Hoffmann and A. Schlaefer}, title = {Unsupervised anomaly detection of paranasal anomalies in the maxillary sinus.}, year = {2023}, volume = {12465.}, pages = {124651B}, editor = {In Khan M. Iftekharuddin and Weijie Chen (Eds.)}, publisher = {SPIE:}, booktitle = {Medical Imaging 2023: Computer-Aided Diagnosis}, organization = {International Society for Optics and Photonics}, doi = {10.1117/12.2651525}, url = {https://doi.org/10.1117/12.2651525}, keywords = {Unsupervised Anomaly Detection, paranasal anomaly, anomaly detection, autoencoder, variational autoencoder, VAE}, abstract = {Deep learning (DL) algorithms can be used to automate paranasal anomaly detection from Magnetic Resonance Imaging (MRI). However, previous works relied on supervised learning techniques to distinguish between normal and abnormal samples. This method limits the type of anomalies that can be classified as the anomalies need to be present in the training data. Further, many data points from normal and anomaly class are needed for the model to achieve satisfactory classification performance. However, experienced clinicians can segregate between normal samples (healthy maxillary sinus) and anomalous samples (anomalous maxillary sinus) after looking at a few normal samples. We mimic the clinicians ability by learning the distribution of healthy maxillary sinuses using a 3D convolutional auto-encoder (cAE) and its variant, a 3D variational autoencoder (VAE) architecture and evaluate cAE and VAE for this task. Concretely, we pose the paranasal anomaly detection as an unsupervised anomaly detection problem. Thereby, we are able to reduce the labelling effort of the clinicians as we only use healthy samples during training. Additionally, we can classify any type of anomaly that differs from the training distribution. We train our 3D cAE and VAE to learn a latent representation of healthy maxillary sinus volumes using L1 reconstruction loss. During inference, we use the reconstruction error to classify between normal and anomalous maxillary sinuses. We extract sub-volumes from larger head and neck MRIs and analyse the effect of different fields of view on the detection performance. Finally, we report which anomalies are easiest and hardest to classify using our approach. Our results demonstrate the feasibility of unsupervised detection of paranasal anomalies from MRIs with an AUPRC of 85% and 80% for cAE and VAE, respectively.} } @conference{bhattacharya2023tissue, author = {D. Bhattacharya and S. Latus and F. Behrendt and F. Thimm and D. Eggert and C. Betz and A. Schlaefer}, title = {Tissue Classification During Needle Insertion Using Self-Supervised Contrastive Learning and Optical Coherence Tomography.}, year = {2023}, pages = {Accepted}, url = {https://doi.org/10.48550/arXiv.2304.13574}, abstract = {Needle positioning is essential for various medical applications such as epidural anaesthesia. Physicians rely on their instincts while navigating the needle in epidural spaces. Thereby, identifying the tissue structures may be helpful to the physician as they can provide additional feedback in the needle insertion process. To this end, we propose a deep neural network that classifies the tissues from the phase and intensity data of complex OCT signals acquired at the needle tip. We investigate the performance of the deep neural network in a limited labelled dataset scenario and propose a novel contrastive pretraining strategy that learns invariant representation for phase and intensity data. We show that with 10% of the training set, our proposed pretraining strategy helps the model achieve an F1 score of 0.84 whereas the model achieves an F1 score of 0.60 without it. Further, we analyse the importance of phase and intensity individually towards tissue classification} } @article{Eggert2023, author = {D. Eggert and D. Bhattacharya and A. Felicio-Briegel and V. Volgger and A. Schlaefer and C. Betz}, title = {Deep-learning-based image acquisition support tool for endoscopic narrow Band Imaging of the Larynx.}, year = {2023}, volume = {102.}, number = {(S 02),}, publisher = {Georg Thieme Verlag:}, doi = {10.1055/s-0043-1767104}, url = {http://www.thieme-connect.com/products/ejournals/abstract/10.1055/s-0043-1767104}, abstract = {Narrow band imaging (NBI) enables a contrast-enhanced imaging of mucosal blow-vessels. Nowadays NBI is a standard feature in many endoscopes. NBI is increasingly being applies in clinical investigations of the head-neck region. Using flexible laryngoscopes different laryngeal lesions can be investigated in awake patients. NBI enables a better recognition and differentiation of different pathologies than white light endoscopy.} } @article{Eggert2023a, author = {D. Eggert and D. Bhattacharya and A. Felicio-Briegel and V. Volgger and A. Schlaefer and C. Betz}, title = {Deep-Learning-basierte Aufnahmeunterstützung für endoskopisches Narrow Band Imaging des Larynx.}, year = {2023}, volume = {102.}, number = {(S 02),}, publisher = {Georg Thieme Verlag:}, doi = {10.1055/s-0043-1766512}, url = {http://www.thieme-connect.com/products/ejournals/abstract/10.1055/s-0043-1766512}, abstract = {Narrow Band Imaging (NBI) ermöglicht die kontrastverstärkte Darstellung von Blutgefäßen in Schleimhäuten. NBI gehört in vielen Endoskopen bereits zur Standardausstattung und findet im Kopf-Hals-Bereich immer stärkere Anwendung. Mit Hilfe von flexiblen Laryngoskopen können verschiedene Schleimhautläsionen des oberen Luft-Speiseweges im Wachzustand untersucht werden. Pathologien können dabei meist besser als bei konventioneller Weißlichtendoskopie erkannt werden. Für eine gute Beurteilbarkeit der NBI-Bilddaten ist eine gute Bildqualität essentiell. Nur wenn die oberflächlichen Blutgefäße klar erkennbar sind, können NBI-Aufnahmen sinnvoll ausgewertet werden.} } @inproceedings{10230753, author = {F. Behrendt and D. Bhattacharya and J. Krüger and R. Roland and A. Schlaefer}, title = {Nodule Detection in Chest Radiographs with Unsupervised Pre-Trained Detection Transformers.}, year = {2023}, pages = {1-4}, month = {April}, booktitle = {2023 IEEE 20th International Symposium on Biomedical Imaging (ISBI)}, doi = {10.1109/ISBI53787.2023.10230753}, abstract = {The detection of pulmonary nodules in chest x-rays is important for early observation and monitoring of lung cancer which is a major reason for death worldwide. However, detecting nodules from x-rays is challenging as nodules are easily overseen by radiologists. Convolutional neural networks (CNN) show promising results in supporting the clinical practice by automatic lung nodule detection and localization. Recently, attention-based Vision Transformers have been successfully applied in computer vision tasks. For object detection, end-to-end solutions have been proposed that reduce the amount of encoded prior knowledge and manual postprocessing. This is desirable particularly for medical applications where data domains often vary.In this work, we evaluate the application of Detection Transformers for nodule detection in chest x-rays and compare them against four CNN-based baseline object detection algorithms. To overcome the data inefficiency of Vision Transformers, we investigate the use of self-supervision from large-scale data sources. Our results demonstrate the high performance of transformer-based object detectors, by consistently outperforming CNN-based baselines on the Node21 data set. Furthermore, we demonstrate that self-supervision improves the detection performance without the costly requirement of collecting annotated data.} } @article{, author = {F. Behrendt and M. Bengs and D. Bhattacharya and J. Krüger and R. Opfer and A. Schlaefer}, title = {A systematic approach to deep learning-based nodule detection in chest radiographs.}, journal = {Scientific Reports.}, year = {2023}, volume = {13.}, number = {(1),}, pages = {10120}, doi = {10.1038/s41598-023-37270-2}, url = {https://doi.org/10.1038/s41598-023-37270-2}, abstract = {Lung cancer is a serious disease responsible for millions of deaths every year. Early stages of lung cancer can be manifested in pulmonary lung nodules. To assist radiologists in reducing the number of overseen nodules and to increase the detection accuracy in general, automatic detection algorithms have been proposed. Particularly, deep learning methods are promising. However, obtaining clinically relevant results remains challenging. While a variety of approaches have been proposed for general purpose object detection, these are typically evaluated on benchmark data sets. Achieving competitive performance for specific real-world problems like lung nodule detection typically requires careful analysis of the problem at hand and the selection and tuning of suitable deep learning models. We present a systematic comparison of state-of-the-art object detection algorithms for the task of lung nodule detection. In this regard, we address the critical aspect of class imbalance and and demonstrate a data augmentation approach as well as transfer learning to boost performance. We illustrate how this analysis and a combination of multiple architectures results in state-of-the-art performance for lung nodule detection, which is demonstrated by the proposed model winning the detection track of the Node21 competition. The code for our approach is available at https://github.com/FinnBehrendt/node21-submit.} } @misc{behrendt2023patched, author = {Finn Behrendt and Debayan Bhattacharya and Julia Krüger and Roland Opfer and Alexander Schlaefer}, title = {Patched Diffusion Models for Unsupervised Anomaly Detection in Brain MRI.}, year = {2023}, abstract = {The use of supervised deep learning techniques to detect pathologies in brain MRI scans can be challenging due to the diversity of brain anatomy and the need for annotated data sets. An alternative approach is to use unsupervised anomaly detection, which only requires sample-level labels of healthy brains to create a reference representation. This reference representation can then be compared to unhealthy brain anatomy in a pixel-wise manner to identify abnormalities. To accomplish this, generative models are needed to create anatomically consistent MRI scans of healthy brains. While recent diffusion models have shown promise in this task, accurately generating the complex structure of the human brain remains a challenge. In this paper, we propose a method that reformulates the generation task of diffusion models as a patch-based estimation of healthy brain anatomy, using spatial context to guide and improve reconstruction. We evaluate our approach on data of tumors and multiple sclerosis lesions and demonstrate a relative improvement of 25.1% compared to existing baselines.} } @article{jimaging9090170, author = {I. Kniep and R. Mieling and M. Gerling and A. Schlaefer and A. Heinemann and B. Ondruschka}, title = {Bayesian Reconstruction Algorithms for Low-Dose Computed Tomography Are Not Yet Suitable in Clinical Context.}, journal = {Journal of Imaging.}, year = {2023}, volume = {9.}, number = {(9),}, doi = {10.3390/jimaging9090170}, url = {https://www.mdpi.com/2313-433X/9/9/170}, abstract = {Computed tomography (CT) is a widely used examination technique that usually requires a compromise between image quality and radiation exposure. Reconstruction algorithms aim to reduce radiation exposure while maintaining comparable image quality. Recently, unsupervised deep learning methods have been proposed for this purpose. In this study, a promising sparse-view reconstruction method (posterior temperature optimized Bayesian inverse model; POTOBIM) is tested for its clinical applicability. For this study, 17 whole-body CTs of deceased were performed. In addition to POTOBIM, reconstruction was performed using filtered back projection (FBP). An evaluation was conducted by simulating sinograms and comparing the reconstruction with the original CT slice for each case. A quantitative analysis was performed using peak signal-to-noise ratio (PSNR) and structural similarity index measure (SSIM). The quality was assessed visually using a modified Ludewig’s scale. In the qualitative evaluation, POTOBIM was rated worse than the reference images in most cases. A partially equivalent image quality could only be achieved with 80 projections per rotation. Quantitatively, POTOBIM does not seem to benefit from more than 60 projections. Although deep learning methods seem suitable to produce better image quality, the investigated algorithm (POTOBIM) is not yet suitable for clinical routine.} } @article{10083305, author = {M. Bengs and J. Sprenger and S. Gerlach and M. Neidhardt and A. Schlaefer}, title = {Real-Time Motion Analysis With 4D Deep Learning for Ultrasound-Guided Radiotherapy.}, journal = {IEEE Transactions on Biomedical Engineering.}, year = {2023}, pages = {1-10}, doi = {10.1109/TBME.2023.3262422}, abstract = {Motion compensation in radiation therapy is a challenging scenario that requires estimating and forecasting motion of tissue structures to deliver the target dose. Ultrasound offers direct imaging of tissue in real-time and is considered for image guidance in radiation therapy. Recently, fast volumetric ultrasound has gained traction, but motion analysis with such high-dimensional data remains difficult. While deep learning could bring many advantages, such as fast data processing and high performance, it remains unclear how to process sequences of hundreds of image volumes efficiently and effectively. We present a 4D deep learning approach for real-time motion estimation and forecasting using long-term 4D ultrasound data. Using motion traces acquired during radiation therapy combined with various tissue types, our results demonstrate that long-term motion estimation can be performed markerless with a tracking error of $0.35\pm 0.2$ mm and with an inference time of less than 5 ms. Also, we demonstrate forecasting directly from the image data up to 900 ms into the future. Overall, our findings highlight that 4D deep learning is a promising approach for motion analysis during radiotherapy.} } @article{, author = {M. Neidhardt and R. Mieling and M. Bengs and A. Schlaefer}, title = {Optical force estimation for interactions between tool and soft tissues.}, journal = {Scientific Reports.}, year = {2023}, volume = {13.}, number = {(1),}, pages = {506}, doi = {10.1038/s41598-022-27036-7}, url = {https://doi.org/10.1038/s41598-022-27036-7}, abstract = {Robotic assistance in minimally invasive surgery offers numerous advantages for both patient and surgeon. However, the lack of force feedback in robotic surgery is a major limitation, and accurately estimating tool-tissue interaction forces remains a challenge. Image-based force estimation offers a promising solution without the need to integrate sensors into surgical tools. In this indirect approach, interaction forces are derived from the observed deformation, with learning-based methods improving accuracy and real-time capability. However, the relationship between deformation and force is determined by the stiffness of the tissue. Consequently, both deformation and local tissue properties must be observed for an approach applicable to heterogeneous tissue. In this work, we use optical coherence tomography, which can combine the detection of tissue deformation with shear wave elastography in a single modality. We present a multi-input deep learning network for processing of local elasticity estimates and volumetric image data. Our results demonstrate that accounting for elastic properties is critical for accurate image-based force estimation across different tissue types and properties. Joint processing of local elasticity information yields the best performance throughout our phantom study. Furthermore, we test our approach on soft tissue samples that were not present during training and show that generalization to other tissue properties is possible.} } @inproceedings{neidhardt2023vrbased, author = {M. Neidhardt and S. Gerlach F. N. Schmidt and I. A. K. Fiedler and S. Grube and B. Busse and A. Schlaefer}, title = {VR-based body tracking to stimulate musculoskeletal training.}, year = {2023}, pages = {Inprint}, booktitle = {CURAC 2023 Tagungsband}, url = {https://arxiv.org/abs/2308.03375}, abstract = {Training helps to maintain and improve sufficient muscle function, body control, and body coordination. These are important to reduce the risk of fracture incidents caused by falls, especially for the elderly or people recovering from injury. Virtual reality training can offer a cost-effective and individualized training experience. We present an application for the HoloLens 2 to enable musculoskeletal training for elderly and impaired persons to allow for autonomous training and automatic progress evaluation. We designed a virtual downhill skiing scenario that is controlled by body movement to stimulate balance and body control. By adapting the parameters of the ski slope, we can tailor the intensity of the training to individual users. In this work, we evaluate whether the movement data of the HoloLens 2 alone is sufficient to control and predict body movement and joint angles during musculoskeletal training. We record the movements of 10 healthy volunteers with external tracking cameras and track a set of body and joint angles of the participant during training. We estimate correlation coefficients and systematically analyze whether whole body movement can be derived from the movement data of the HoloLens 2. No participant reports movement sickness effects and all were able to quickly interact and control their movement during skiing. Our results show a high correlation between HoloLens 2 movement data and the external tracking of the upper body movement and joint angles of the lower limbs.} } @proceedings{neidhardt2023vrbased, author = {M. Neidhardt and S. Gerlach F. N. Schmidt and I. A. K. Fiedler and S. Grube and B. Busse and A. Schlaefer}, title = {VR-based body tracking to stimulate musculoskeletal training.}, year = {2023} } @article{, author = {M. Stender and J. Ohlsen and H. Geisler and A. Chabchoub and N. Hoffmann and A. Schlaefer}, title = {Up-Net: a generic deep learning-based time stepper for parameterized spatio-temporal dynamics.}, journal = {Computational Mechanics.}, year = {2023}, volume = {71.}, number = {(6),}, pages = {1227-1249}, doi = {10.1007/s00466-023-02295-x}, url = {https://doi.org/10.1007/s00466-023-02295-x}, abstract = {In the age of big data availability, data-driven techniques have been proposed recently to compute the time evolution of spatio-temporal dynamics. Depending on the required a priori knowledge about the underlying processes, a spectrum of black-box end-to-end learning approaches, physics-informed neural networks, and data-informed discrepancy modeling approaches can be identified. In this work, we propose a purely data-driven approach that uses fully convolutional neural networks to learn spatio-temporal dynamics directly from parameterized datasets of linear spatio-temporal processes. The parameterization allows for data fusion of field quantities, domain shapes, and boundary conditions in the proposed U$$^p$$-Net architecture. Multi-domain U$$^p$$-Net models, therefore, can generalize to different scenes, initial conditions, domain shapes, and domain sizes without requiring re-training or physical priors. Numerical experiments conducted on a universal and two-dimensional wave equation and the transient heat equation for validation purposes show that the proposed U$$^p$$-Net outperforms classical U-Net and conventional encoder-decoder architectures of the same complexity. Owing to the scene parameterization, the U$$^p$$-Net models learn to predict refraction and reflections arising from domain inhomogeneities and boundaries. Generalization properties of the model outside the physical training parameter distributions and for unseen domain shapes are analyzed. The deep learning flow map models are employed for long-term predictions in a recursive time-stepping scheme, indicating the potential for data-driven forecasting tasks. This work is accompanied by an open-sourced code.} } @inproceedings{10161377, author = {R. Mieling and M. Neidhardt and S. Latus and C. Stapper and S. Gerlach and I. Kniep and A. Heinemann and B. Ondruschka and A. Schlaefer}, title = {Collaborative Robotic Biopsy with Trajectory Guidance and Needle Tip Force Feedback.}, year = {2023}, pages = {6893-6900}, month = {May}, booktitle = {2023 IEEE International Conference on Robotics and Automation (ICRA)}, doi = {10.1109/ICRA48891.2023.10161377}, abstract = {The diagnostic value of biopsies is highly dependent on the placement of needles. Robotic trajectory guidance has been shown to improve needle positioning, but feedback for real-time navigation is limited. Haptic display of needle tip forces can provide rich feedback for needle navigation by enabling localization of tissue structures along the insertion path. We present a collaborative robotic biopsy system that combines trajectory guidance with kinesthetic feedback to assist the physician in needle placement. The robot aligns the needle while the insertion is performed in collaboration with a medical expert who controls the needle position on site. We present a needle design that senses forces at the needle tip based on optical coherence tomography and machine learning for real-time data processing. Our robotic setup allows operators to sense deep tissue interfaces independent of frictional forces to improve needle placement relative to a desired target structure. We first evaluate needle tip force sensing in ex-vivo tissue in a phantom study. We characterize the tip forces during insertions with constant velocity and demonstrate the ability to detect tissue interfaces in a collaborative user study. Participants are able to detect 91 percent of ex-vivo tissue interfaces based on needle tip force feedback alone. Finally, we demonstrate that even smaller, deep target structures can be accurately sampled by performing post-mortem in situ biopsies of the pancreas.} } @inproceedings{10.1007/978-3-031-43996-4_58, author = {R. Mieling and S. Latus and M. Fischer and F. Behrendt and A. Schlaefer}, title = {Optical Coherence Elastography Needle for Biomechanical Characterization of Deep Tissue.}, year = {2023}, pages = {607-617}, editor = {In Greenspan, Hayit and Madabhushi, Anant and Mousavi, Parvin and Salcudean, Septimiu and Duncan, James and Syeda-Mahmood, Tanveer and Taylor, Russell (Eds.)}, publisher = {Springer Nature Switzerland:}, address = {Cham}, isbn = {978-3-031-43996-4}, booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2023}, doi = {10.1007/978-3-031-43996-4_58}, abstract = {Compression-based optical coherence elastography (OCE) enables characterization of soft tissue by estimating elastic properties. However, previous probe designs have been limited to surface applications. We propose a bevel tip OCE needle probe for percutaneous insertions, where biomechanical characterization of deep tissue could enable precise needle placement, e.g., in prostate biopsy. We consider a dual-fiber OCE needle probe that provides estimates of local strain and load at the tip. Using a novel setup, we simulate deep tissue indentations where frictional forces and bulk sample displacement can affect biomechanical characterization. Performing surface and deep tissue indentation experiments, we compare our approach with external force and needle position measurements at the needle shaft. We consider two tissue mimicking materials simulating healthy and cancerous tissue and demonstrate that our probe can be inserted into deep tissue layers. Compared to surface indentations, external force-position measurements are strongly affected by frictional forces and bulk displacement and show a relative error of 49.2{\%} and 42.4{\%} for soft and stiff phantoms, respectively. In contrast, quantitative OCE measurements show a reduced relative error of 26.4{\%} and 4.9{\%} for deep indentations of soft and stiff phantoms, respectively. Finally, we demonstrate that the OCE measurements can be used to effectively discriminate the tissue mimicking phantoms.} } @article{, author = {S. A. Hoffmann and D. Bhattacharya and B. Becker and D. Beyersdorff and E. Petersen and M. Petersen and D. Eggert and A. Schläfer and C. Betz}, title = {Machbarkeitsanalyse eines automatisierten KI-basierten Klassifikationssystems zur Erkennung von Kieferhöhlenbefunden.}, journal = {Laryngo-Rhino-Otologie.}, year = {2023}, volume = {102.}, number = {(S 02),}, publisher = {Georg Thieme Verlag:}, doi = {10.1055/s-0043-1766502}, url = {http://www.thieme-connect.com/products/ejournals/abstract/10.1055/s-0043-1766502}, abstract = {Studien zeigen eine erhöhte Inzidenz von Verschattungen der Nasennebenhöhlen im cMRT ohne entsprechende Symptomatik. Dabei ist es von Interesse, ob abklärungsbedürftigte Befunde vorliegen. Der Einsatz von KI-basierten Methoden kann die Erkennung von Verschattungen automatisieren und dadurch die Arbeitsbelastung von ärzten reduzieren. In dieser Arbeit wurde eine Methode zur KI-basierten Klassifikation von Kieferhöhlenverschattungen entwickelt.} } @article{Hoffmann2023, author = {S. A. Hoffmann and D. Bhattacharya and B. Becker and D. Beyersdorff and E. Petersen and M. Petersen and D. Eggert and A. Schläfer and C. Betz}, title = {Analysing the feasibility of an automated AI-based classifier for detecting paranasal anomalies in the maxillary sinus.}, year = {2023}, volume = {102.}, number = {(S 02),}, publisher = {Georg Thieme Verlag:}, doi = {10.1055/s-0043-1767093}, url = {http://www.thieme-connect.com/products/ejournals/abstract/10.1055/s-0043-1767093}, abstract = {Large scale population studies have been performed to analyse the rate of finding sinus opacities in cranial MRIs. It is of interest whether there are findings requiring clarification. Using AI-based methods can automate the detection of the sinus opacities and reduce the workload of clinicians. In this work, a method for AI-based classification was developed in order to automatically recognise paranasal sinus opacities.} } @article{https://doi.org/10.1002/mp.16381, author = {S. Gerlach, T. Hofmann, C. Fürweger, A. Schlaefer}, title = {Towards fast adaptive replanning by constrained reoptimization for intra-fractional non-periodic motion during robotic SBRT.}, journal = {Medical Physics.}, year = {2023}, volume = {50.}, number = {(7),}, pages = {4613-4622}, doi = {https://doi.org/10.1002/mp.16381}, url = {https://aapm.onlinelibrary.wiley.com/doi/abs/10.1002/mp.16381}, keywords = {intra-fractional adaption, robotic radiation therapy, treatment planning}, abstract = {Abstract Background Periodic and slow target motion is tracked by synchronous motion of the treatment beams in robotic stereotactic body radiation therapy (SBRT). However, spontaneous, non-periodic displacement or drift of the target may completely change the treatment geometry. Simple motion compensation is not sufficient to guarantee the best possible treatment, since relative motion between the target and organs at risk (OARs) can cause substantial deviations of dose in the OARs. This is especially evident when considering the temporally heterogeneous dose delivery by many focused beams which is typical for robotic SBRT. Instead, a reoptimization of the remaining treatment plan after a large target motion during the treatment could potentially reduce the actually delivered dose to OARs and improve target coverage. This reoptimization task, however, is challenging due to time constraints and limited human supervision. Purpose To study the detrimental effect of spontaneous target motion relative to surrounding OARs on the delivered dose distribution and to analyze how intra-fractional constrained replanning could improve motion compensated robotic SBRT of the prostate. Methods We solve the inverse planning problem by optimizing a linear program. When considering intra-fractional target motion resulting in a change of geometry, we adapt the linear program to account for the changed dose coefficients and delivered dose. We reduce the problem size by only reweighting beams from the reference treatment plan without motion. For evaluation we simulate target motion and compare our approach for intra-fractional replanning to the conventional compensation by synchronous beam motion. Results are generated retrospectively on data of 50 patients. Results Our results show that reoptimization can on average retain or improve coverage in case of target motion compared to the reference plan without motion. Compared to the conventional compensation, coverage is improved from 87.83 \% to 94.81 \% for large target motion. Our approach for reoptimization ensures fixed upper constraints on the dose even after motion, enabling safer intra-fraction adaption, compared to conventional motion compensation where overdosage in OARs can lead to 21.79 \% higher maximum dose than planned. With an average reoptimization time of 6 s for 200 reoptimized beams our approach shows promising performance for intra-fractional application. Conclusions We show that intra-fractional constrained reoptimization for adaption to target motion can improve coverage compared to the conventional approach of beam translation while ensuring that upper dose constraints on VOIs are not violated.} } @inproceedings{10.1117/12.2653833, author = {S. Grube and M. Bengs and M. Neidhardt and S. Latus and A. Schlaefer}, title = {Ultrasound shear wave velocity estimation in a small field of view via spatio-temporal deep learning.}, year = {2023}, volume = {12464.}, pages = {1246425}, editor = {In Olivier Colliot and Ivana Išgum (Eds.)}, publisher = {SPIE:}, booktitle = {Medical Imaging 2023: Image Processing}, organization = {International Society for Optics and Photonics}, doi = {10.1117/12.2653833}, url = {https://doi.org/10.1117/12.2653833}, keywords = {Ultrasound, Shear Wave Elastography, High-Frequency US imaging, Field of View, Deep Learning, Tissue Elasticity}, abstract = {A change in tissue stiffness can indicate pathological diseases and therefore supports physicians in diagnosis and treatment. Ultrasound shear wave elastography (US-SWEI) can be used to quantify tissue stiffness by estimating the velocity of propagating shear waves. While a linear US probe with a lateral imaging width of approximately 40 mm is commonly used and US-SWEI has been successfully demonstrated, some clinical applications, such as laparoscopic or endoscopic interventions, require small probes. This limits the lateral image width to the millimeter range and reduces the available information in the US images substantially. In this work, we systematically analyze the effect of a reduced lateral imaging width for shear wave velocity estimation using the conventional time-of-flight (ToF) method and spatio-temporal convolutional neural networks (ST-CNNs). For our study, we use tissue mimicking gelatin phantoms with varying stiffness and resulting shear wave velocities in the range from 3.63 m/s to 7.09 m/s. We find that lateral imaging width has a substantial impact on the performance of ToF, while shear wave velocity estimation with ST-CNNs remains robust. Our results show that shear wave velocity estimation with ST-CNN can even be performed for a lateral imaging width of 2.1 mm resulting in a mean absolute error of 0.81 ± 0.61 m/s.} } @article{article, author = {S. Kolibová and E. Wölfel and H. Hemmatian and P. Milovanovic and H. Mushumba and B. Wulff and M. Neidhardt and K. Püschel and A. Failla and A. Vlug and A. Schlaefer and B. Ondruschka and M. Amling and L. Hofbauer and M. Rauner and B. Busse and K. Jähn-Rickert}, title = {Osteocyte apoptosis and cellular micropetrosis signify skeletal aging in type 1 diabetes.}, journal = {Acta Biomaterialia.}, year = {2023}, month = {03}, pmid = {36878337}, doi = {10.1016/j.actbio.2023.02.037}, abstract = {Bone fragility is a profound complication of type 1 diabetes mellitus (T1DM), increasing patient morbidity. Within the mineralized bone matrix, osteocytes build a mechanosensitive network that orchestrates bone remodeling; thus, osteocyte viability is crucial for maintaining bone homeostasis. In human cortical bone specimens from individuals with T1DM, we found signs of accelerated osteocyte apoptosis and local mineralization of osteocyte lacunae (micropetrosis) compared with samples from age-matched controls. Such morphological changes were seen in the relatively young osteonal bone matrix on the periosteal side, and micropetrosis coincided with microdamage accumulation, implying that T1DM drives local skeletal aging and thereby impairs the biomechanical competence of the bone tissue. The consequent dysfunction of the osteocyte network hampers bone remodeling and decreases bone repair mechanisms, potentially contributing to the enhanced fracture risk seen in individuals with T1DM. STATEMENT OF SIGNIFICANCE: Type 1 diabetes mellitus (T1DM) is a chronic autoimmune disease that causes hyperglycemia. Increased bone fragility is one of the complications associated with T1DM. Our latest study on T1DM-affected human cortical bone identified the viability of osteocytes, the primary bone cells, as a potentially critical factor in T1DM-bone disease. We linked T1DM with increased osteocyte apoptosis and local accumulation of mineralized lacunar spaces and microdamage. Such structural changes in bone tissue suggest that T1DM speeds up the adverse effects of aging, leading to the premature death of osteocytes and potentially contributing to diabetes-related bone fragility.} } @article{10122996, author = {S. Latus and S. Grube and T. Eixmann and M. Neidhardt and S. Gerlach and R. Mieling and G. Hüttmann and M. Lutz and A. Schlaefer}, title = {A Miniature Dual-Fiber Probe for Quantitative Optical Coherence Elastography.}, journal = {IEEE Transactions on Biomedical Engineering.}, year = {2023}, volume = {70.}, number = {(11),}, pages = {3064-3072}, month = {Nov}, doi = {10.1109/TBME.2023.3275539}, abstract = {Objective: Optical coherence elastography (OCE) allows for high resolution analysis of elastic tissue properties. However, due to the limited penetration of light into tissue, miniature probes are required to reach structures inside the body, e.g., vessel walls. Shear wave elastography relates shear wave velocities to quantitative estimates of elasticity. Generally, this is achieved by measuring the runtime of waves between two or multiple points. For miniature probes, optical fibers have been integrated and the runtime between the point of excitation and a single measurement point has been considered. This approach requires precise temporal synchronization and spatial calibration between excitation and imaging. Methods: We present a miniaturized dual-fiber OCE probe of $1 \,\mathrm{m}\mathrm{m}$ diameter allowing for robust shear wave elastography. Shear wave velocity is estimated between two optics and hence independent of wave propagation between excitation and imaging. We quantify the wave propagation by evaluating either a single or two measurement points. Particularly, we compare both approaches to ultrasound elastography. Results: Our experimental results demonstrate that quantification of local tissue elasticities is feasible. For homogeneous soft tissue phantoms, we obtain mean deviations of $0.15 \,\mathrm{m}\mathrm{s}^{-1}$ and $0.02 \,\mathrm{m}\mathrm{s}^{-1}$ for single-fiber and dual-fiber OCE, respectively. In inhomogeneous phantoms, we measure mean deviations of up to $0.54 \,\mathrm{m}\mathrm{s}^{-1}$ and $0.03 \,\mathrm{m}\mathrm{s}^{-1}$ for single-fiber and dual-fiber OCE, respectively. Conclusion: We present a dual-fiber OCE approach that is much more robust in inhomogeneous tissues. Moreover, we demonstrate the feasibility of elasticity quantification in ex-vivo coronary arteries. Significance: This study introduces an approach for robust elasticity quantification from within the tissue.} } @inproceedings{10.1007/978-3-031-16437-8_41, author = {D. Bhattacharya and B. T. Becker and F. Behrendt and M. Bengs and D. Beyersdorff and D. Eggert and E. Petersen and F. Jansen and M. Petersen and B. Cheng and C. Betz and A. Schlaefer and A. S. Hoffmann}, title = {Supervised Contrastive Learning to Classify Paranasal Anomalies in the Maxillary Sinus.}, year = {2022}, pages = {429-438}, editor = {In Wang, Linwei and Dou, Qi and Fletcher, P. Thomas and Speidel, Stefanie and Li, Shuo (Eds.)}, publisher = {Springer Nature Switzerland:}, address = {Cham}, isbn = {978-3-031-16437-8}, booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2022}, doi = {10.1007/978-3-031-16437-8_41}, url = {https://link.springer.com/chapter/10.1007/978-3-031-16437-8_41}, abstract = {Using deep learning techniques, anomalies in the paranasal sinus system can be detected automatically in MRI images and can be further analyzed and classified based on their volume, shape and other parameters like local contrast. However due to limited training data, traditional supervised learning methods often fail to generalize. Existing deep learning methods in paranasal anomaly classification have been used to diagnose at most one anomaly. In our work, we consider three anomalies. Specifically, we employ a 3D CNN to separate maxillary sinus volumes without anomaly from maxillary sinus volumes with anomaly. To learn robust representations from a small labelled dataset, we propose a novel learning paradigm that combines contrastive loss and cross-entropy loss. Particularly, we use a supervised contrastive loss that encourages embeddings of maxillary sinus volumes with and without anomaly to form two distinct clusters while the cross-entropy loss encourages the 3D CNN to maintain its discriminative ability. We report that optimising with both losses is advantageous over optimising with only one loss. We also find that our training strategy leads to label efficiency. With our method, a 3D CNN classifier achieves an AUROC of 0.85 {\textpm} 0.03 while a 3D CNN classifier optimised with cross-entropy loss achieves an AUROC of 0.66 {\textpm} 0.1. Our source code is available at https://github.com/dawnofthedebayan/SupConCE{\_}MICCAI{\_}22.} } @article{ima.22795, author = {D. Bhattacharya and D. Eggert and C. Betz and A. Schlaefer}, title = {Squeeze and multi-context attention for polyp segmentation.}, journal = {International Journal of Imaging Systems and Technology.}, year = {2022}, volume = {n/a.}, number = {(n/a),}, doi = {https://doi.org/10.1002/ima.22795}, url = {https://onlinelibrary.wiley.com/doi/abs/10.1002/ima.22795}, keywords = {attention, attention gate, polyp segmentation, squeeze and excite, squeeze and multi-context, U-Net}, abstract = {Artificial Intelligence-based Computer Aided Diagnostics (AI-CADx) have been proposed to help physicians in reducing misdetection of polyps in colonoscopy examination. The heterogeneity of a polyp\'s appearance makes detection challenging for physicians and AI-CADx. Towards building better AI-CADx, we propose an attention module called Squeeze and Multi-Context Attention (SMCA) that re-calibrates a feature map by providing channel and spatial attention by taking into consideration highly activated features and context of the features at multiple receptive fields simultaneously. We test the effectiveness of SMCA by incorporating it into the encoder of five popular segmentation models. We use five public datasets and construct intra-dataset and inter-dataset test sets to evaluate the generalizing capability of models with SMCA. Our intra-dataset evaluation shows that U-Net with SMCA and without SMCA has a precision of 0.86 +/- 0.01 and 0.76 +/- 0.02 respectively on CVC-ClinicDB. Our inter-dataset evaluation reveals that U-Net with SMCA and without SMCA has a precision of 0.62 +/- 0.01 and 0.55 +/- 0.09 respectively when trained on Kvasir-SEG and tested on CVC-ColonDB. Similar results are observed using other segmentation models and other public datasets. In conclusion, we demonstrate that incorporating SMCA into the segmentation models leads to an increase in generalizing capability of the segmentation models} } @inproceedings{bhattacharya2022learning, author = {D. Bhattacharya and F. Behrendt and A. Felicio-Briegel and V. Volgger and D. Eggert and C. Betz and A. Schlaefer}, title = {Learning Robust Representation for Laryngeal Cancer Classification in Vocal Folds from Narrow Band Images..}, year = {2022}, pages = {Accepted}, booktitle = {Medical Imaging with Deep Learning}, url = {https://openreview.net/forum?id=nJd70UxI5hH} } @article{Eggert2021, author = {D. Eggert and M. Bengs and S. Westermann and N. Gessert and A. O. H. Gerstner and N. A. Mueller and J. Bewarder and A. Schlaefer and C. Betz, and W. Laffers}, title = {In vivo detection of head and neck tumors by hyperspectral imaging combined with deep learning methods.}, journal = {Journal of Biophotonics.}, year = {2022}, volume = {15.}, number = {(3),}, pages = {e202100167}, doi = {https://doi.org/10.1002/jbio.202100167}, url = {https://onlinelibrary.wiley.com/doi/abs/10.1002/jbio.202100167}, keywords = {convolutional neural network, head and neck cancer, hyperspectral imaging, intraoperative imaging, optical biopsy}, abstract = {Abstract Currently, there are no fast and accurate screening methods available for head and neck cancer, the eighth most common tumor entity. For this study, we used hyperspectral imaging, an imaging technique for quantitative and objective surface analysis, combined with deep learning methods for automated tissue classification. As part of a prospective clinical observational study, hyperspectral datasets of laryngeal, hypopharyngeal and oropharyngeal mucosa were recorded in 98 patients before surgery in vivo. We established an automated data interpretation pathway that can classify the tissue into healthy and tumorous using convolutional neural networks with 2D spatial or 3D spatio-spectral convolutions combined with a state-of-the-art Densenet architecture. Using 24 patients for testing, our 3D spatio-spectral Densenet classification method achieves an average accuracy of 81\%, a sensitivity of 83\% and a specificity of 79\%.} } @article{BehrendtBhattacharyaKrügerOpferSchlaefer+2022+34+3, author = {F. Behrendt and D. Bhattacharya and J. Krüger and R. Opfer and A. Schlaefer}, title = {Data-Efficient Vision Transformers for Multi-Label Disease Classification on Chest Radiographs.}, journal = {Current Directions in Biomedical Engineering.}, year = {2022}, volume = {8.}, number = {(1),}, pages = {34--37}, doi = {doi:10.1515/cdbme-2022-0009}, url = {https://doi.org/10.1515/cdbme-2022-0009}, abstract = {Radiographs are a versatile diagnostic tool for thedetection and assessment of pathologies, for treatment plan-ning or for navigation and localization purposes in clinical in-terventions. However, their interpretation and assessment byradiologists can be tedious and error-prone. Thus, a wide va-riety of deep learning methods have been proposed to supportradiologists interpreting radiographs.Mostly, these approaches rely on convolutional neural net-works (CNN) to extract features from images. Especially forthe multi-label classification of pathologies on chest radio-graphs (Chest X-Rays, CXR), CNNs have proven to be wellsuited. On the Contrary, Vision Transformers (ViTs) have notbeen applied to this task despite their high classification per-formance on generic images and interpretable local saliencymaps which could add value to clinical interventions. ViTs donot rely on convolutions but on patch-based self-attention andin contrast to CNNs, no prior knowledge of local connectivityis present. While this leads to increased capacity, ViTs typi-cally require an excessive amount of training data which rep-resents a hurdle in the medical domain as high costs are asso-ciated with collecting large medical data sets.In this work, we systematically compare the classification per-formance of ViTs and CNNs for different data set sizes andevaluate more data-efficient ViT variants (DeiT). Our resultsshow that while the performance between ViTs and CNNs ison par with a small benefit for ViTs, DeiTs outperform theformer if a reasonably large data set is available for training} } @inproceedings{behrendt2022, author = {F. Behrendt and M. Bengs and D. Bhattacharya and J. Krüger and R. Opfer and A. Schlaefer}, title = {Capturing Inter-Slice Dependencies of 3D Brain MRI-Scans for Unsupervised Anomaly Detection.}, year = {2022}, booktitle = {Medical Imaging with Deep Learning.}, url = {https://openreview.net/pdf?id=db8wDgKH4p4}, abstract = {The increasing workloads for radiologists in clinical practice lead to the need for an automatic support tool for anomaly detection in brain MRI-scans. While supervised learning methods can detect and localize lesions in brain MRI-scans, the need for large, balanced data sets with pixel-level annotations limits their use. In contrast, unsupervised anomaly detection (UAD) models only require healthy brain data for training. Despite the inherent 3D structure of brain MRI-scans, most UAD studies focus on slicewise processing. In this work, we capture the inter-slice dependencies of the human brain using recurrent neural networks (RNN) and transformer-based self-attention mechanisms together with variational autoencoders (VAE). We show that by this we can improve both reconstruction quality and UAD performance while the number of parameters remain similar to the 2D approach where the slices are processed individually.} } @inproceedings{9761443, author = {F. Behrendt and M. Bengs and F. Rogge and J. Krüger and R. Opfer and A. Schlaefer}, title = {Unsupervised Anomaly Detection in 3D Brain MRI Using Deep Learning with Impured Training Data.}, year = {2022}, pages = {1-4}, booktitle = {2022 IEEE 19th International Symposium on Biomedical Imaging (ISBI)}, doi = {10.1109/ISBI52829.2022.9761443}, abstract = {The detection of lesions in magnetic resonance imaging MRI-scans of human brains remains challenging, timeconsuming and error-prone. Recently, unsupervised anomaly detection UAD methods have shown promising results for this task. These methods rely on training data sets that solely contain healthy samples. Compared to supervised approaches, this significantly reduces the need for an extensive amount of labeled training data. However, data labelling remains error-prone. We study how unhealthy samples within the training data affect anomaly detection performance for brain MRI-scans. For our evaluations, we consider autoencoders AE as a well-established baseline method for UAD. We systematically evaluate the effect of impured training data by injecting different quantities of unhealthy samples to our training set of healthy samples. We evaluate a method to identify falsely labeled samples directly during training based on the reconstruction error of the AE. Our results show that training with impured data decreases the UAD performance notably even with few falsely labeled samples. By performing outlier removal directly during training based on the reconstruction-loss, we demonstrate that falsely labeled data can be detected and that this mitigates the effect of falsely labeled data.} } @article{, author = {Gerlach, Stefan and Schlaefer, Alexander}, title = {Robotic Systems in Radiotherapy and Radiosurgery.}, journal = {Current Robotics Reports.}, year = {2022}, doi = {10.1007/s43154-021-00072-3}, url = {https://doi.org/10.1007/s43154-021-00072-3}, abstract = {This review provides an overview of robotic systems in radiotherapy and radiosurgery, with a focus on medical devices and recently proposed research systems. We summarize the key motivation for using robotic systems and illustrate the potential advantages.} } @article{, author = {J. Sprenger and M. Bengs and S. Gerlach and M. Neidhardt and A. Schlaefer}, title = {Systematic analysis of volumetric ultrasound parameters for markerless 4D motion tracking.}, journal = {International Journal of Computer Assisted Radiology and Surgery.}, year = {2022}, doi = {10.1007/s11548-022-02665-5}, url = {https://doi.org/10.1007/s11548-022-02665-5}, abstract = {Motion compensation is an interesting approach to improve treatments of moving structures. For example, target motion can substantially affect dose delivery in radiation therapy, where methods to detect and mitigate the motion are widely used. Recent advances in fast, volumetric ultrasound have rekindled the interest in ultrasound for motion tracking. We present a setup to evaluate ultrasound based motion tracking and we study the effect of imaging rate and motion artifacts on its performance.} } @article{SprengerNeidhardtLatusGrubeFischerSchlaefer+2022+6, author = {J. Sprenger and M. Neidhardt and S. Latus and S. Grube and M. Fischer and A. Schlaefer}, title = {Surface Scanning for Navigation Using High-Speed Optical Coherence Tomography.}, journal = {Current Directions in Biomedical Engineering.}, year = {2022}, volume = {8.}, number = {(1),}, pages = {62-65}, doi = {doi:10.1515/cdbme-2022-0016}, url = {https://doi.org/10.1515/cdbme-2022-0016}, abstract = {Medical interventions are often guided by opti-cal tracking systems and optical coherence tomography hasshown promising results for markerless tracking of soft tissue.The high spatial resolution and subsurface information con-tain valuable information about the underlying tissue structureand tracking of certain target structures is in principle possible.However, the small field-of-view complicates the selection ofsuitable regions-of-interest for tracking. Therefore, we extendan experimental setup and perform volumetric surface scan-ning of target structures to enlarge the field-of-view. We showthat the setup allows for data acquisition and that precise merg-ing of the volumes is possible with mean absolute errors from0.041 mmto0.097 mm.} } @article{MaackHolsteinSchlaefer+2022+17+20, author = {L. Maack and L. Holstein and A. Schlaefer}, title = {GANs for generation of synthetic ultrasound images from small datasets.}, journal = {Current Directions in Biomedical Engineering.}, year = {2022}, volume = {8.}, number = {(1),}, pages = {17--20}, doi = {doi:10.1515/cdbme-2022-0005}, url = {https://doi.org/10.1515/cdbme-2022-0005}, abstract = {The task of medical image classification is increas-ingly supported by algorithms. Deep learning methods likeconvolutional neural networks (CNNs) show superior perfor-mance in medical image analysis but need a high-quality train-ing dataset with a large number of annotated samples. Partic-ularly in the medical domain, the availability of such datasetsis rare due to data privacy or the lack of data sharing practicesamong institutes. Generative adversarial networks (GANs) areable to generate high quality synthetic images. This work in-vestigates the capabilities of different state-of-the-art GAN ar-chitectures in generating realistic breast ultrasound images ifonly a small amount of training data is available. In a secondstep, these synthetic images are used to augment the real ul-trasound image dataset utilized for training CNNs. The train-ing of both GANs and CNNs is conducted with systemati-cally reduced dataset sizes. The GAN architectures are ca-pable of generating realistic ultrasound images. GANs usingdata augmentation techniques outperform the baseline Style-GAN2 with respect to the Fréchet Inception distance by upto64.2%. CNN models trained with additional synthetic dataoutperform the baseline CNN model using only real data fortraining by up to15.3%with respect to the F1 score, espe-cially for datasets containing less than 100 images. As a con-clusion, GANs can successfully be used to generate syntheticultrasound images of high quality and diversity, improve clas-sification performance of CNNs and thus provide a benefit tocomputer-aided diagnostics} } @inproceedings{10.1117/12.2608120, author = {M. Bengs and F. Behrendt and M.-H. Laves and J. Krüger and R. Opfer and A. Schlaefer}, title = {Unsupervised anomaly detection in 3D brain MRI using deep learning with multi-task brain age prediction.}, year = {2022}, volume = {12033.}, pages = {1203314}, editor = {In Karen Drukker and Khan M. Iftekharuddin and Hongbing Lu and Maciej A. Mazurowski and Chisako Muramatsu and Ravi K. Samala (Eds.)}, publisher = {SPIE:}, booktitle = {Medical Imaging 2022: Computer-Aided Diagnosis}, organization = {International Society for Optics and Photonics}, doi = {10.1117/12.2608120}, url = {https://doi.org/10.1117/12.2608120}, keywords = {Anomaly Detection, Brain Age Prediction, Unsupervised, Brain MRI, 3D Autoencoder}, abstract = {Lesion detection in brain Magnetic Resonance Images (MRIs) remains a challenging task. MRIs are typically read and interpreted by domain experts, which is a tedious and time-consuming process. Recently, unsupervised anomaly detection (UAD) in brain MRI with deep learning has shown promising results to provide a quick, initial assessment. So far, these methods only rely on the visual appearance of healthy brain anatomy for anomaly detection. Another biomarker for abnormal brain development is the deviation between the brain age and the chronological age, which is unexplored in combination with UAD. We propose deep learning for UAD in 3D brain MRI considering additional age information. We analyze the value of age information during training, as an additional anomaly score, and systematically study several architecture concepts. Based on our analysis, we propose a novel deep learning approach for UAD with multi-task age prediction. We use clinical T1-weighted MRIs of 1735 healthy subjects and the publicly available BraTs 2019 data set for our study. Our novel approach significantly improves UAD performance with an AUC of 92.60% compared to an AUC-score of 84.37% using previous approaches without age information.} } @article{LAVES2022102382, author = {M. H. Laves and M. Tölle and A. Schlaefer and S. Engelhardt}, title = {Posterior temperature optimized Bayesian models for inverse problems in medical imaging.}, journal = {Medical Image Analysis.}, year = {2022}, volume = {78.}, pages = {102382}, doi = {https://doi.org/10.1016/j.media.2022.102382}, url = {https://www.sciencedirect.com/science/article/pii/S1361841522000342}, keywords = {Variational inference, Hallucination, Deep learning}, abstract = {We present Posterior Temperature Optimized Bayesian Inverse Models (POTOBIM), an unsupervised Bayesian approach to inverse problems in medical imaging using mean-field variational inference with a fully tempered posterior. Bayesian methods exhibit useful properties for approaching inverse tasks, such as tomographic reconstruction or image denoising. A suitable prior distribution introduces regularization, which is needed to solve the ill-posed problem and reduces overfitting the data. In practice, however, this often results in a suboptimal posterior temperature, and the full potential of the Bayesian approach is not being exploited. In POTOBIM, we optimize both the parameters of the prior distribution and the posterior temperature with respect to reconstruction accuracy using Bayesian optimization with Gaussian process regression. Our method is extensively evaluated on four different inverse tasks on a variety of modalities with images from public data sets and we demonstrate that an optimized posterior temperature outperforms both non-Bayesian and Bayesian approaches without temperature optimization. The use of an optimized prior distribution and posterior temperature leads to improved accuracy and uncertainty estimation and we show that it is sufficient to find these hyperparameters per task domain. Well-tempered posteriors yield calibrated uncertainty, which increases the reliability in the predictions. Our source code is publicly available at github.com/Cardio-AI/mfvi-dip-mia.} } @article{9760236, author = {M. Neidhardt and M. Bengs and S. Latus and S. Gerlach and C. J. Cyron and J. Sprenger and A. Schlaefer}, title = {Ultrasound Shear Wave Elasticity Imaging with Spatio-Temporal Deep Learning.}, journal = {IEEE Transactions on Biomedical Engineering.}, year = {2022}, pages = {1-1}, doi = {10.1109/TBME.2022.3168566}, url = {https://arxiv.org/abs/2204.05745}, abstract = {Ultrasound shear wave elasticity imaging is a valuable tool for quantifying the elastic properties of tissue. Typically, the shear wave velocity is derived and mapped to an elasticity value, which neglects information such as the shape of the propagating shear wave or push sequence characteristics. We present 3D spatio-temporal CNNs for fast local elasticity estimation from ultrasound data. This approach is based on retrieving elastic properties from shear wave propagation within small local regions. A large training data set is acquired with a robot from homogeneous gelatin phantoms ranging from 17.42 kPa to 126.05 kPa with various push locations. The results show that our approach can estimate elastic properties on a pixelwise basis with a mean absolute error of 5.014.37 kPa. Furthermore, we estimate local elasticity independent of the push location and can even perform accurate estimates inside the push region. For phantoms with embedded inclusions, we report a 53.93% lower MAE (7.50 kPa) and on the background of 85.24% (1.64 kPa) compared to a conventional shear wave method. Overall, our method offers fast local estimations of elastic properties with small spatio-temporal window sizes.} } @article{9693976, author = {M. Neidhardt and S. Gerlach and R. Mieling and M.-H. Laves and T. Weiß, and M. Gromniak and A. Fitzek and D. Möbius and I. Kniep and A. Ron and J. Schädler and A. Heinemann K. and Püschel and B. Ondruschka and A. Schlaefer}, title = {Robotic Tissue Sampling for Safe Post-Mortem Biopsy in Infectious Corpses.}, journal = {IEEE Transactions on Medical Robotics and Bionics.}, year = {2022}, volume = {4.}, number = {(1),}, pages = {94-105}, month = {Feb}, doi = {10.1109/TMRB.2022.3146440}, url = {https://arxiv.org/abs/2201.12168}, abstract = {In pathology and legal medicine, the histopathological and microbiological analysis of tissue samples from infected deceased is a valuable information for developing treatment strategies during a pandemic such as COVID-19. However, a conventional autopsy carries the risk of disease transmission and may be rejected by relatives. We propose minimally invasive biopsy with robot assistance under CT guidance to minimize the risk of disease transmission during tissue sampling and to improve accuracy. A flexible robotic system for biopsy sampling is presented, which is applied to human corpses placed inside protective body bags. An automatic planning and decision system estimates optimal insertion point. Heat maps projected onto the segmented skin visualize the distance and angle of insertions and estimate the minimum cost of a puncture while avoiding bone collisions. Further, we test multiple insertion paths concerning feasibility and collisions. A custom end effector is designed for inserting needles and extracting tissue samples under robotic guidance. Our robotic post-mortem biopsy (RPMB) system is evaluated in a study during the COVID-19 pandemic on 20 corpses and 10 tissue targets, 5 of them being infected with SARS-CoV-2. The mean planning time including robot path planning is 5.72±167s. Mean needle placement accuracy is 7.19± 422mm.} } @inproceedings{10.1007/978-3-031-06249-0_34, author = {R. Mieling and C. Stapper and S. Gerlach and M. Neidhardt and S. Latus and M. Gromniak and P. Breitfeld and A. Schlaefer}, title = {Proximity-Based Haptic Feedback for Collaborative Robotic Needle Insertion.}, year = {2022}, pages = {301-309}, editor = {In Seifi, Hasti and Kappers, Astrid M. L. and Schneider, Oliver and Drewing, Knut and Pacchierotti, Claudio and Abbasimoshaei, Alireza and Huisman, Gijs and Kern, Thorsten A. (Eds.)}, publisher = {Springer International Publishing:}, address = {Cham}, isbn = {978-3-031-06249-0}, booktitle = {Haptics: Science, Technology, Applications}, abstract = {Collaborative robotic needle insertions have the potential to improve placement accuracy and safety, e.g., during epidural anesthesia. Epidural anesthesia provides effective regional pain management but can lead to serious complications, such as nerve injury or cerebrospinal fluid leakage. Robotic assistance might prevent inadvertent puncture by providing haptic feedback to the physician. Haptic feedback can be realized on the basis of force measurements at the needle. However, contact should be avoided for delicate structures. We propose a proximity-based method to provide feedback prior to contact. We measure the distance to boundary layers, visualize the proximity for the operator and further feedback it as a haptic resistance. We compare our approach to haptic feedback based on needle forces and visual feedback without haptics. Participants are asked to realize needle insertions with each of the three feedback modes. We use phantoms that mimic the structures punctured during epidural anesthesia. We show that visual feedback improves needle placement, but only proximity-based haptic feedback reduces accidental puncture. The puncture rate is 62{\%} for force-based haptic feedback, 60{\%} for visual feedback and 6{\%} for proximity-based haptic feedback. Final needle placement inside the epidural space is achieved in 38{\%}, 70{\%} and 96{\%} for force-based haptic, visual and proximity-based haptic feedback, respectively. Our results suggest that proximity-based haptic feedback could improve needle placement safety in the context of epidural anesthesia.} } @article{xxx, author = {S. Gerlach and A. Schlaefer}, title = {Robotic Systems in Radiotherapy and Radiosurgery.}, journal = {Current Robotics Reports.}, year = {2022}, doi = {10.1007/s43154-021-00072-3}, url = {https://doi.org/10.1007/s43154-021-00072-3}, abstract = {This review provides an overview of robotic systems in radiotherapy and radiosurgery, with a focus on medical devices and recently proposed research systems. We summarize the key motivation for using robotic systems and illustrate the potential advantages} } @inproceedings{gerlach2022fast, author = {S. Gerlach and T. Hofmann and C. Fuerweger and A. Schlaefer}, title = {TH-B-206-02: Fast Adaptive Replanning by Constrained Reoptimization for Intra-Fractional Non-Periodic Motion During SBRT of the Prostate.}, year = {2022}, volume = {49.}, number = {(6),}, pages = {E570-E570}, booktitle = {Medical Physics}, url = {https://aapm.onlinelibrary.wiley.com/doi/epdf/10.1002/mp.15769}, abstract = {Purpose: Periodic motion of the target can be compensated by translational motion of the treatment beams in robotic SBRT. However, spontaneous, non-periodic displacement of the target may completely change the treatment geometry. In this case, translation is not sufficient since relative motion between the PTV and OARs can cause substantial deviations of dose in the OARs. Instead, solving a new optimization problem is required after partial dose delivery. We demonstrate this effect and propose a method for adaption by replanning which accounts for the change in the geometry. Methods:In contrast to typical adaptive strategies, our approach is based on complete and constrained replanning of the optimization problem which guarantees that no side effects such as higher doses than prescribed can occur in the treatment plan. We adapt the linear program to account for the changed treatment geometry which allows for fast reoptimization. For evaluation, we translate the target with random direction and length sampled from a truncated normal distribution with mean values from 12.5 to 30mm without overlap with OARs. We stu dy treatment plans with approximately 300 treatment beams and consider the motion to occur after 100 delivered beams. We solve in total 40,950 inverse planning problems for 45 patients. Re sults: Replanning can compensate for coverage loss and avoid constraint violation. Runtime of reoptimization is on average 14s. When not compensating for movement, coverage can decrease from 95% to 20%. While translation of the beam source can compensate for loss in coverage, dose constraints can be violated. E.g. maximum dose in the rectum is violated in 62% of treatment plans with translational compensation. Conclusion: For non-periodic target displacements, translational compensation can lead to suboptimal treatment plan delivery. Constrained replanning after partially delivery of the treatment plan can compensate for the negative impact on the delivered dose distribution} } @article{, author = {S. Gerlach and T. Hofmann and C. Fürweger and A. Schlaefer}, title = {AI-based optimization for US-guided radiation therapy of the prostate.}, journal = {International Journal of Computer Assisted Radiology and Surgery.}, year = {2022}, doi = {10.1007/s11548-022-02664-6}, url = {https://doi.org/10.1007/s11548-022-02664-6}, abstract = {Fast volumetric ultrasound presents an interesting modality for continuous and real-time intra-fractional target tracking in radiation therapy of lesions in the abdomen. However, the placement of the ultrasound probe close to the target structures leads to blocking some beam directions.} } @article{GrubeNeidhardtLatusSchlaefer+2022+42+45, author = {S. Grube and M. Neidhardt and S. Latus and A. Schlaefer}, title = {Influence of the Field of View on Shear Wave Velocity Estimation.}, journal = {Current Directions in Biomedical Engineering.}, year = {2022}, volume = {8.}, number = {(1),}, pages = {42--45}, doi = {doi:10.1515/cdbme-2022-0011}, url = {https://doi.org/10.1515/cdbme-2022-0011}, abstract = {Tissue elasticity contains important informationfor physicians in diagnosis and treatment, and, e.g., can help intumor detection because tumors are stiffer than healthy tissue.Ultrasound shear wave elastography imaging (US-SWEI) canbe used to estimate tissue stiffness by measuring the velocityof induced shear waves. Commonly, a linear US probe is usedto track shear waves at a high imaging frequency in 2D. Real-time US-SWEI is limited by the required time for data process-ing. Hence, reducing the imaging field of view (FOV) is ben-eficial as it decreases the size of the acquired data and therebythe acquisition, transfer and processing time. However, a de-crease in the FOV has the disadvantage that shear waves aretracked over a smaller distance and thus, also fewer samplingpoints are available for velocity estimation. This trade-off be-tween a smaller FOV and thus, a smaller data size, and theaccuracy of shear wave velocity estimation is investigated inthis work. For this purpose, shear waves were tracked with alinear US probe in gelatin phantoms with four different stiff-ness values. During data processing, we reduced the FOV vir-tually from38.1 mmto2.1 mm. It was found that a reductionof the FOV to4.5 mmleads to an overestimation of up to fivetimes larger shear wave velocities but still allows to distinguishbetween phantoms of different stiffness. However, not all esti-mated velocity values could be clearly assigned to the correctstiffness value. The smallest studied FOV of2.1 mmwas notsufficient for distinguishing between the phantoms anymore} } @inproceedings{Lehmann2022, author = {S. Lehmann, A. Rogalla, M. Neidhardt, A. Reinecke, A. Schlaefer, S. Schupp}, title = {Modeling R3 Needle Steering in Uppaal.}, year = {2022}, volume = {355.}, pages = {40-59}, editor = {In Dubslaff, Clemens and Luttik, Bas (Eds.)}, publisher = {Open Publishing Association:}, series = {Electronic Proceedings in Theoretical Computer Science}, booktitle = {{\rm Proceedings Fifth Workshop on} Models for Formal Analysis of Real Systems, {\rm Munich, Germany, 2nd April 2022}}, doi = {10.4204/EPTCS.355.4}, abstract = {Medical cyber-physical systems are safety-critical, and as such, require ongoing verification of their correct behavior, as system failure during run time may cause severe (or even fatal) personal damage. However, creating a verifiable model often conflicts with other application requirements, most notably regarding data precision and model accuracy, as efficient model checking promotes discrete data (over continuous) and abstract models to reduce the state space. In this paper, we approach the task of medical needle steering in soft tissue around potential obstacles. We design a verifiable model of needle motion (implemented in Uppaal Stratego) and a framework embedding the model for online needle steering. We mitigate the conflict by imposing boundedness on both the data types, reducing from R^3 to Z^3 when needed, and the motion and environment models, reducing the set of allowed local actions and global paths. In experiments, we successfully apply the static model alone, as well as the dynamic framework in scenarios with varying environment complexity and both a virtual and real needle setting, where up to 100% of targets were reached depending on the scenario and needle. } } @inproceedings{11420_14525, author = {T. Sonntag and M. Bauer and J. Sprenger and S. Gerlach and P. Breitfeld and A. Schlaefer}, title = {Deep learning based segmentation of cervical blood vessels in ultrasound images.}, journal = {European journal of anaesthesiology.}, year = {2022}, volume = {39.}, number = {(e-Supplement 60),}, pages = {41-41}, isbn = {0265-0215}, booktitle = {The European Anaesthesiology Congress, Euroanaesthesia 2022}, url = {http://hdl.handle.net/11420/14525}, abstract = {Puncture of central vessels is a frequently used therapeutic and diagnostic procedure. The use of ultrasound (US) during needle insertion has become the gold standard. Handling the US probe and needle is challenging, especially in difficult anatomic conditions. Our long-term vision is a deep learning based and augmented reality (AR) assisted needle puncture. We aim to visualize the vessel structures in 3D based on 2D US image segmentation. While punctuating, the relative needle tip position and relevant vessels can be highlighted via AR lenses to optimize the image guidance process} } @misc{Ellebrecht2021, author = {D. B. Ellebrecht and N. Hessler and A. Schlaefer and N. Gessert}, title = {Confocal Laser Microscopy for in vivo Intraoperative Application: Diagnostic Accuracy of Investigator and Machine Learning Strategies.}, year = {2021}, volume = {37.}, number = {(6),}, pages = {533-541}, doi = {10.1159/000517146}, url = {https://www.karger.com/DOI/10.1159/000517146}, abstract = {Background: Confocal laser microscopy (CLM) is one of the optical techniques that are promising methods of intraoperative in vivo real-time tissue examination based on tissue fluorescence. However, surgeons might struggle interpreting CLM images intraoperatively due to different tissue characteristics of different tissue pathologies in clinical reality. Deep learning techniques enable fast and consistent image analysis and might support intraoperative image interpretation. The objective of this study was to analyze the diagnostic accuracy of newly trained observers in the evaluation of normal colon and peritoneal tissue and colon cancer and metastasis, respectively, and to compare it with that of convolutional neural networks (CNNs). Methods: Two hundred representative CLM images of the normal and malignant colon and peritoneal tissue were evaluated by newly trained observers (surgeons and pathologists) and CNNs (VGG-16 and Densenet121), respectively, based on tissue dignity. The primary endpoint was the correct detection of the normal and cancer/metastasis tissue measured by sensitivity and specificity of both groups. Additionally, positive predictive values (PPVs) and negative predictive values (NPVs) were calculated for the newly trained observer group. The interobserver variability of dignity evaluation was calculated using kappa statistic. The F1-score and area under the curve (AUC) were used to evaluate the performance of image recognition of the CNNs\’ training scenarios. Results: Sensitivity and specificity ranged between 0.55 and 1.0 (pathologists: 0.66-0.97; surgeons: 0.55-1.0) and between 0.65 and 0.96 (pathologists: 0.68-0.93; surgeons: 0.65-0.96), respectively. PPVs were 0.75 and 0.90 in the pathologists\’ group and 0.73-0.96 in the surgeons\’ group, respectively. NPVs were 0.73 and 0.96 for pathologists\’ and between 0.66 and 1.00 for surgeons\’ tissue analysis. The overall interobserver variability was 0.54. Depending on the training scenario, cancer/metastasis tissue was classified with an AUC of 0.77-0.88 by VGG-16 and 0.85-0.89 by Densenet121. Transfer learning improved performance over training from scratch. Conclusions: Newly trained investigators are able to learn CLM images features and interpretation rapidly, regardless of their clinical experience. Heterogeneity in tissue diagnosis and a moderate interobserver variability reflect the clinical reality more realistic. CNNs provide comparable diagnostic results as clinical observers and could improve surgeons’ intraoperative tissue assessment.} } @conference{bhattacharya2021selfsupervised, author = {D. Bhattacharya and C. Betz and D. Eggert and A. Schlaefer}, title = {Self-Supervised U-Net for Segmenting Flat and Sessile Polyps..}, year = {2021}, booktitle = {SPIE Medical Imaging Symposium 2021}, url = {https://arxiv.org/abs/2110.08776#}, abstract = {Colorectal Cancer(CRC) poses a great risk to public health. It is the third most common cause of cancer in the US. Development of colorectal polyps is one of the earliest signs of cancer. Early detection and resection of polyps can greatly increase survival rate to 90%. Manual inspection can cause misdetections because polyps vary in color, shape, size and appearance. To this end, Computer-Aided Diagnosis systems(CADx) has been proposed that detect polyps by processing the colonoscopic videos. The system acts a secondary check to help clinicians reduce misdetections so that polyps may be resected before they transform to cancer. Polyps vary in color, shape, size, texture and appearance. As a result, the miss rate of polyps is between 6% and 27% despite the prominence of CADx solutions. Furthermore, sessile and flat polyps which have diameter less than 10 mm are more likely to be undetected. Convolutional Neural Networks(CNN) have shown promising results in polyp segmentation. However, all of these works have a supervised approach and are limited by the size of the dataset. It was observed that smaller datasets reduce the segmentation accuracy of ResUNet++. We train a U-Net to inpaint randomly dropped out pixels in the image as a proxy task. The dataset we use for pre-training is Kvasir-SEG dataset. This is followed by a supervised training on the limited Kvasir-Sessile dataset. Our experimental results demonstrate that with limited annotated dataset and a larger unlabeled dataset, self-supervised approach is a better alternative than fully supervised approach. Specifically, our self-supervised U-Net performs better than five segmentation models which were trained in supervised manner on the Kvasir-Sessile dataset.} } @article{deb-nmi-2021, author = {D. Bhattacharya and C. Betz and D. Eggert and A. Schlaefer}, title = {Dual Parallel Reverse Attention Edge Network : DPRA-EdgeNet..}, journal = {Nordic Machine Intelligence, MedAI2021.}, year = {2021}, volume = {1..}, number = {((1),),}, pages = {11-13 Second place in challenge task}, doi = {https://doi.org/10.5617/nmi.9116}, keywords = {polyp segmentation, attention gates, deep learning, colon cancer, polyp, medical image segmentation}, abstract = {In this paper, we propose the Dual Parallel Reverse AttentionEdge Network (DPRA-EdgeNet), an architecture that jointlylearns to segment an object and its edges. We compareour model against three popular segmentation models anddemonstrate that our model improves the segmentationaccuracy on the Kvasir-SEG dataset and the Kvasir-Instrumentdataset} } @article{SCHMIDT2021100950, author = {F. N. Schmidt and S. Gerlach and M. Issleib and A. Schlaefer and B. Busse}, title = {Development of a virtual reality-based training for the elderly with increased fracture risk to prevent falls and improve their balance.}, journal = {Bone Reports.}, year = {2021}, volume = {14.}, pages = {100950}, note = {Abstracts of the ECTS Congress 2021}, doi = {https://doi.org/10.1016/j.bonr.2021.100950}, url = {https://www.sciencedirect.com/science/article/pii/S2352187221002059} } @article{doi:10.1098/rsif.2021.0411, author = {G. A. Holzapfel and K. Linka and S. Sherifova and C. J. Cyron}, title = {Predictive constitutive modelling of arteries by deep learning.}, journal = {Journal of The Royal Society Interface.}, year = {2021}, volume = {18.}, number = {(182),}, pages = {20210411}, doi = {10.1098/rsif.2021.0411}, url = {https://royalsocietypublishing.org/doi/abs/10.1098/rsif.2021.0411}, abstract = {The constitutive modelling of soft biological tissues has rapidly gained attention over the last 20 years. Current constitutive models can describe the mechanical properties of arterial tissue. Predicting these properties from microstructural information, however, remains an elusive goal. To address this challenge, we are introducing a novel hybrid modelling framework that combines advanced theoretical concepts with deep learning. It uses data from mechanical tests, histological analysis and images from second-harmonic generation. In this first proof of concept study, our hybrid modelling framework is trained with data from 27 tissue samples only. Even such a small amount of data is sufficient to be able to predict the stress–stretch curves of tissue samples with a median coefficient of determination of R2 = 0.97 from microstructural information, as long as one limits the scope to tissue samples whose mechanical properties remain in the range commonly encountered. This finding suggests that deep learning could have a transformative impact on the way we model the constitutive properties of soft biological tissues.} } @article{9508399, author = {J. F. Fast and H. R. Dava and A. K. Rüppel and D. Kundrat and M. Krauth and M.-H. Laves and S. Spindeldreier and L. A. Kahrs and M. Ptok}, title = {Stereo Laryngoscopic Impact Site Prediction for Droplet-Based Stimulation of the Laryngeal Adductor Reflex.}, journal = {IEEE Access.}, year = {2021}, volume = {9.}, pages = {112177-112192}, doi = {10.1109/ACCESS.2021.3103049}, abstract = {The laryngeal adductor reflex (LAR) is a vital reflex of the human larynx. LAR malfunctions may cause life-threatening aspiration events. An objective, noninvasive, and reproducible method for LAR assessment is still lacking. Stimulation of the larynx by droplet impact, termed Microdroplet Impulse Testing of the LAR (MIT-LAR), may remedy this situation. However, droplet instability and imprecise stimulus application thus far prevented MIT-LAR from gaining clinical relevance. We present a system comprising two alternative, custom-built stereo laryngoscopes, each offering a distinct set of properties, a droplet applicator module, and image/point cloud processing algorithms to enable a targeted, droplet-based LAR stimulation. Droplet impact site prediction (ISP) is achieved by droplet trajectory identification and spatial target reconstruction. The reconstruction and ISP accuracies were experimentally evaluated. Global spatial reconstruction errors at the glottal area of (0.3±0.3) mm and (0.4±0.3) mm and global ISP errors of (0.9±0.6) mm and (1.3±0.8) mm were found for a rod lens-based and an alternative, fiberoptic laryngoscope, respectively. In the case of the rod lens-based system, 96% of all observed ISP error values are inferior to 2 mm; a value of 80% was found with the fiberoptic assembly. This contribution represents an important step towards introducing a reproducible and objective LAR screening method into the clinical routine.} } @article{Krueger2021, author = {J. Krüger and A. C. Ostwaldt and L. Spies and B. Geisler and A. Schlaefer, and H. H. Kitzler and S. Schippling and R. Opfer}, title = {Infratentorial lesions in multiple sclerosis patients: intra- and inter-rater variability in comparison to a fully automated segmentation using 3D convolutional neural networks.}, year = {2021}, doi = {10.1007/s00330-021-08329-3}, url = {https://doi.org/10.1007/s00330-021-08329-3}, abstract = {Automated quantification of infratentorial multiple sclerosis lesions on magnetic resonance imaging is clinically relevant but challenging. To overcome some of these problems, we propose a fully automated lesion segmentation algorithm using 3D convolutional neural networks (CNNs).} } @article{pamm.202000148, author = {J. Ohlsen and M. Neidhardt and A. Schlaefer and N. Hoffmann}, title = {Modelling shear wave propagation in soft tissue surrogates using a finite element- and finite difference method.}, journal = {PAMM.}, year = {2021}, volume = {20.}, number = {(1),}, pages = {e202000148}, doi = {https://doi.org/10.1002/pamm.202000148}, url = {https://onlinelibrary.wiley.com/doi/abs/10.1002/pamm.202000148}, abstract = {Abstract Shear Wave Elasticity Imaging (SWEI) has become a popular medical imaging technique [1] in which soft tissue is excited by the acoustic radiation forces of a focused ultrasonic beam. Tissue stiffness can then be derived from measurements of shear wave propagation speeds [2]. The main objective of this work is a comparison of a finite element (FEM) and a finite difference method (FDM) in terms of their computational efficiency when modeling shear wave propagation in tissue phantoms. Moreover, the propagation of shear waves is examined in experiments with ballistic gelatin to assess the simulation results. In comparison to the FEM the investigated FDM proves to be significantly more performant for this computing task} } @article{SprengerPetersenNeumannReichenspurnerRussDetterSch, author = {J. Sprenger and J. Petersen and N. Neumann and H. Reichenspurner and D. Russ and C. Detter and A. Schlaefer}, title = {Tracking heart surface features to determine myocardial contrast agent enrichment:.}, journal = {Current Directions in Biomedical Engineering.}, year = {2021}, volume = {7.}, number = {(1),}, pages = {53-57}, doi = {doi:10.1515/cdbme-2021-1012}, url = {https://doi.org/10.1515/cdbme-2021-1012}, abstract = {Fluorescent cardiac imaging can be applied for intraoperative quality control after a coronary bypass grafting surgery to ensure the myocardial perfusion by evaluating the increasing contrast agent enrichment in the heart. The motion due to the beating heart impedes the interpretation of the contrast agent enrichment in the vessels and leads to noisy enrichment curves. We propose tracking of the heart surface features to compensate for the motion of the beating heart and thereby improve the analysis of the contrast agent enrichment. Furthermore, we propose a vessel segmentation pipeline for a local evaluation of contrast agent enrichment directly in the vessels} } @inproceedings{sprenger-spi2021, author = {J. Sprenger and M. Neidhardt and M. Schlüter and S. Latus and T. Gosau and J. Kemmling and S. Feldhaus and U. Schumacher and A. Schlaefer}, title = {In-vivo markerless motion detection from volumetric optical coherence tomography data using CNNs.}, year = {2021}, volume = {11598.}, pages = {345 - 350}, editor = {In Cristian A. Linte and Jeffrey H. Siewerdsen (Eds.)}, publisher = {SPIE:}, booktitle = {Medical Imaging 2021: Image-Guided Procedures, Robotic Interventions, and Modeling}, organization = {International Society for Optics and Photonics}, doi = {10.1117/12.2581023}, url = {https://doi.org/10.1117/12.2581023}, keywords = {deep learning, optical coherence tomography, markerless motion detection}, abstract = {Precise navigation is an important task in robot-assisted and minimally invasive surgery. The need for optical markers and a lack of distinct anatomical features on skin or organs complicate tissue tracking with commercial tracking systems. Previous work has shown the feasibility of a 3D optical coherence tomography based system for this purpose. Furthermore, convolutional neural networks have been proven to precisely detect shifts between volumes. However, most experiments have been performed with phantoms or ex-vivo tissue. We introduce an experimental setup and perform measurements on perfused and non-perfused (dead) tissue of in-vivo xenograft tumors. We train 3D siamese deep learning models and evaluate the precision of the motion prediction. The network\'s ability to predict shifts for different motion magnitudes and also the performance for the different volume axes are compared. The root-mean-square errors are 0:12mm and 0:08mm on perfused and non-perfused tumor tissue, respectively} } @inproceedings{sprenger_ieee_sbmi_2021, author = {J. Sprenger and T. Saathoff and A. Schlaefer}, title = {Automated robotic surface scanning with optical coherence tomography.}, year = {2021}, pages = {1137-1140}, booktitle = {IEEE 18th International Symposium on Biomedical Imaging}, organization = {IEEE}, abstract = {Optical coherence tomography (OCT) is a near-infrared light based imaging modality that enables depth scans with a high spatial resolution. By scanning along the lateral dimensions, high-resolution volumes can be acquired. This allows to characterize tissue and precisely detect abnormal structures in medical scenarios. However, its small field of view (FOV) limits the applicability of OCT for medical examinations. We therefore present an automated setup to move an OCT scan head over arbitrary surfaces. By mounting the scan head to a highly accurate robot arm, we obtain precise information about the position of the acquired volumes. We implement a geometric approach to stitch the volumes and generate the surface scans. Our results show that a precise stitching of the volumes is achieved with mean absolute errors of 0\.078mm and 0\.098mm in the lateral directions and 0\.037mm in the axial direction. We can show that our setup provides automated surface scanning with OCT of samples and phantoms larger than the usual FOV} } @article{LINKA2021110010, author = {K. Linka and M. Hillgärtner and K. P. Abdolazizi and R. C. Aydin and M. Itskov and C. J. Cyron}, title = {Constitutive artificial neural networks: A fast and general approach to predictive data-driven constitutive modeling by deep learning.}, journal = {Journal of Computational Physics.}, year = {2021}, volume = {429.}, pages = {110010}, doi = {https://doi.org/10.1016/j.jcp.2020.110010}, url = {https://www.sciencedirect.com/science/article/pii/S0021999120307841}, keywords = {Deep learning, Data-driven, Constitutive modeling, Hyperelasticity}, abstract = {In this paper we introduce constitutive artificial neural networks (CANNs), a novel machine learning architecture for data-driven modeling of the mechanical constitutive behavior of materials. CANNs are able to incorporate by their very design information from three different sources, namely stress-strain data, theoretical knowledge from materials theory, and diverse additional information (e.g., about microstructure or materials processing). CANNs can easily and efficiently be implemented in standard computational software. They require only a low-to-moderate amount of training data and training time to learn without human guidance the constitutive behavior also of complex nonlinear and anisotropic materials. Moreover, in a simple academic example we demonstrate how the input of microstructural data can endow CANNs with the ability to describe not only the behavior of known materials but to predict also the properties of new materials where no stress-strain data are available yet. This ability may be particularly useful for the future in-silico design of new materials. The developed source code of the CANN architecture and accompanying example data sets are available at https://github.com/ConstitutiveANN/CANN.} } @article{https://doi.org/10.1002/pamm.202000284, author = {K. P. Abdolazizi and K. Linka and J. Sprenger and M. Neidhardt and A. Schlaefer and C. J. Cyron}, title = {Concentration-Specific Constitutive Modeling of Gelatin Based on Artificial Neural Networks.}, journal = {PAMM.}, year = {2021}, volume = {20.}, number = {(1),}, pages = {e202000284}, doi = {https://doi.org/10.1002/pamm.202000284}, url = {https://onlinelibrary.wiley.com/doi/abs/10.1002/pamm.202000284}, abstract = {Abstract Gelatin phantoms are frequently used in the development of surgical devices and medical imaging techniques. They exhibit mechanical properties similar to soft biological tissues [1] but can be handled at a much lower cost. Moreover, they enable a better reproducibility of experiments. Accurate constitutive models for gelatin are therefore of great interest for biomedical engineering. In particular it is important to capture the dependence of mechanical properties of gelatin on its concentration. Herein we propose a simple machine learning approach to this end. It uses artificial neural networks (ANN) for learning from indentation data the relation between the concentration of ballistic gelatin and the resulting mechanical properties.} } @article{11420_11562, author = {K. P. Abdolazizi and K. Linka and J. Sprenger and M. Neidhardt and A. Schlaefer and C. J. Cyron}, title = {Identification of the concentration‐dependent viscoelastic constitutive parameters of gelatin by combining computational mechanics and machine learning.}, journal = {Proceedings in applied mathematics and mechanics.}, year = {2021}, volume = {21.}, number = {(1),}, pages = {e202100250}, url = {https://onlinelibrary.wiley.com/doi/abs/10.1002/pamm.202100250}, abstract = {Since the mechanical properties of gelatin are similar to those of soft biological tissues, gelatin is a commonly used surrogate for real tissues, for example in safety engineering or medical engineering. Additional advantages of gelatin over real tissues are lower costs and better reproducibility of experiments. Therefore, constitutive models of gelatin are of great interest. In particular, it is important to capture the concentration dependence of the mechanical properties since the gelatin mass concentration significantly affects the constitutive behavior. To this end, we propose a hybrid approach linking artificial neural networks (ANN) and classical constitutive modeling to relate the gelatin\'s concentration to its viscoelastic material properties using indentation data} } @article{BargstenKlischRiedlWisselBrunnerSchaefersGrassBlan, author = {L. Bargsten and D. Klisch and K. A. Riedl and T. Wissel and F. J. Brunner and K. Schaefers and M. Grass and S. Blankenberg and M. Seiffert and A. Schlaefer}, title = {Deep learning for guidewire detection in intravascular ultrasound images:.}, journal = {Current Directions in Biomedical Engineering.}, year = {2021}, volume = {7.}, number = {(1),}, pages = {106-110}, doi = {doi:10.1515/cdbme-2021-1023}, url = {https://doi.org/10.1515/cdbme-2021-1023}, abstract = {Algorithms for automated analysis of intravascular ultrasound (IVUS) images can be disturbed by guidewires, which are often encountered when treating bifurcations in percutaneous coronary interventions. Detecting guidewires in advance can therefore help avoiding potential errors. This task is not trivial, since guidewires appear rather small compared to other relevant objects in IVUS images. We employed CNNs with additional multi-task learning as well as different guidewire-specific regularizations to enable and improve guidewire detection. In this context, we developed a network block which generates heatmaps that highlight guidewires without the need of localization annotations. The guidewire detection results reach values of 0.931 in terms of the F1-score and 0.996 in terms of area under curve (AUC). Comparing thresholded guidewire heatmaps with ground truth segmentation masks leads to a Dice score of 23.1 % and an average Hausdorff distance of 1.45 mm. Guidewire detection has proven to be a task that CNNs can handle quite well. Employing multi-task learning and guidewire-specific regularizations further improve detection results and enable generation of heatmaps that indicate the position of guidewires without actual labels} } @inproceedings{Bargsten12.2580720, author = {L. Bargsten and K. A. Riedl and T. Wissel and F. J. Brunner and K. Schaefers and J. Sprenger and M. Grass and M. Seiffert and S. Blankenberg and A. Schlaefer}, title = {Tailored methods for segmentation of intravascular ultrasound images via convolutional neural networks.}, year = {2021}, volume = {11602.}, pages = {1-7}, editor = {In Brett C. Byram and Nicole V. Ruiter (Eds.)}, publisher = {SPIE:}, booktitle = {Medical Imaging 2021: Ultrasonic Imaging and Tomography}, organization = {International Society for Optics and Photonics}, doi = {10.1117/12.2580720}, url = {https://doi.org/10.1117/12.2580720}, keywords = {Intravascular ultrasound, Convolutional neural networks, Segmentation, Speckle statistics, Shape priors, Domain knowledge, Deep learning}, abstract = {Automatic delineation of relevant structures in intravascular imaging can support percutaneous coronary interventions (PCIs), especially when dealing with rather demanding cases. We found three major error types which occur regularly when segmenting lumen and wall of morphologically complex vessels with convolutional neural networks (CNNs). In order to reduce these three error types, we developed three IVUS-specific methods which are able to improve generalizability of state-of-the-art CNNs for IVUS segmentation tasks. These methods are based on three concepts: speckle statistics, artery shape priors via independent component analysis (ICA) and the concentricity condition of lumen and vessel wall. We found that all three methods outperform the baseline. Since all three concepts can be readily transferred to intravascular optical coherence tomography (IVOCT), we expect these findings can support the segmentation of corresponding images as well} } @article{BargstenRiedlWisselBrunnerSchaefersGrassBlankenber, author = {L. Bargsten and K. A. Riedl and T. Wissel and F. J. Brunner and K. Schaefers and M. Grass and S. Blankenberg and M. Seiffert and A. Schlaefer}, title = {Deep learning for calcium segmentation in intravascular ultrasound images:.}, journal = {Current Directions in Biomedical Engineering.}, year = {2021}, volume = {7.}, number = {(1),}, pages = {96-100}, doi = {doi:10.1515/cdbme-2021-1021}, url = {https://doi.org/10.1515/cdbme-2021-1021}, abstract = {Knowing the shape of vascular calcifications is crucial for appropriate planning and conductance of percutaneous coronary interventions. The clinical workflow can therefore benefit from automatic segmentation of calcified plaques in intravascular ultrasound (IVUS) images. To solve segmentation problems with convolutional neural networks (CNNs), large datasets are usually required. However, datasets are often rather small in the medical domain. Hence, developing and investigating methods for increasing CNN performance on small datasets can help on the way towards clinically relevant results. We compared two state-of-the-art CNN architectures for segmentation, U-Net and DeepLabV3, and investigated how incorporating auxiliary image data with vessel wall and lumen annotations improves the calcium segmentation performance by using these either for pre-training or multi-task training. DeepLabV3 outperforms U-Net with up to 6.3 % by means of the Dice coefficient and 36.5 % by means of the average Hausdorff distance. Using auxiliary data improves the segmentation performance in both cases, whereas the multi-task approach outperforms the pre-training approach. The improvements of the multi-task approach in contrast to not using auxiliary-data at all is 5.7 % for the Dice coefficient and 42.9 % for the average Hausdorff distance. Automatic segmentation of calcified plaques in IVUS images is a demanding task due to their relatively small size compared to the image dimensions and due to visual ambiguities with other image structures. We showed that this problem can generally be tackled by CNNs. Furthermore, we were able to improve the performance by a multi-task learning approach with auxiliary segmentation data} } @inproceedings{pmlr-v143-bargsten21a, author = {L. Bargsten and K. A. Riedl and T. Wissel and F. J. Brunner and K. Schaefers and M. Grass and S. Blankenberg and M. Seiffert and A. Schlaefer}, title = {Attention via Scattering Transforms for Segmentation of Small Intravascular Ultrasound Data Sets.}, year = {2021}, volume = {143.}, pages = {34-47}, month = {07-09 Jul}, editor = {In Heinrich, Mattias and Dou, Qi and de Bruijne, Marleen and Lellmann, Jan and Schläfer, Alexander and Ernst, Floris (Eds.)}, publisher = {PMLR:}, series = {Proceedings of Machine Learning Research}, booktitle = {Proceedings of the Fourth Conference on Medical Imaging with Deep Learning}, url = {https://proceedings.mlr.press/v143/bargsten21a.html}, abstract = {Using intracoronary imaging modalities like intravascular ultrasound (IVUS) has a positive impact on the results of percutaneous coronary interventions. Efficient extraction of important vessel metrics like lumen diameter, vessel wall thickness or plaque burden via automatic segmentation of IVUS images can improve the clinical workflow. State-of-the-art segmentation results are usually achieved by data-driven methods like convolutional neural networks (CNNs). However, clinical data sets are often rather small leading to extraction of image features which are not very meaningful and thus decreasing performance. This is also the case for some applications which inherently allow for only small amounts of available data, e.g., detection of diseases with extremely small prevalence or online-adaptation of an existing algorithm to individual patients. In this work we investigate how integrating scattering transformations - as special forms of wavelet transformations - into CNNs could improve the extraction of meaningful features. To this end, we developed a novel network module which uses features of a scattering transform for an attention mechanism. We observed that this approach improves the results of calcium segmentation up to 8.2% (relatively) in terms of the Dice coefficient and 24.8% in terms of the modified Hausdorff distance. In the case of lumen and vessel wall segmentation, the improvements are up to 2.3% (relatively) in terms of the Dice coefficient and 30.8% in terms of the modified Hausdorff distance.Incorporating scattering transformations as a component of an attention block into CNNs improves the segmentation results on small IVUS segmentation data sets. In general, scattering transformations can help in situations where efficient feature extractors can not be learned via the training data. This makes our attention module an interesting candidate for applications like few-shot learning for patient adaptation or detection of rare diseases.} } @article{Bargsten2021, author = {L. Bargsten and S. Raschka and A. Schlaefer}, title = {Capsule networks for segmentation of small intravascular ultrasound image datasets.}, journal = {International Journal of Computer Assisted Radiology and Surgery.}, year = {2021}, volume = {16.}, number = {(8),}, pages = {1243-1254}, doi = {10.1007/s11548-021-02417-x}, url = {https://doi.org/10.1007/s11548-021-02417-x}, abstract = {Intravascular ultrasound (IVUS) imaging is crucial for planning and performing percutaneous coronary interventions. Automatic segmentation of lumen and vessel wall in IVUS images can thus help streamlining the clinical workflow. State-of-the-art results in image segmentation are achieved with data-driven methods like convolutional neural networks (CNNs). These need large amounts of training data to perform sufficiently well but medical image datasets are often rather small. A possibility to overcome this problem is exploiting alternative network architectures like capsule networks.} } @article{bengs2021three, author = {M. Bengs and F. Behrendt and J. Krüger and R. Opfer and A. Schlaefer}, title = {Three-dimensional deep learning with spatial erasing for unsupervised anomaly segmentation in brain MRI.}, journal = {International journal of computer assisted radiology and surgery.}, year = {2021}, volume = {16.}, number = {(9),}, pages = {1413-1423}, publisher = {Springer:}, doi = {https://doi.org/10.1007/s11548-021-02451-9}, abstract = {Brain Magnetic Resonance Images (MRIs) are essential for the diagnosis of neurological diseases. Recently, deep learning methods for unsupervised anomaly detection (UAD) have been proposed for the analysis of brain MRI. These methods rely on healthy brain MRIs and eliminate the requirement of pixel-wise annotated data compared to supervised deep learning. While a wide range of methods for UAD have been proposed, these methods are mostly 2D and only learn from MRI slices, disregarding that brain lesions are inherently 3D and the spatial context of MRI volumes remains unexploited} } @article{, author = {M. Bengs and F. Behrendt and J. Krüger and R. Opfer and A. Schlaefer}, title = {Three-dimensional deep learning with spatial erasing for unsupervised anomaly segmentation in brain MRI.}, journal = {International Journal of Computer Assisted Radiology and Surgery.}, year = {2021}, volume = {16.}, number = {(9),}, pages = {1413-1423}, doi = {10.1007/s11548-021-02451-9}, url = {https://doi.org/10.1007/s11548-021-02451-9}, abstract = {Brain Magnetic Resonance Images (MRIs) are essential for the diagnosis of neurological diseases. Recently, deep learning methods for unsupervised anomaly detection (UAD) have been proposed for the analysis of brain MRI. These methods rely on healthy brain MRIs and eliminate the requirement of pixel-wise annotated data compared to supervised deep learning. While a wide range of methods for UAD have been proposed, these methods are mostly 2D and only learn from MRI slices, disregarding that brain lesions are inherently 3D and the spatial context of MRI volumes remains unexploited.} } @inproceedings{10.1117/12.2580717, author = {M. Bengs and M. Bockmayr and U. Schüller and A. Schlaefer}, title = {Medulloblastoma tumor classification using deep transfer learning with multi-scale EfficientNets.}, year = {2021}, volume = {11603.}, pages = {70-75}, editor = {In John E. Tomaszewski and Aaron D. Ward (Eds.)}, publisher = {SPIE:}, booktitle = {Medical Imaging 2021: Digital Pathology}, organization = {International Society for Optics and Photonics}, doi = {10.1117/12.2580717}, url = {https://doi.org/10.1117/12.2580717}, keywords = {transfer learning, convolutional neural networks, digital pathology, histopathology, image analysis, medulloblastoma}, abstract = {Medulloblastoma (MB) is the most common malignant brain tumor in childhood. The diagnosis is generally based on the microscopic evaluation of histopathological tissue slides. However, visual-only assessment of histopathological patterns is a tedious and time-consuming task and is also affected by observer variability. Hence, automated MB tumor classification could assist pathologists by promoting consistency and robust quantification. Recently, convolutional neural networks (CNNs) have been proposed for this task, while transfer learning has shown promising results. In this work, we propose an end-to-end MB tumor classification and explore transfer learning with various input sizes and matching network dimensions. We focus on differentiating between the histological subtypes classic and desmoplastic/nodular. For this purpose, we systematically evaluate recently proposed EfficientNets, which uniformly scale all dimensions of a CNN. Using a data set with 161 cases, we demonstrate that pre-trained EfficientNets with larger input resolutions lead to significant performance improvements compared to commonly used pre-trained CNN architectures. Also, we highlight the importance of transfer learning, when using such large architectures. Overall, our best performing method achieves an F1-Score of 80.1%} } @article{BengsPantBockmayrSchüllerSchlaefer+2021+63+66, author = {M. Bengs and S. Pant and M. Bockmayr and U. Schüller and A. Schlaefer}, title = {Multi-Scale Input Strategies for Medulloblastoma Tumor Classification using Deep Transfer Learning.}, journal = {Current Directions in Biomedical Engineering.}, year = {2021}, volume = {7.}, number = {(1),}, pages = {63-66}, doi = {doi:10.1515/cdbme-2021-1014}, url = {https://doi.org/10.1515/cdbme-2021-1014}, abstract = {Medulloblastoma (MB) is a primary central nervous system tumor and the most common malignant brain cancer among children. Neuropathologists perform microscopic inspection of histopathological tissue slides under a microscope to assess the severity of the tumor. This is a time-consuming task and often infused with observer variability. Recently, pre-trained convolutional neural networks (CNN) have shown promising results for MB subtype classification. Typically, high-resolution images are divided into smaller tiles for classification, while the size of the tiles has not been systematically evaluated. We study the impact of tile size and input strategy and classify the two major histopathological subtypes—Classic and Desmoplastic/Nodular. To this end, we use recently proposed EfficientNets and evaluate tiles with increasing size combined with various downsampling scales. Our results demonstrate using large input tiles pixels followed by intermediate downsampling and patch cropping significantly improves MB classification performance. Our top-performing method achieves the AUC-ROC value of 90.90% compared to 84.53% using the previous approach with smaller input tiles} } @article{NeidhardtOhlsenHoffmannSchlaefer+2021+35+38, author = {M. Neidhardt and J. Ohlsen and N. Hoffmann and A. Schlaefer}, title = {Parameter Identification for Ultrasound Shear Wave Elastography Simulation:.}, journal = {Current Directions in Biomedical Engineering.}, year = {2021}, volume = {7.}, number = {(1),}, pages = {35-38}, doi = {doi:10.1515/cdbme-2021-1008}, url = {https://doi.org/10.1515/cdbme-2021-1008}, abstract = {Elasticity of soft tissue is a valuable information to physicians in treatment and diagnosis of diseases. The elastic properties of tissue can be estimated with ultrasound (US) shear wave imaging (SWEI). In US-SWEI, a force push is applied inside the tissue and the resulting shear wave is detected by high-frequency imaging. The properties of the wave such as the shear wave velocity can be mapped to tissue elasticity. Commonly, wave features are extracted by tracking the peak of the shear wave, estimating the phase velocity or with machine learning methods. To tune and test these methods, often simulation data is employed since material properties and excitation can be accurately controlled. Subsequent validation on real US-SWEI data is in many cases performed on tissue phantoms such as gelatine. Clearly, validation performance of these procedures is dependent on the accuracy of the simulated tissue phantom and a thorough comparison of simulation and experimental data is needed. In this work, we estimate wave parameters from 400 US-SWEI data sets acquired in various homogeneous gelatine phantoms. We tune a linear material model to these parameters. We report an absolute percentage error for the shear wave velocity between simulation and phantom experiment of } } @article{NeidhardtGerlachLavesLatusStapperGromniakSchlaefer, author = {M. Neidhardt and S. Gerlach and M.-H. Laves and S. Latus and C. Stapper and M. Gromniak and A. Schlaefer}, title = {Collaborative robot assisted smart needle placement.}, journal = {Current Directions in Biomedical Engineering.}, year = {2021}, volume = {7.}, number = {(2),}, pages = {472--475}, doi = {doi:10.1515/cdbme-2021-2120}, url = {https://doi.org/10.1515/cdbme-2021-2120}, abstract = {Needles are key tools to realize minimally invasive interventions. Physicians commonly rely on subjectively perceived insertion forces at the distal end of the needle when advancing the needle tip to the desired target. However, detecting tissue transitions at the distal end of the needle is difficult since the sensed forces are dominated by shaft forces. Disentangling insertion forces has the potential to substantially improve needle placement accuracy. We propose a collaborative system for robotic needle insertion, relaying haptic information sensed directly at the needle tip to the physician by haptic feedback through a light weight robot. We integrate optical fibers into medical needles and use optical coherence tomography to image a moving surface at the tip of the needle. Using a convolutional neural network, we estimate forces acting on the needle tip from the optical coherence tomography data. We feed back forces estimated at the needle tip for real time haptic feedback and robot control. When inserting the needle at constant velocity, the force change estimated at the tip when penetrating tissue layers is up to 94 % between deep tissue layers compared to the force change at the needle handle of 2.36 %. Collaborative needle insertion results in more sensible force change at tissue transitions with haptic feedback from the tip (49.79 ± 25.51) % compared to the conventional shaft feedback (15.17 ± 15.92) %. Tissue transitions are more prominent when utilizing forces estimated at the needle tip compared to the forces at the needle shaft, indicating that a more informed advancement of the needle is possible with our system} } @phdthesis{11420_10432, author = {M. Schlüter}, title = {Analysis of ultrasound and optical coherence tomography for markerless volumetric image guidance in robotic radiosurgery.}, year = {2021}, doi = {10.15480/882.3798}, url = {http://hdl.handle.net/11420/10432}, type = {doctoralThesis}, school = {Technische Universität Hamburg}, keywords = {Medical Imaging; Ultrasound Imaging; Optical Coherence Tomography; Radiation Therapy; Tracking Systems; Motion Compensation;}, abstract = {An accurate dose delivery in radiosurgery requires to reliably detect and compensate any motion of the target during the treatment. In this thesis, we study approaches for markerless volumetric image guidance. For abdominal targets, we analyze and optimize the impact of robotic transabdominal ultrasound imaging. For cranial targets, we describe a novel setup using optical coherence tomography.} } @article{MielingSprengerLatusBargstenSchlaefer+2021+21+25, author = {R. Mieling and J. Sprenger and S. Latus and L. Bargsten and A. Schlaefer}, title = {A novel optical needle probe for deep learning-based tissue elasticity characterization:.}, journal = {Current Directions in Biomedical Engineering.}, year = {2021}, volume = {7.}, number = {(1),}, pages = {21-25}, doi = {doi:10.1515/cdbme-2021-1005}, url = {https://doi.org/10.1515/cdbme-2021-1005}, abstract = {The distinction between malignant and benign tumors is essential to the treatment of cancer. The tissue\'s elasticity can be used as an indicator for the required tissue characterization. Optical coherence elastography (OCE) probes have been proposed for needle insertions but have so far lacked the necessary load sensing capabilities. We present a novel OCE needle probe that provides simultaneous optical coherence tomography (OCT) imaging and load sensing at the needle tip. We demonstrate the application of the needle probe in indentation experiments on gelatin phantoms with varying gelatin concentrations. We further implement two deep learning methods for the end-to-end sample characterization from the acquired OCT data. We report the estimation of gelatin sample weight ratios [wt%] in unseen samples with a mean error of 1.21 ± 0.91 wt%. Both evaluated deep learning models successfully provide sample characterization with different advantages regarding the accuracy and inference time} } @article{GerlachNeidhardtWeißLavesStapperGromniakKniepMöbiu, author = {S. Gerlach and M. Neidhardt and T. Weiß and M.-H. Laves and C. Stapper and M. Gromniak and I. Kniep and D. Möbius and A. Heinemann and B. Ondruschka and A. Schlaefer}, title = {Needle insertion planning for obstacle avoidance in robotic biopsy.}, journal = {Current Directions in Biomedical Engineering.}, year = {2021}, volume = {7.}, number = {(2),}, pages = {779--782}, doi = {doi:10.1515/cdbme-2021-2199}, url = {https://doi.org/10.1515/cdbme-2021-2199}, abstract = {Understanding the underlying pathology in different tissues and organs is crucial when fighting pandemics like COVID-19. During conventional autopsy, large tissue sample sets of multiple organs can be collected from cadavers. However, direct contact with an infectious corpse is associated with the risk of disease transmission and relatives of the deceased might object to a conventional autopsy. To overcome these drawbacks, we consider minimally invasive autopsies with robotic needle placement as a practical alternative. One challenge in needle based biopsies is avoidance of dense obstacles, including bones or embedded medical devices such as pacemakers. We demonstrate an approach for automated planning and visualising suitable needle insertion points based on computed tomography (CT) scans. Needle paths are modeled by a line between insertion and target point and needle insertion path occlusion from obstacles is determined by using central projections from the biopsy target to the surface of the skin. We project the maximum and minimum CT attenuation, insertion depth, and standard deviation of CT attenuation along the needle path and create two-dimensional intensity-maps projected on the skin. A cost function considering these metrics is introduced and minimized to find an optimal biopsy needle path. Furthermore, we disregard insertion points without sufficient room for needle placement. For visualisation, we display the color-coded cost function so that suitable points for needle insertion become visible. We evaluate our system on 10 post mortem CTs with six biopsy targets in abdomen and thorax annotated by medical experts. For all patients and targets an optimal insertion path is found. The mean distance to the target ranges from (49.9 ± 12.9) mm for the spleen to (90.1 ± 25.8) mm for the pancreas } } @article{9366987, author = {S. Latus and J. Sprenger and M. Neidhardt and J. Schadler and A. Ron and A. Fitzek and M. Schlüter and P. Breitfeld and A. Heinemann and K. Püschel and A. Schlaefer}, title = {Rupture detection during needle insertion using complex OCT data and CNNs.}, journal = {IEEE Transactions on Biomedical Engineering.}, year = {2021}, volume = {68.}, number = {(10),}, pages = {3059-3067}, month = {October}, isbn = {1558-2531}, doi = {10.1109/TBME.2021.3063069}, keywords = {Needles;Phantoms;Shafts;Sensors;Probes;Force;Optical sensors;Needle Navigation;Optical Coherence Tomography;Relative Tissue Motion;Deep Learning}, abstract = {Objective: Soft tissue deformation and ruptures complicate needle placement. However, ruptures at tissue inter- faces also contain information which helps physicians to navigate through different layers. This navigation task can be challenging, whenever ultrasound (US) image guidance is hard to align and externally sensed forces are superimposed by friction. Methods: We propose an experimental setup for reproducible needle insertions, applying optical coherence tomography (OCT) directly at the needle tip as well as external US and force measurements. Processing the complex OCT data is challenging as the penetration depth is limited and the data can be difficult to interpret. Using a machine learning approach, we show that ruptures can be detected in the complex OCT data without additional external guidance or measurements after training with multi-modal ground-truth from US and force. Results: We can detect ruptures with accuracies of 0.94 and 0.91 on homogeneous and inhomogeneous phantoms, respectively, and 0.71 for ex-situ tissues. Conclusion: We propose an experimental setup and deep learning based rupture detection for the complex OCT data in front of the needle tip, even in deeper tissue structures without the need for US or force sensor guiding. Significance: This study promises a suitable approach to comple- ment a robust robotic needle placement} } @article{2021, author = {S. Lehmann, and A. Rogalla and M. Neidhardt and A. Schlaefer S. and Schupp}, title = {Online Strategy Synthesis for Safe and Optimized Control of Steerable Needles.}, journal = {Electronic Proceedings in Theoretical Computer Science.}, year = {2021}, volume = {348.}, pages = {128-135}, month = {Oct}, publisher = {Open Publishing Association:}, doi = {10.4204/eptcs.348.9}, url = {http://dx.doi.org/10.4204/EPTCS.348.9}, abstract = {Autonomous systems are often applied in uncertain environments, which require prospective action planning and retrospective data evaluation for future planning to ensure safe operation. Formal approaches may support these systems with safety guarantees, but are usually expensive and do not scale well with growing system complexity. In this paper, we introduce online strategy synthesis based on classical strategy synthesis to derive formal safety guarantees while reacting and adapting to environment changes. To guarantee safety online, we split the environment into region types which determine the acceptance of action plans and trigger local correcting actions. Using model checking on a frequently updated model, we can then derive locally safe action plans (prospectively), and match the current model against new observations via reachability checks (retrospectively). As use case, we successfully apply online strategy synthesis to medical needle steering, i.e., navigating a (flexible and beveled) needle through tissue towards a target without damaging its surroundings.} } @inproceedings{mars2020, author = {A. Rogalla and S. Lehmann and M. Neidhardt and J. Sprenger and M. Bengs and A. Schlaefer and S. Schupp}, title = {Synthesizing Strategies for Needle Steering in Gelatin Phantoms.}, journal = {MARS 2020.}, year = {2020}, booktitle = {Models for Formal Analysis of Real Systems (MARS 2020)}, doi = {10.4204/EPTCS.316.10}, url = {http://hdl.handle.net/11420/6107}, abstract = {In medicine, needles are frequently used to deliver treatments to subsurface targets or to take tissue samples from the inside of an organ. Current clinical practice is to insert needles under image guidance or haptic feedback, although that may involve reinsertions and adjustments since the needle and its interaction with the tissue during insertion cannot be completely controlled. (Automated) needle steering could in theory improve the accuracy with which a target is reached and thus reduce surgical traumata especially for minimally invasive procedures, e.g., brachytherapy or biopsy. Yet, flexible needles and needle-tissue interaction are both complex and expensive to model and can often be computed approximatively only. In this paper we propose to employ timed games to navigate flexible needles with a bevel tip to reach a fixed target in tissue. We use a simple non-holonomic model of needle-tissue interaction, which abstracts in particular from the various physical forces involved and appears to be simplistic compared to related models from medical robotics. Based on the model, we synthesize strategies from which we can derive sufficiently precise motion plans to steer the needle in soft tissue. However, applying those strategies in practice, one is faced with the problem of an unpredictable behavior of the needle at the initial insertion point. Our proposal is to implement a preprocessing step to initialize the model based on data from the real system, once the needle is inserted. Taking into account the actual needle tip angle and position, we generate strategies to reach the desired target. We have implemented the model in Uppaal Stratego and evaluated it on steering a flexible needle in gelatin phantoms; gelatin phantoms are commonly used in medical technology to simulate the behavior of soft tissue. The experiments show that strategies can be synthesized for both generated and measured needle motions with a maximum deviation of 1.84mm} } @inproceedings{rogalla2020, author = {A. Rogalla and T. Kamph and U. Bulmann and K. Billerbeck and M. Blumreiter and S. Schupp}, title = {Designing And Analyzing Open Application-Oriented Labs in Software-Verification Education.}, journal = {Proceedings of the 48th Annual SEFI Conference.}, year = {2020}, volume = {49.}, pages = {444-453}, month = {September}, booktitle = {Annual Conference of European Society for Engineering Education (SEFI). Enschede (the Netherlands)}, organization = {Annual Conference of European Society for Engineering Education (SEFI). Enschede (the Netherlands)}, abstract = {The daily work of a software engineer frequently includes the design and implementation of systems in non-software-engineering disciplines, like medical technology, often in interdisciplinary teams. In order to successfully select and apply the appropriate theoretical concept to perform the task, it is necessary to understand the actual problem, possibly outside one’s personal subject area, and to find an appropriate abstraction. However, software-engineering education often focuses on the theoretical concepts alone, ignoring the necessary skills to solve interdisciplinary tasks. We argue that the use of open, application-oriented labs creates a synergetic effect in understanding of theoretical concepts and the ability to apply them to solve practical issues. The subject of this study is a lab in the master’s program module “Software Verification” at a German university of technology. Therein, student groups solve openly-formulated, application-oriented modeling tasks in the field of medical technology. In this paper, we present the design and an analysis of this lab by means of a student questionnaire after completion of the lab and a document analysis of 32 laboratory reports. Our survey results show that more than 90 % of the respondents state that the practical labs helped them to understand the theoretical content of the lectures. The evaluation of the lab reports shows that around half of the student groups were able to understand, abstract, and model the task correctly. We conclude that the inclusion of open, application-oriented labs in software-engineering education is beneficial to both, understanding of theoretical concepts and ability to solve interdisciplinary tasks} } @article{viscmed2020, author = {D.B. Ellebrecht and S. Latus and A. Schlaefer and T. Keck and N. Gessert}, title = {Towards an Optical Biopsy during Visceral Surgical Interventions.}, journal = {Visceral Medicine.}, year = {2020}, booktitle = {Visceral Medicine}, doi = {10.1159/000505938}, abstract = {Cancer will replace cardiovascular diseases as the most frequent cause of death. Therefore, the goals of cancer treatment are prevention strategies and early detection by cancer screening and ideal stage therapy. From an oncological point of view, complete tumor resection is a significant prognostic factor. Optical coherence tomography (OCT) and confocal laser microscopy (CLM) are two techniques that have the potential to complement intraoperative frozen section analysis as in vivo and real-time optical biopsies. Summary: In this review we present both procedures and review the progress of evaluation for intraoperative application in visceral surgery. For visceral surgery, there are promising studies evaluating OCT and CLM; however, application during routine visceral surgical interventions is still lacking. Key Message: OCT and CLM are not competing but complementary approaches of tissue analysis to intraoperative frozen section analysis. Although intraoperative application of OCT and CLM is at an early stage, they are two promising techniques of intraoperative in vivo and real-time tissue examination. Additionally, deep learning strategies provide a significant supplement for automated tissue detection} } @article{Behrend-2020, author = {F. Behrendt and N. Gessert and A. Schlaefer}, title = {Generalization of spatio-temporal deep learning for vision-based force estimation.}, journal = {Current Directions in Biomedical Engineering.}, year = {2020}, volume = {6.}, number = {(1),}, pages = {20200024}, month = {May}, publisher = {De Gruyter:}, address = {Berlin, Boston}, doi = {https://doi.org/10.1515/cdbme-2020-0024}, url = {https://www.degruyter.com/view/journals/cdbme/6/1/article-20200024.xml}, abstract = {Robot-assisted minimally-invasive surgery is increasingly used in clinical practice. Force feedback offers potential to develop haptic feedback for surgery systems. Forces can be estimated in a vision-based way by capturing deformation observed in 2D-image sequences with deep learning models. Variations in tissue appearance and mechanical properties likely influence force estimation methods’ generalization. In this work, we study the generalization capabilities of different spatial and spatio-temporal deep learning methods across different tissue samples. We acquire several data-sets using a clinical laparoscope and use both purely spatial and also spatio-temporal deep learning models. The results of this work show that generalization across different tissues is challenging. Nevertheless, we demonstrate that using spatio-temporal data instead of individual frames is valuable for force estimation. In particular, processing spatial and temporal data separately by a combination of a ResNet and GRU architecture shows promising results with a mean absolute error of 15.450 compared to 19.744 mN of a purely spatial CNN} } @article{10.1371/journal.pone.0230821, author = {F. Griese AND S. Latus AND M. Schlüter AND M. Graeser AND M. Lutz AND A. Schlaefer AND T. Knopp}, title = {In-Vitro MPI-guided IVOCT catheter tracking in real time for motion artifact compensation.}, journal = {PLOS ONE.}, year = {2020}, volume = {15.}, number = {(3),}, pages = {e0230821}, month = {03}, publisher = {Public Library of Science:}, doi = {10.1371/journal.pone.0230821}, url = {https://arxiv.org/abs/1911.12226}, abstract = {Purpose Using 4D magnetic particle imaging (MPI), intravascular optical coherence tomography (IVOCT) catheters are tracked in real time in order to compensate for image artifacts related to relative motion. Our approach demonstrates the feasibility for bimodal IVOCT and MPI in-vitro experiments. Material and methods During IVOCT imaging of a stenosis phantom the catheter is tracked using MPI. A 4D trajectory of the catheter tip is determined from the MPI data using center of mass sub-voxel strategies. A custom built IVOCT imaging adapter is used to perform different catheter motion profiles: no motion artifacts, motion artifacts due to catheter bending, and heart beat motion artifacts. Two IVOCT volume reconstruction methods are compared qualitatively and quantitatively using the DICE metric and the known stenosis length. Results The MPI-tracked trajectory of the IVOCT catheter is validated in multiple repeated measurements calculating the absolute mean error and standard deviation. Both volume reconstruction methods are compared and analyzed whether they are capable of compensating the motion artifacts. The novel approach of MPI-guided catheter tracking corrects motion artifacts leading to a DICE coefficient with a minimum of 86% in comparison to 58% for a standard reconstruction approach. Conclusions IVOCT catheter tracking with MPI in real time is an auspicious method for radiation free MPI-guided IVOCT interventions. The combination of MPI and IVOCT can help to reduce motion artifacts due to catheter bending and heart beat for optimized IVOCT volume reconstructions.} } @article{KRUGER2020102445, author = {J. Krüger and R. Opfer and N. Gessert and A.-C. Ostwaldt and P. Manogaran and H. H. Kitzler and A. Schlaefer and S. Schippling}, title = {Fully automated longitudinal segmentation of new or enlarged multiple sclerosis lesions using 3D convolutional neural networks.}, journal = {NeuroImage: Clinical.}, year = {2020}, volume = {28.}, pages = {102445}, doi = {https://doi.org/10.1016/j.nicl.2020.102445}, url = {http://www.sciencedirect.com/science/article/pii/S2213158220302825}, keywords = {Multiple sclerosis, Lesion activity, Convolutional neural network, U-net, Lesion segmentation}, abstract = {The quantification of new or enlarged lesions from follow-up MRI scans is an important surrogate of clinical disease activity in patients with multiple sclerosis (MS). Not only is manual segmentation time consuming, but inter-rater variability is high. Currently, only a few fully automated methods are available. We address this gap in the field by employing a 3D convolutional neural network (CNN) with encoder-decoder architecture for fully automatic longitudinal lesion segmentation. Input data consist of two fluid attenuated inversion recovery (FLAIR) images (baseline and follow-up) per patient. Each image is entered into the encoder and the feature maps are concatenated and then fed into the decoder. The output is a 3D mask indicating new or enlarged lesions (compared to the baseline scan). The proposed method was trained on 1809 single point and 1444 longitudinal patient data sets and then validated on 185 independent longitudinal data sets from two different scanners. From the two validation data sets, manual segmentations were available from three experienced raters, respectively. The performance of the proposed method was compared to the open source Lesion Segmentation Toolbox (LST), which is a current state-of-art longitudinal lesion segmentation method. The mean lesion-wise inter-rater sensitivity was 62%, while the mean inter-rater number of false positive (FP) findings was 0.41 lesions per case. The two validated algorithms showed a mean sensitivity of 60% (CNN), 46% (LST) and a mean FP of 0.48 (CNN), 1.86 (LST) per case. Sensitivity and number of FP were not significantly different (p } } @article{Bargsten2020, author = {L. Bargsten and A. Schlaefer}, title = {SpeckleGAN: a generative adversarial network with an adaptive speckle layer to augment limited training data for ultrasound image processing.}, journal = {International Journal of Computer Assisted Radiology and Surgery.}, year = {2020}, volume = {15.}, number = {(9),}, pages = {1427-1436}, month = {Sept}, booktitle = {International Journal of Computer Assisted Radiology and Surgery}, doi = {10.1007/s11548-020-02203-1}, url = {https://doi.org/10.1007/s11548-020-02203-1}, abstract = {In the field of medical image analysis, deep learning methods gained huge attention over the last years. This can be explained by their often improved performance compared to classic explicit algorithms. In order to work well, they need large amounts of annotated data for supervised learning, but these are often not available in the case of medical image data. One way to overcome this limitation is to generate synthetic training data, e.g., by performing simulations to artificially augment the dataset. However, simulations require domain knowledge and are limited by the complexity of the underlying physical model. Another method to perform data augmentation is the generation of images by means of neural networks} } @inproceedings{bengs2020deep, author = {M. Bengs and N. Gessert and A. Schlaefer}, title = {A Deep Learning Approach for Motion Forecasting Using 4D OCT Data.}, year = {2020}, pages = {2004.10121}, booktitle = {International Conference on Medical Imaging with Deep Learning}, url = {https://arxiv.org/abs/2004.10121}, abstract = {Forecasting motion of a specific target object is a common problem for surgical interventions, e.g. for localization of a target region, guidance for surgical interventions, or motion compensation. Optical coherence tomography (OCT) is an imaging modality with a high spatial and temporal resolution. Recently, deep learning methods have shown promising performance for OCT\-based motion estimation based on two volumetric images. We extend this approach and investigate whether using a time series of volumes enables motion forecasting. We propose 4D spatio-temporal deep learning for end-to-end motion forecasting and estimation using a stream of OCT volumes. We design and evaluate five different 3D and 4D deep learning methods using a tissue data set. Our best performing 4D method achieves motion forecasting with an overall average correlation coefficient of 97.41\%, while also improving motion estimation performance by a factor of 2.5 compared to a previous 3D approach.} } @misc{https://doi.org/10.48550/arxiv.2004.10165, author = {M. Bengs and N. Gessert and A. Schlaefer}, title = {4D Spatio-Temporal Deep Learning with 4D fMRI Data for Autism Spectrum Disorder Classification.}, year = {2020}, publisher = {arXiv:}, doi = {10.48550/ARXIV.2004.10165}, url = {https://arxiv.org/abs/2004.10165}, keywords = {Image and Video Processing (eess.IV), Computer Vision and Pattern Recognition (cs.CV), FOS: Electrical engineering, electronic engineering, information engineering, FOS: Electrical engineering, electronic engineering, information engineering, FOS: Computer and information sciences, FOS: Computer and information sciences}, abstract = {Autism spectrum disorder (ASD) is associated with behavioral and communication problems. Often, functional magnetic resonance imaging (fMRI) is used to detect and characterize brain changes related to the disorder. Recently, machine learning methods havecbeen employed to reveal new patterns by trying to classify ASD from spatio-temporal fMRIcimages. Typically, these methods have either focused on temporal or spatial informationcprocessing. Instead, we propose a 4D spatio-temporal deep learning approach for ASD classification where we jointly learn from spatial and temporal data. We employ 4D convolutional neural networks and convolutional-recurrent models which outperform a previous approach with an F1-score of 0.71 compared to an F1-score of 0.65.} } @article{bengs2020spatiotemporal, author = {M. Bengs and N. Gessert and M. Schlüter and A. Schlaefer}, title = {Spatio-Temporal Deep Learning Methods for Motion Estimation Using 4D OCT Image Data.}, journal = {International Journal of Computer Assisted Radiology and Surgery.}, year = {2020}, volume = {15.}, number = {(6),}, pages = {943-952}, month = {Jun}, booktitle = {International Journal of Computer Assisted Radiology and Surgery}, doi = {10.1007/s11548-020-02178-z}, url = {https://arxiv.org/abs/2004.10114}, abstract = {Purpose. Localizing structures and estimating the motion of a specific target region are common problems for navigation during surgical interventions. Optical coherence tomography (OCT) is an imaging modality with a high spatial and temporal resolution that has been used for intraoperative imaging and also for motion estimation, for example, in the context of ophthalmic surgery or cochleostomy. Recently, motion estimation between a template and a moving OCT image has been studied with deep learning methods to overcome the shortcomings of conventional, feature\-based methods. Methods. We investigate whether using a temporal stream of OCT image volumes can improve deep learning\-based motion estimation performance. For this purpose, we design and evaluate several 3D and 4D deep learning methods and we propose a new deep learning approach. Also, we propose a temporal regularization strategy at the model output. Results. Using a tissue dataset without additional markers, our deep learning methods using 4D data outperform previous approaches. The best performing 4D architecture achieves an correlation coefficient (aCC) of 98.58\% compared to 85.0\% of a previous 3D deep learning method. Also, our temporal regularization strategy at the output further improves 4D model performance to an aCC of 99.06\%. In particular, our 4D method works well for larger motion and is robust towards image rotations and motion distortions. Conclusions. We propose 4D spatio\-temporal deep learning for OCT\-based motion estimation. On a tissue dataset, we find that using 4D information for the model input improves performance while maintaining reasonable inference times. Our regularization strategy demonstrates that additional temporal information is also beneficial at the model output. } } @inproceedings{10.1007/978-3-030-59716-0_66, author = {M. Bengs and N. Gessert and W. Laffers and D. Eggert and S. Westermann and N.A. Mueller and A.O.H. Gerstners and C. Betz and A. Schlaefer}, title = {Spectral-spatial Recurrent-Convolutional Networks for In-Vivo Hyperspectral Tumor Type Classification.}, year = {2020}, pages = {690-699}, publisher = {Springer International Publishing:}, address = {Cham}, isbn = {978-3-030-59716-0}, booktitle = {Medical Image Computing and Computer Assisted Intervention - MICCAI 2020}, abstract = {Early detection of cancerous tissue is crucial for long-term patient survival. In the head and neck region, a typical diagnostic procedure is an endoscopic intervention where a medical expert manually assesses tissue using RGB camera images. While healthy and tumor regions are generally easier to distinguish, differentiating benign and malignant tumors is very challenging. This requires an invasive biopsy, followed by histological evaluation for diagnosis. Also, during tumor resection, tumor margins need to be verified by histological analysis. To avoid unnecessary tissue resection, a non-invasive, image-based diagnostic tool would be very valuable. Recently, hyperspectral imaging paired with deep learning has been proposed for this task, demonstrating promising results on ex-vivo specimens. In this work, we demonstrate the feasibility of in-vivo tumor type classification using hyperspectral imaging and deep learning. We analyze the value of using multiple hyperspectral bands compared to conventional RGB images and we study several machine learning models\' ability to make use of the additional spectral information. Based on our insights, we address spectral and spatial processing using recurrent-convolutional models for effective spectral aggregating and spatial feature learning. Our best model achieves an AUC of 76.3% significantly outperforming previous conventional and deep learning methods} } @inproceedings{10.1117/12.2549251, author = {M. Bengs and S. Westermann and N. Gessert and D. Eggert and A. O. H. Gerstner and N. A. Mueller and C. Betz and W. Laffers and A. Schlaefer}, title = {Spatio-spectral deep learning methods for in-vivo hyperspectral laryngeal cancer detection.}, year = {2020}, volume = {11314.}, pages = {113141L}, editor = {In Horst K. Hahn and Maciej A. Mazurowski (Eds.)}, publisher = {SPIE:}, booktitle = {Medical Imaging 2020: Computer-Aided Diagnosis}, organization = {International Society for Optics and Photonics}, doi = {10.1117/12.2549251}, url = {https://doi.org/10.1117/12.2549251}, keywords = {hyperspectral imaging, convolutional neural networks, optical biopsy, intraoperative imaging, head and neck cancer}, abstract = {Early detection of head and neck tumors is crucial for patient survival. Often, diagnoses are made based on endoscopic examination of the larynx followed by biopsy and histological analysis, leading to a high interobserver variability due to subjective assessment. In this regard, early non-invasive diagnostics independent of the clinician would be a valuable tool. A recent study has shown that hyperspectral imaging (HSI) can be used for non-invasive detection of head and neck tumors, as precancerous or cancerous lesions show specific spectral signatures that distinguish them from healthy tissue. However, HSI data processing is challenging due to high spectral variations, various image interferences, and the high dimensionality of the data. Therefore, performance of automatic HSI analysis has been limited and so far, mostly ex-vivo studies have been presented with deep learning. In this work, we analyze deep learning techniques for in-vivo hyperspectral laryngeal cancer detection. For this purpose we design and evaluate convolutional neural networks (CNNs) with 2D spatial or 3D spatio-spectral convolutions combined with a state-of-the-art Densenet architecture. For evaluation, we use an in-vivo data set with HSI of the oral cavity or oropharynx. Overall, we present multiple deep learning techniques for in-vivo laryngeal cancer detection based on HSI and we show that jointly learning from the spatial and spectral domain improves classification accuracy notably. Our 3D spatio-spectral Densenet achieves an average accuracy of 81%.} } @conference{bengs-2020-1, author = {M. Bengs and S. Westermann and N. Gessert and D. Eggert and A. O. H. Gerstner, N. A. Mueller and C. Betz and W. Laffers and A. Schlaefer}, title = {Spatio-spectral deep learning methods for in-vivohyperspectral laryngeal cancer detection.}, journal = {SPIE Medical Imaging 2020: Computer-Aided Diagnosis.}, year = {2020}, pages = {in print}, booktitle = {SPIE Medical Imaging 2020: Computer-Aided Diagnosis} } @article{11420_7722, author = {M. Bengs and T. Gessert and A. Schlaefer}, title = {4D spatio-temporal convolutional networks for object position estimation in OCT volumes.}, journal = {Current directions in biomedical engineering.}, year = {2020}, volume = {6.}, number = {(1),}, pages = {20200001}, doi = {10.15480/882.3036}, url = {http://hdl.handle.net/11420/7722}, keywords = {convolutional neural networks; optical coherence tomography; position estimation; spatio-temporal data;}, abstract = {Tracking and localizing objects is a central problem in computer-assisted surgery. Optical coherence tomography (OCT) can be employed as an optical tracking system, due to its high spatial and temporal resolution. Recently, 3D convolutional neural networks (CNNs) have shown promising performance for pose estimation of a marker object using single volumetric OCT images. While this approach relied on spatial information only, OCT allows for a temporal stream of OCT image volumes capturing the motion of an object at high volumes rates. In this work, we systematically extend 3D CNNs to 4D spatio-temporal CNNs to evaluate the impact of additional temporal information for marker object tracking. Across various architectures, our results demonstrate that using a stream of OCT volumes and employing 4D spatio-temporal convolutions leads to a 30% lower mean absolute error compared to single volume processing with 3D CNNs} } @article{NeedleplacementaccuracyinCTguidedroboticpostmortem, author = {M. Gromniak and M. Neidhardt and A. Heinemann and K. Püschel and A. Schlaefer}, title = {Needle placement accuracy in CT-guided robotic post mortem biopsy.}, journal = {Current Directions in Biomedical Engineering.}, year = {2020}, volume = {6.}, number = {(1),}, pages = {20200031}, month = {May}, publisher = {De Gruyter:}, address = {Berlin, Boston}, doi = {https://doi.org/10.1515/cdbme-2020-0031}, url = {https://www.degruyter.com/view/journals/cdbme/6/1/article-20200031.xml}, abstract = {Forensic autopsies include a thorough examination of the corpse to detect the source or alleged manner of death as well as to estimate the time since death. However, a full autopsy may be not feasible due to limited time, cost or ethical objections by relatives. Hence, we propose an automated minimal invasive needle biopsy system with a robotic arm, which does not require any online calibrations during a procedure. The proposed system can be easily integrated into the workflow of a forensic biopsy since the robot can be flexibly positioned relative to the corpse. With our proposed system, we performed needle insertions into wax phantoms and livers of two corpses and achieved an accuracy of 4.34 ± 1.27 mm and 10.81 ± 4.44 mm respectively} } @article{, author = {M. Gromniak and N. Gessert and T. Saathoff and A. Schlaefer}, title = {Needle tip force estimation by deep learning from raw spectral OCT data.}, journal = {International Journal of Computer Assisted Radiology and Surgery.}, year = {2020}, volume = {15.}, pages = {1699-1702}, booktitle = {International Journal of Computer Assisted Radiology and Surgery}, doi = {10.1007/s11548-020-02224-w}, url = {https://link.springer.com/article/10.1007%2Fs11548-020-02224-w}, abstract = {Purpose Needle placement is a challenging problem for applications such as biopsy or brachytherapy. Tip force sensing can provide valuable feedback for needle navigation inside the tissue. For this purpose, fiber\-optical sensors can be directly integrated into the needle tip. Optical coherence tomography \(OCT\) can be used to image tissue. Here, we study how to calibrate OCT to sense forces, e.g., during robotic needle placement. Methods We investigate whether using raw spectral OCT data without a typical image reconstruction can improve a deep learning\-based calibration between optical signal and forces. For this purpose, we consider three different needles with a new, more robust design which are calibrated using convolutional neural networks \(CNNs\). We compare training the CNNs with the raw OCT signal and the reconstructed depth profiles. Results We find that using raw data as an input for the largest CNN model outperforms the use of reconstructed data with a mean absolute error of 5.81 mN compared to 8.04 mN. Conclusions We find that deep learning with raw spectral OCT data can improve learning for the task of force estimation. Our needle design and calibration approach constitute a very accurate fiber\-optical sensor for measuring forces at the needle tip} } @inproceedings{neidhard-cars-2020, author = {M. Neidhardt and M. Bengs and S. Latus and M. Schlüter and T. Saathoff and A. Schlaefer}, title = {4D Deep learning for real-time volumetric optical coherence elastography.}, journal = {International Journal of Computer Assisted Radiology and Surgery.}, year = {2020}, pages = {1861-6429}, booktitle = {International Journal of Computer Assisted Radiology and Surgery 2020}, doi = {10.1007/s11548-020-02261-5}, url = {https://doi.org/10.1007/s11548-020-02261-5}, abstract = {Elasticity of soft tissue provides valuable information to physicians during treatment and diagnosis of diseases. A number of approaches have been proposed to estimate tissue stiffness from the shear wave velocity. Optical coherence elastography offers a particularly high spatial and temporal resolution. However, current approaches typically acquire data at different positions sequentially, making it slow and less practical for clinical application} } @inproceedings{neidhardt20isbi, author = {M. Neidhardt and M. Bengs and S. Latus and M. Schlüter and T. Saathoff and A. Schlaefer}, title = {Deep Learning for High Speed Optical Coherence Elastography.}, year = {2020}, pages = {1583-1586}, booktitle = {IEEE International Symposium on Biomedical Imaging}, doi = {10.1109/ISBI45749.2020.9098422}, abstract = {Mechanical properties of tissue provide valuable information for identifying lesions. One approach to obtain quantitative estimates of elastic properties is shear wave elastography with optical coherence elastography (OCE). However, given the shear wave velocity, it is still difficult to estimate elastic properties. Hence, we propose deep learning to directly predict elastic tissue properties from OCE data. We acquire 2D images with a frame rate of 30 kHz and use convolutional neural networks to predict gelatin concentration, which we use as a surrogate for tissue elasticity. We compare our deep learning approach to predictions from conventional regression models, using the shear wave velocity as a feature. Mean absolut prediction errors for the conventional approaches range from 1.32±0.98 p.p. to 1.57±1.30 p.p. whereas we report an error of 0.90±0.84 p.p. for the convolutional neural network with 3D spatio-temporal input. Our results indicate that deep learning on spatio-temporal data outperforms elastography based on explicit shear wave velocity estimation.} } @article{Neidhardt-cdibe, author = {M. Neidhardt and N. Gessert and T. Gosau and J. Kemmling and S. Feldhaus and U. Schumacher and A. Schlaefer}, title = {Force estimation from 4D OCT data in a human tumor xenograft mouse model.}, journal = {Current Directions in Biomedical Engineering.}, year = {2020}, volume = {6.}, number = {(1),}, pages = {20200022}, month = {May}, publisher = {De Gruyter:}, address = {Berlin, Boston}, doi = {https://doi.org/10.1515/cdbme-2020-0022}, url = {https://www.degruyter.com/view/journals/cdbme/6/1/article-20200022.xml}, abstract = {Minimally invasive robotic surgery offer benefits such as reduced physical trauma, faster recovery and lesser pain for the patient. For these procedures, visual and haptic feedback to the surgeon is crucial when operating surgical tools without line-of-sight with a robot. External force sensors are biased by friction at the tool shaft and thereby cannot estimate forces between tool tip and tissue. As an alternative, vision-based force estimation was proposed. Here, interaction forces are directly learned from deformation observed by an external imaging system. Recently, an approach based on optical coherence tomography and deep learning has shown promising results. However, most experiments are performed on ex-vivo tissue. In this work, we demonstrate that models trained on dead tissue do not perform well in in vivo data. We performed multiple experiments on a human tumor xenograft mouse model, both on in vivo, perfused tissue and dead tissue. We compared two deep learning models in different training scenarios. Training on perfused, in vivo data improved model performance by 24% for in vivo force estimation} } @inproceedings{schlueter20isbi, author = {M. Schlüter and L. Glandorf and J. Sprenger and M. Gromniak and M. Neidhardt and T. Saathoff and A. Schlaefer}, title = {High-Speed Markerless Tissue Motion Tracking Using Volumetric Optical Coherence Tomography Images.}, year = {2020}, pages = {1979-1982}, booktitle = {IEEE International Symposium on Biomedical Imaging}, doi = {10.1109/ISBI45749.2020.9098448}, abstract = {Modern optical coherence tomography (OCT) devices provide volumetric images with micrometer-scale spatial resolution and a temporal resolution beyond video rate. In this work, we analyze an OCT-based prototypical tracking system which processes 831 volumes per second, estimates translational motion, and automatically adjusts the field-of-view, which has a size of few millimeters, to follow a sample even along larger distances. The adjustment is realized by two galvo mirrors and a motorized reference arm, such that no mechanical movement of the scanning setup is necessary. Without requiring a marker or any other knowledge about the sample, we demonstrate that reliable tracking of velocities up to 25 mm s -1 is possible with mean tracking errors in the order of 0.25 mm. Further, we report successful tracking of lateral velocities up to 70 mm s -1 with errors below 0.3 mm} } @article{Schlueter20b, author = {M. Schlüter and L. Glandorf and M. Gromniak and T. Saathoff and A. Schlaefer}, title = {Concept for Markerless 6D Tracking Employing Volumetric Optical Coherence Tomography.}, journal = {Sensors.}, year = {2020}, volume = {20.}, number = {(9),}, pages = {2678}, doi = {10.3390/s20092678}, abstract = {Optical tracking systems are widely used, for example, to navigate medical interventions. Typically, they require the presence of known geometrical structures, the placement of artificial markers, or a prominent texture on the target’s surface. In this work, we propose a 6D tracking approach employing volumetric optical coherence tomography (OCT) images. OCT has a micrometer-scale resolution and employs near-infrared light to penetrate few millimeters into, for example, tissue. Thereby, it provides sub-surface information which we use to track arbitrary targets, even with poorly structured surfaces, without requiring markers. Our proposed system can shift the OCT’s field-of-view in space and uses an adaptive correlation filter to estimate the motion at multiple locations on the target. This allows one to estimate the target’s position and orientation. We show that our approach is able to track translational motion with root-mean-squared errors below 0.25mm and in-plane rotations with errors below 0.3°. For out-of-plane rotations, our prototypical system can achieve errors around 0.6°} } @article{Bargsten_2020, author = {M. Seemann and L. Bargsten and A. Schlaefer}, title = {Data augmentation for computed tomography angiography via synthetic image generation and neural domain adaptation.}, journal = {Current Directions in Biomedical Engineering.}, year = {2020}, volume = {6.}, number = {(1),}, pages = {20200015}, publisher = {De Gruyter:}, address = {Berlin, Boston}, doi = {https://doi.org/10.1515/cdbme-2020-0015}, url = {https://www.degruyter.com/view/journals/cdbme/6/1/article-20200015.xml}, abstract = {Deep learning methods produce promising results when applied to a wide range of medical imaging tasks, including segmentation of artery lumen in computed tomography angiography (CTA) data. However, to perform sufficiently, neural networks have to be trained on large amounts of high quality annotated data. In the realm of medical imaging, annotations are not only quite scarce but also often not entirely reliable. To tackle both challenges, we developed a two-step approach for generating realistic synthetic CTA data for the purpose of data augmentation. In the first step moderately realistic images are generated in a purely numerical fashion. In the second step these images are improved by applying neural domain adaptation. We evaluated the impact of synthetic data on lumen segmentation via convolutional neural networks (CNNs) by comparing resulting performances. Improvements of up to 5% in terms of Dice coefficient and 20% for Hausdorff distance represent a proof of concept that the proposed augmentation procedure can be used to enhance deep learning-based segmentation for artery lumen in CTA images} } @misc phdthesis{11420_8296, author = {N. Gessert}, title = {Deep learning with multi-dimensional medical image data.}, journal = {TUHH Open Research.}, year = {2020}, month = {Dec}, publisher = {TUHH Open Research:}, address = {Hamburg, Germany}, booktitle = {TUHH Open Research}, doi = {10.15480/882.3216}, url = {http://hdl.handle.net/11420/8296}, type = {doctoralThesis}, school = {Technische Universität Hamburg}, keywords = {medical imaging; Deep learning; machine learning; Optical coherence tomography; Magnetic resonance imaging;}, abstract = {In this work, we explore deep learning model design and application in the context of multi-dimensional data in medical image analysis. A lot of medical image analysis problems come with 3D or even 4D spatio-temporal data that requires appropriate processing. While higher-dimensional processing allows for exploiting a lot of context, model design becomes very challenging due to exponentially increasing model parameters and risk of overfitting. Therefore, we design a variety of deep learning models for low- and high-dimensional data processing, including 1D up to 4D convolutional neural networks, convolutional-recurrent models, and Siamese architectures. Across a large number of applications, we find that using high-dimensional data is often effective when using well-designed deep learning models.} } @article{2019arXiv190804181G, author = {N. Gessert and A. Schlaefer}, title = {Left Ventricle Quantification Using Direct Regression with Segmentation Regularization and Ensembles of Pretrained 2D and 3D CNNs.}, journal = {arXiv e-prints.}, year = {2020}, pages = {Statistical Atlases and Computational Models of the Heart. Multi-Sequence CMR Segmentation, CRT-EPiggy and LV Full Quantification Challenges. STACOM@MICCAI 2019. Lecture Notes in Computer Science. 375-383}, month = {Aug}, doi = {10.1007/978-3-030-39074-7_39}, url = {https://arxiv.org/abs/1908.04181}, keywords = {Electrical Engineering and Systems Science - Image and Video Processing, Computer Science - Computer Vision and Pattern Recognition}, abstract = {Cardiac left ventricle (LV) quantification provides a tool for diagnosing cardiac diseases. Automatic calculation of all relevant LV indices from cardiac MR images is an intricate task due to large variations among patients and deformation during the cardiac cycle. Typical methods are based on segmentation of the myocardium or direct regression from MR images. To consider cardiac motion and deformation, recurrent neural networks and spatio\-temporal convolutional neural networks (CNNs) have been proposed. We study an approach combining stateof\-the\-art models and emphasizing transfer learning to account for the small dataset provided for the LVQuan19 challenge. We compare 2D spatial and 3D spatio\-temporal CNNs for LV indices regression and cardiac phase classification. To incorporate segmentation information, we propose an architecture\-independent segmentation\-based regularization. To improve the robustness further, we employ a search scheme that identifies the optimal ensemble from a set of architecture variants. Evaluating on the LVQuan19 Challenge training dataset with 5\-fold cross\-validation, we achieve mean absolute errors of 111 \± 76 mm2, 1:84 \± 0:90 mm and 1:22 \± 0:60 mm for area, dimension and regional wall thickness regression, respectively. The error rate for cardiac phase classification is 6:7 \%} } @article{ngessert2020-cmiag, author = {N. Gessert and J. Krüger and R. Opfer and A.-C. Ostwaldt and P. Manogaran and H. H. Kitzler and S. Schippling and A. Schlaefer}, title = {Multiple Sclerosis Lesion Activity Segmentation with Attention-Guided Two-Path CNNs.}, journal = {Computerized Medical Imaging and Graphics.}, year = {2020}, volume = {84.}, number = {(101772),}, month = {Sept}, booktitle = {Computerized Medical Imaging and Graphics}, doi = {10.1016/j.compmedimag.2020.101772}, url = {https://arxiv.org/abs/2008.02001}, abstract = {Multiple sclerosis is an inflammatory autoimmune demyelinating disease that is characterized by lesions in the central nervous system. Typically, magnetic resonance imaging \(MRI\) is used for tracking disease progression. Automatic image processing methods can be used to segment lesions and derive quantitative lesion parameters. So far, methods have focused on lesion segmentation for individual MRI scans. However, for monitoring disease progression, lesion activity in terms of new and enlarging lesions between two time points is a crucial biomarker. For this problem, several classic methods have been proposed, e.g., using difference volumes. Despite their success for single-volume lesion segmentation, deep learning approaches are still rare for lesion activity segmentation. In this work, convolutional neural networks \(CNNs\) are studied for lesion activity segmentation from two time points. For this task, CNNs are designed and evaluated that combine the information from two points in different ways. In particular, two-path architectures with attention-guided interactions are proposed that enable effective information exchange between the two time point\'s processing paths. It is demonstrated that deep learning\-based methods outperform classic approaches and it is shown that attention\-guided interactions significantly improve performance. Furthermore, the attention modules produce plausible attention maps that have a masking effect that suppresses old, irrelevant lesions. A lesion\-wise false positive rate of 26.4\% is achieved at a true positive rate of 74.2\%, which is not significantly different from the interrater performance} } @inproceedings{gessert2019melanoma, author = {N. Gessert and M. Bengs and A. Schlaefer}, title = {Melanoma detection with electrical impedance spectroscopy and dermoscopy using joint deep learning models.}, year = {2020}, pages = {in print}, booktitle = {SPIE Medical Imaging 2020}, url = {https://arxiv.org/abs/1911.02322v2}, keywords = {Computer Vision and Pattern Recognition }, abstract = {The initial assessment of skin lesions is typically based on dermoscopic images. As this is a difficult and timeconsuming task, machine learning methods using dermoscopic images have been proposed to assist human experts. Other approaches have studied electrical impedance spectroscopy \(EIS\) as a basis for clinical decision support systems. Both methods represent different ways of measuring skin lesion properties as dermoscopy relies on visible light and EIS uses electric currents. Thus, the two methods might carry complementary featuThe initial assessment of skin lesions is typically based on dermoscopic images. As this is a difficult and timeconsuming task, machine learning methods using dermoscopic images have been proposed to assist human experts. Other approaches have studied electrical impedance spectroscopy \(EIS\) as a basis for clinical decision support systems. Both methods represent different ways of measuring skin lesion properties as dermoscopy relies on visible light and EIS uses electric currents. Thus, the two methods might carry complementary features for lesion classification. Therefore, we propose joint deep learning models considering both EIS and dermoscopy for melanoma detection. For this purpose, we first study machine learning methods for EIS that incorporate domain knowledge and previously used heuristics into the design process. As a result, we propose a recurrent model with state\-max\-pooling which automatically learns the relevance of different EIS measurements. Second, we combine this new model with different convolutional neural networks that process dermoscopic images. We study ensembling approaches and also propose a cross\-attention module guiding information exchange between the EIS and dermoscopy model. In general, combinations of EIS and dermoscopy clearly outperform models that only use either EIS or dermoscopy. We show that our attention\-based, combined model outperforms other models with specificities of 34.4 \% \(CI 31.3\-38.4\), 34.7 \% \(CI 31.0\-38.8\) and 53.7 \% \(CI 50.1\-57.6\) for dermoscopy, EIS and the combined model, respectively, at a clinically relevant sensitivity of 98 \%} } @inproceedings{gessert2020d, author = {N. Gessert and M. Bengs and J. Krüger and R. Opfer and A.-C. Ostwaldt and P. Manogaran and S. Schippling and A. Schlaefer}, title = {4D Deep Learning for Multiple-Sclerosis Lesion Activity Segmentation.}, year = {2020}, pages = {accepted}, booktitle = {Medical Imaging with Deep Learning}, url = {https://openreview.net/forum?id=sMsAIWBSvg}, abstract = {Multiple sclerosis lesion activity segmentation is the task of detecting new and enlarging lesions that appeared between a baseline and a follow\-up brain MRI scan. While deep learning methods for single\-scan lesion segmentation are common, deep learning approaches for lesion activity have only been proposed recently. Here, a two\-path architecture processes two 3D MRI volumes from two time points. In this work, we investigate whether extending this problem to full 4D deep learning using a history of MRI volumes and thus an extended baseline can improve performance. For this purpose, we design a recurrent multi\-encoder\-decoder architecture for processing 4D data. We find that adding more temporal information is beneficial and our proposed architecture outperforms previous approaches with a lesion\-wise true positive rate of 0.84 at a lesion\-wise false positive rate of 0.19.} } @article{ngessert2020-mia, author = {N. Gessert and M. Bengs and M. Schlüter and A. Schlaefer }, title = {Deep learning with 4D spatio-temporal data representations for OCT-based force estimation.}, journal = {Medical Image Analysis.}, year = {2020}, volume = {64.}, number = {(101730),}, booktitle = {Medical Image Analysis}, pmid = {32492583}, doi = {10.1016/j.media.2020.101730}, url = {https://doi.org/10.1016/j.media.2020.101730}, abstract = {Estimating the forces acting between instruments and tissue is a challenging problem for robot-assisted minimally-invasive surgery. Recently, numerous vision-based methods have been proposed to replace electro-mechanical approaches. Moreover, optical coherence tomography (OCT) and deep learning have been used for estimating forces based on deformation observed in volumetric image data. The method demonstrated the advantage of deep learning with 3D volumetric data over 2D depth images for force estimation. In this work, we extend the problem of deep learning-based force estimation to 4D spatio-temporal data with streams of 3D OCT volumes. For this purpose, we design and evaluate several methods extending spatio-temporal deep learning to 4D which is largely unexplored so far. Furthermore, we provide an in-depth analysis of multi-dimensional image data representations for force estimation, comparing our 4D approach to previous, lower-dimensional methods. Also, we analyze the effect of temporal information and we study the prediction of short-term future force values, which could facilitate safety features. For our 4D force estimation architectures, we find that efficient decoupling of spatial and temporal processing is advantageous. We show that using 4D spatio-temporal data outperforms all previously used data representations with a mean absolute error of 10.7 mN. We find that temporal information is valuable for force estimation and we demonstrate the feasibility of force prediction} } @article{GESSERT2020100864, author = {N. Gessert and M. Nielsen and M. Shaikh and R. Werner and A. Schlaefer}, title = {Skin lesion classification using ensembles of multi-resolution EfficientNets with meta data.}, journal = {MethodsX.}, year = {2020}, volume = {7.}, pages = {100864}, doi = {https://doi.org/10.1016/j.mex.2020.100864}, url = {http://www.sciencedirect.com/science/article/pii/S2215016120300832}, keywords = {Deep Learning, Multi-class skin lesion classification, Convolutional neural networks}, abstract = {In this paper, we describe our method for the ISIC 2019 Skin Lesion Classification Challenge. The challenge comes with two tasks. For task 1, skin lesions have to be classified based on dermoscopic images. For task 2, dermoscopic images and additional patient meta data are used. Our deep learning\-based method achieved first place for both tasks. The are several problems we address with our method. First, there is an unknown class in the test set which we cover with a data\-driven approach. Second, there is a severe class imbalance that we address with loss balancing. Third, there are images with different resolutions which motivates two different cropping strategies and multi\-crop evaluation. Last, there is patient meta data available which we incorporate with a dense neural network branch. • We address skin lesion classification with an ensemble of deep learning models including EfficientNets, SENet, and ResNeXt WSL, selected by a search strategy. • We rely on multiple model input resolutions and employ two cropping strategies for training. We counter severe class imbalance with a loss balancing approach. • We predict an additional, unknown class with a data\-driven approach and we make use of patient meta data with an additional input branch.} } @article{8710336, author = {N. Gessert and T. Sentker and F. Madesta and R. Schmitz and H. Kniep and I. Baltruschat and R. Werner and A. Schlaefer}, title = {Skin Lesion Classification Using CNNs With Patch-Based Attention and Diagnosis-Guided Loss Weighting.}, journal = {IEEE Transactions on Biomedical Engineering.}, year = {2020}, volume = {67.}, number = {(2),}, pages = {495-503}, month = {Feb}, doi = {10.1109/TBME.2019.2915839}, url = {https://arxiv.org/abs/1905.02793}, keywords = {Lesions;Skin;Computer architecture;Medical diagnostic imaging;Image resolution;Sensitivity;Skin lesion classification;deep learning;attention;dermoscopy}, abstract = {Objective: This paper addresses two key problems of skin lesion classification. The first problem is the effective use of high-resolution images with pretrained standard architectures for image classification. The second problem is the high-class imbalance encountered in real-world multi-class datasets. Methods: To use high-resolution images, we propose a novel patch-based attention architecture that provides global context between small, high-resolution patches. We modify three pretrained architectures and study the performance of patch-based attention. To counter class imbalance problems, we compare oversampling, balanced batch sampling, and class-specific loss weighting. Additionally, we propose a novel diagnosis-guided loss weighting method that takes the method used for ground-truth annotation into account. Results: Our patch-based attention mechanism outperforms previous methods and improves the mean sensitivity by 7% . Class balancing significantly improves the mean sensitivity and we show that our diagnosis-guided loss weighting method improves the mean sensitivity by 3% over normal loss balancing. Conclusion: The novel patch-based attention mechanism can be integrated into pretrained architectures and provides global context between local patches while outperforming other patch-based methods. Hence, pretrained architectures can be readily used with high-resolution images without downsampling. The new diagnosis-guided loss weighting method outperforms other methods and allows for effective training when facing class imbalance. Significance: The proposed methods improve automatic skin lesion classification. They can be extended to other clinical applications where high-resolution image data and class imbalance are relevant} } @article{mieleng2020, author = {R. Mieling and S. Latus and N. Gessert and M. Lutz and A. Schlaefer }, title = {Deep learning-based rotation frequency estimation and NURD correction for IVOCT image data.}, journal = {(Suppl1) International Journal of CARS'2020.}, year = {2020}, volume = {15.}, number = {(1),}, pages = {162-163}, month = {June}, booktitle = {(Suppl1) International Journal of CARS'2020}, doi = {https://doi.org/10.1007/s11548-020-02171-6}, abstract = {Atherosclerotic plaque in coronary arteries can lead to myocardial infarction and is one of the leading causes of death. Intravascular optical coherence tomography (IVOCT) can be used to image the affected blood vessels for assessment and treatment. However, catheter bending often causes changes in the rotation frequency of the optical probe during acquisition. The resulting non-uniform rotation distortion (NURD) artefacts complicate the image interpretation and may affect the diagnosis. Deep learning methods have been proposed to analyze IVOCT image data, including plaque detection [1] and feature extraction [2]. We present a novel approach to directly estimate the rotation frequency of the optical probe from a sequence of IVOCT images. We illustrate that this allows a proper correction of NURD artefacts} } @article{gerlach-2020, author = {S. Gerlach and C. Fürweger and T. Hofmann and A. Schlaefer}, title = {Multicriterial CNN based beam generation for robotic radiosurgery of the prostate.}, journal = {Current Directions in Biomedical Engineering.}, year = {2020}, volume = {6.}, number = {(1),}, pages = {20200030}, month = {may}, publisher = {De Gruyter:}, address = {Berlin, Boston}, doi = {https://doi.org/10.1515/cdbme-2020-0030}, url = {https://www.degruyter.com/view/journals/cdbme/6/1/article-20200030.xml}, abstract = {Although robotic radiosurgery offers a flexible arrangement of treatment beams, generating treatment plans is computationally challenging and a time consuming process for the planner. Furthermore, different clinical goals have to be considered during planning and generally different sets of beams correspond to different clinical goals. Typically, candidate beams sampled from a randomized heuristic form the basis for treatment planning. We propose a new approach to generate candidate beams based on deep learning using radiological features as well as the desired constraints. We demonstrate that candidate beams generated for specific clinical goals can improve treatment plan quality. Furthermore, we compare two approaches to include information about constraints in the prediction. Our results show that CNN generated beams can improve treatment plan quality for different clinical goals, increasing coverage from 91.2 to 96.8% for 3,000 candidate beams on average. When including the clinical goal in the training, coverage is improved by 1.1% points} } @article{Gerlach-14331, author = {S. Gerlach and C. Fürweger and T. Hofmann and A. Schlaefer}, title = {Feasibility and analysis of CNN-based candidate beam generation for robotic radiosurgery.}, journal = {Medical Physics.}, year = {2020}, volume = {47.}, number = {(9),}, pages = {3806-3815}, doi = {https://doi.org/10.1002/mp.14331}, url = {https://aapm.onlinelibrary.wiley.com/doi/abs/10.1002/mp.14331}, keywords = {beam optimization, machine learning, radiation therapy, robotic radiosurgery, treatment planning}, abstract = {Purpose Robotic radiosurgery offers the flexibility of a robotic arm to enable high conformity to the target and a steep dose gradient. However, treatment planning becomes a computationally challenging task as the search space for potential beam directions for dose delivery is arbitrarily large. We propose an approach based on deep learning to improve the search for treatment beams. Methods In clinical practice, a set of candidate beams generated by a randomized heuristic forms the basis for treatment planning. We use a convolutional neural network to identify promising candidate beams. Using radiological features of the patient, we predict the influence of a candidate beam on the delivered dose individually and let this prediction guide the selection of candidate beams. Features are represented as projections of the organ structures which are relevant during planning. Solutions to the inverse planning problem are generated for random and CNN-predicted candidate beams. Results The coverage increases from 95.35\% to 97.67\% for 6000 heuristically and CNN-generated candidate beams, respectively. Conversely, a similar coverage can be achieved for treatment plans with half the number of candidate beams. This results in a patient-dependent reduced averaged computation time of 20.28\%–45.69\%. The number of active treatment beams can be reduced by 11.35\% on average, which reduces treatment time. Constraining the maximum number of candidate beams per beam node can further improve the average coverage by 0.75 percentage points for 6000 candidate beams. Conclusions We show that deep learning based on radiological features can substantially improve treatment plan quality, reduce computation runtime, and treatment time compared to the heuristic approach used in clinics} } @article{mp.14316, author = {S. Gerlach and F. Siebert and A. Schlaefer}, title = {BReP-SNAP-T-54: Efficient Stochastic Optimization Accounting for Uncertainty in HDR Prostate Brachytherapy Needle Placement.}, journal = {Medical Physics.}, year = {2020}, volume = {47.}, number = {(6),}, pages = {e458}, doi = {https://doi.org/10.1002/mp.14316}, url = {https://aapm.onlinelibrary.wiley.com/doi/abs/10.1002/mp.14316}, abstract = {Purpose: Uncertainty due to tissue deformation affects treatment planningfor HDR prostate brachytherapy. Hence, position and orientation of the nee-dles are typically not optimized in inverse planning. Stochastic linear pro-gramming (SLP) has been proposed to consider uncertainty duringoptimization. Conventionally, it draws samples from a probability distribu-tion but increases the problem size substantially. We propose an efficientscheme allowing for fast identification of robust needle configurations. Methods: We account for uncertainty along the needle axis by deformingthe target using B-Spline interpolation and a random displacement of thevoxel at the needle tip. Conventional SLP adds constraints for each sample.The new weighted SLP (WSLP) scheme first creates all spatial distributionsand then establishes one discretized optimization problem where weights inthe objective function represent the likelihood of voxels falling into grid ele-ments. Both approaches and the original deterministic problem are comparedon a set of 5 patient cases. Moreover, we use WSLP on a large set of ran-domly generated needles to select a robust subset of needles. Evaluations aredone on 100 independently sampled deformations. Results: Depending onthe deformation and needle count, SLP and WSLP improve the target cover-age by 1.5 to 10.9 percentage points compared to deterministic optimization.There is no significant difference in target coverage between plans for SLPand WSLP (p = 0.98) but WLSP is substantially more efficient, taking belowten seconds instead of more than four hours when considering 100 sampleddeformations. Using WSLP to identify robust needle configurations, cover-age can be improved 0.7 to 3.3 percentage points over the most promisingneedle configurations identified by deterministic optimization. Conclusion: WSLP allows for fast optimization considering a dense sample of possibledeformations. Using WSLP, it is feasible to realize inverse planning incorpo-rating uncertainty in needle placement and to identify robust needle sets} } @article{latus-euroana-2020, author = {S. Latus and P. Breitfeld and M. Neidhardt and W. Reip and C. Zöllner and A. Schlaefer}, title = {Boundary prediction during epidural punctures based on OCT relative motion analysis.}, journal = {EUR J ANAESTH.}, year = {2020}, volume = {2020.}, number = {(Volume 37 | e-Supplement 58 | June 2020),}, month = {6}, publisher = {Lippincott Williams and Wilkins:}, abstract = {1 Background and Goal of StudyUntil today, physicians mainly use their haptic impression for the correct po-sitioning of the needle during epidural punctures. ?Blind? techniques such asLoss-of-Resistance (LOR) and saline drop methods [1, 2] help to identify theepidural space, but the error rate is highly dependent on the physician?s exper-tise. The challenge of a direct needle insertion through different tissues into theepidural space with reliable identification of the same and without bone contactrequires a lot of experience from the performer. A penetration into the spinalcolumn space has to be absolutely avoided [3]. Different imaging modalities suchas ultrasound (US) [4] are used to assist needle navigation task in a few anesthe-sia cases. However, an external tracking of the needle path is limited due to theanatomy around the spinal cord. Hence, optical fibers are integrated in epiduralneedles to enable high-resolution optical coherence tomography (OCT) imagingof the punctured tissue [5]. Deep learning approaches based on morphologicalinformation of OCT intensity data facilitate an online identification of tissuestructures along the needle trajectory [6]. Furthermore, the fiber bragg gratings(FBG) are integrated to measure the forces during punctures [7]. In this study,we propose a novel method to determine both the morphological and mechanicproperties of tissues during epidural punctures. In addition to the intensity theOCT phase data is used to differentiate between tissue structures and identifyruptures and deformations in front of the needle tip in order to detect relevanttissue structures and boundaries.2 Material and MethodsWe perform ex-vivo epidural punctures in a pig cadaver model using an adaptedepidural Tuohy-needle with an integrated forward facing OCT fiber (Fig. 1).During manual punctures we capture ground-truth information of the needlepose and interacting forces at the needle shaft using an optical tracking system(fusion track 500, Atracsys) and a force-torque (FT) sensor (SRI, Sunrise In-struments), respectively. In addition, the physician feeds his haptic impression2S. Latus and P. Breitfeld(e.g. harder or softer resistance in tissues and the sensing of slight click by pen-etrating the ligamentum flavum) and expectation of the boundary interactionsback. We allocate one dimensional OCT depth scans (A-scans) throughout theFig. 1.Setup for OCT deformation analysis during ex-vivo epidural punctures. Theepidural needle is attached to a force torque (FT) sensor with associated trackingmarker. An optical fiber is integrated in the Tuohy-needle enabling a forward facingA-scan acquisitions (bottom right, red dashed line). A physician navigates the OCTneedle towards the epidural space, meanwhile OCT, FT, and tracking data is capturedsynchronously. During epidural puncture the needle needs to be navigated through skin(orange), supraspinous ligament, fat/muscle tissue, and ligamentum flavum (brown)preventing a rupture of the Dura (blue) or a collision with bone structures (gray).punctures applying a spectral domain OCT system (Telesto, Thorlabs) with aconstant A-scan frequency of 91 kHz.Using both the acquired OCT intensity and phase data, we are able to mag-nify the tissue properties in front of the needle towards the haptic feedbacksensed by the FT sensor and physician. We analyze the tissue interactions infront of the needle by means of the relative motion derived from the OCT phasedifferences [8]. An increase of the determined relative motion can be relatedto deformations and following ruptures at tissue boundaries, whereas negativevalues are associated to negative needle motion w.r.t. the tissue. We use thehaptic impression of the physician as ground-truth information for the detectedboundary interactions.Title Suppressed Due to Excessive Length33 Results and DiscussionExemplary, OCT intensity data is shown with overlayed relative motion (greenand red) and measured force in needle direction (blue line) for one epiduralpuncture (Fig. 2). In case of boundary deformations and following ruptures therelative motion increases rapidly (time points B, C, and E). Especially, duringthe puncture of the ligamentum flavum (E to F) multiple ruptures are detecteduntil a LOR is measured with the FT sensor. Between C and D several smallruptures appear due to sinews and muscle structures. In contrast, the highlightedtime points without tissue boundary ruptures (A and D) do not show an increaseof the estimated relative motion. After the bone contact (D) the needle is pulledbackwards and an obvious negative relative motion (red) follows.Fig. 2.OCT intensity and relative motion estimations related to the externally mea-sured forces for an exemplary epidural puncture. The relative motion is depicted ingreen and red for positive and negative values, respectively. The time points (A-F)are related to different needle-tissue interactions: A) Re-orientation of needle, B) firstrupture at skin, C) second rupture at supraspinous ligament, D) bone contact and fol-lowing needle re-orientation, E) start of ruptures at ligamentum flavum, and F) LORafter ligamentum flavum.4 ConclusionWe propose a forward facing OCT needle design to enable the evaluation of both,OCT speckle due to different tissue structures and the relative motion in orderto determine relevant deformations and ruptures at tissue boundaries. Hence,we are able to sense and thereby magnify the tissue mechanics during and priorboundary punctures without additional sensors such as FBGs.} } @COMMENT{Bibtex file generated on 2026-5-28 with typo3 si_bibtex plugin. Data from https://www.tuhh.de/mtec/publications/2024-2020 }