19 results
Now showing1 - 10 of 19
- Some of the metrics are blocked by yourconsent settings
Item type:Publication, Access status: without files , Deep denoising of volumetric OCT images for in vivo motion detection(IEEE, 2025-07); ; ; ; Optical Coherence Tomography (OCT) is an imaging modality with high temporal and spatial resolution. There are different applications, e.g. in ophthalmology, dermatology, cardiology and recently OCT image based motion compensation has been considered. However, OCT suffers from speckle noise which affects image quality and subsequent interpretation. We propose denoising based on unsupervised 3D convolutional Autoencoder (AE) and systematically evaluate the model on in vivo and post mortem datasets. Moreover, we study motion detection as a subsequent task and to investigate the effects of AE based denoising. Our results demonstrate that speckle noise in the OCT images can lead to substantial outliers. Denoising based on AEs is effective in reducing the outliers and results in improved motion detection. Furthermore, the proposed AE processes OCT volumes in 0.5 ms, making it suitable for real-time applications. Finally, the results illustrate that the AE can effectively improve motion detection performance on in vivo data, despite being trained on different data. In conclusion, our AE model presents a simple and unsupervised deep learning approach to obtain fast denoising adapted to OCT imaging. For motion detection, denoising can be crucial to avoid artifacts due to outliers.Clinical relevance: This work provides an efficient denoising Autoencoder for fast data processing that can be applied in various clinical scenarios to improve noisy 3D Optical Coherence Tomography images.Publicationtype: Conference Paper7 - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Access status: without files , A-scan sequence transformers for palpation with optical coherence elastography(2025-05-01); ; ; ; Recognizing the properties of elastic tissue can facilitate surgical navigation, e.g., when localizing lesions by palpation. However, palpation is very subjective and often unavailable in minimally invasive surgery. High-speed optical coherence elastography (OCE) adapted for intraoperative use could enable elasticity estimation by measuring the propagation of mechanically stimulated waves. However, robust estimation of wave velocity can be challenging, and reconstruction of the elastic modulus is highly dependent on the correct modeling of wave propagation. We therefore consider deep learning for the end-to-end estimation of elasticity from OCE phase data. Since optical coherence tomography inherently produces a temporal sequence of one-dimensional axial scans (A-scans), we consider transformer-based deep learning models to directly process A-scan sequences. For homogeneous tissue phantoms with known elastic properties, we obtain a mean error of 1.64 kPa, which significantly improves elasticity reconstruction compared to conventional processing and the best CNN-based approach with 7.80 kPa and 5.55 kPa, respectively. Furthermore, we demonstrate generalization to heterogeneous phantoms with inclusions and assess the elasticity of soft tissue samples, including heart, kidney, and liver. The results show that transformer architectures are well suited for reconstructing elasticity from A-scan sequences in OCE.Publicationtype: Journal Article26 - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Access status: without files , Efficient anatomy segmentation in laparoscopic surgery using multi-teacher knowledge distillation(Microtome Publishing, 2024-07); ; ; ; Automatic segmentation of anatomical structures in laparoscopic images or videos is an important prerequisite for visual assistance tools which are designed to increase efficiency and safety during an intervention. In order to be used in a realistic clinical scenario, both high accuracy and real-time capability are required. Current deep learning networks for anatomy segmentation show high accuracy, but are not suitable for real-time clinical application due to their large size. As smaller, real-time capable deep learning networks show lower segmentation performance, we propose a multi-teacher knowledge distillation approach applicable to partially labeled datasets. We leverage the knowledge of multiple anatomy-specific, high-accuracy teacher networks to improve the segmentation performance of a single and efficient student network capable of segmenting multiple anatomies simultaneously. To do so, we minimize the Kullback-Leibler divergence between the normalized anatomy-specific teacher logits and the respective normalized logits of the student. We conduct experiments on the Dresden Surgical Anatomy Dataset, which provides multiple subsets of binary segmented anatomical structures. Results show that our approach can increase the overall Dice score for different real-time capable network architectures for anatomy segmentation.Publicationtype: Conference Paper18 - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Access status: without files , Diffusion models with ensembled structure-based anomaly scoring for unsupervised anomaly detectionSupervised deep learning techniques show promise in medical image analysis. However, they require comprehensive annotated data sets, which poses challenges, particularly for rare diseases. Consequently, unsupervised anomaly detection (UAD) emerges as a viable alternative for pathology segmentation, as only healthy data is required for training. However, recent UAD anomaly scoring functions often focus on intensity only and neglect structural differences, which impedes the segmentation performance. This work investigates the potential of Structural Similarity (SSIM) to bridge this gap. SSIM captures both intensity and structural disparities and can be advantageous over the classical l1 error. However, we show that there is more than one optimal kernel size for the SSIM calculation for different pathologies. Therefore, we investigate an adaptive ensembling strategy for various kernel sizes to offer a more pathology-agnostic scoring mechanism. We demonstrate that this ensembling strategy can enhance the performance of DMs and mitigate the sensitivity to different kernel sizes across varying pathologies, highlighting its promise for brain MRI anomaly detection.Publicationtype: Conference Paper10 - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Access status: without files , Can complex-valued neural networks improve force sensing with optical coherence tomography?(IEEE, 2024-05); ; ; ; Deep learning has proven highly effective in processing optical coherence tomography (OCT) images. However, current approaches often treat complex OCT data as real-valued representations and mostly focus on the amplitude component. In contrast, OCT phase can detect submicrometer displacements, surpassing the resolution of amplitude-based methods. In this study, we investigate complex-valued and real-valued neural networks (cvNNs and rvNNs) for end-to-end needle tip force estimation from complex OCT data. We consider predicting absolute forces and force differences over time. We compare cvNNs with two rvNNs where amplitude and phase are either stacked in the channel dimension or processed separately in a dual-path architecture. Our results show that the amplitude-focused estimation of the absolute force is still best achieved with rvNNs. However, when estimating small force differences, where phase-sensitive OCT is particularly valuable, cvNNs outperform rvNNs with stacked inputs. Similar improvements can be achieved with separate processing in the dual-path models. The results emphasize the importance of model design in processing complex OCT signals and demonstrate the potential of cvNNs for phase-sensitive OCT. However, our work also highlights the current limitations of cvNNs, particularly computational cost and data dependency.Publicationtype: Conference Paper20 - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Access status: without files , Nodule detection and generation on chest X-rays: NODE21 challenge(IEEE, 2024-03-26); ; ; ; Pulmonary nodules may be an early manifestation of lung cancer, the leading cause of cancer-related deaths among both men and women. Numerous studies have established that deep learning methods can yield high-performance levels in the detection of lung nodules in chest X-rays. However, the lack of gold-standard public datasets slows down the progression of the research and prevents benchmarking of methods for this task. To address this, we organized a public research challenge, NODE21, aimed at the detection and generation of lung nodules in chest X-rays. While the detection track assesses state-of-the-art nodule detection systems, the generation track determines the utility of nodule generation algorithms to augment training data and hence improve the performance of the detection systems. This paper summarizes the results of the NODE21 challenge and performs extensive additional experiments to examine the impact of the synthetically generated nodule training images on the detection algorithm performance.Publicationtype: Journal Article22 - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Access status: without files , Convolutional transformer network for paranasal anomaly classification in the maxillary sinus(SPIE, 2024-02); ; ; ; Large-scale population studies have examined the detection of sinus opacities in cranial MRIs. Deep learning methods, specifically 3D convolutional neural networks (CNNs), have been used to classify these anomalies. However, CNNs have limitations in capturing long-range dependencies across the low and high level features, potentially reducing performance. To address this, we propose an end-to-end pipeline using a novel deep learning network called ConTra-Net. ConTra-Net combines the strengths of CNNs and self-attention mechanisms of transformers to classify paranasal anomalies in the maxillary sinuses. Our approach outperforms 3D CNNs and 3D Vision Transformer (ViT), with relative improvements in F1 score of 11.68% and 53.5%, respectively. Our pipeline with ConTra-Net could serve as an alternative to reduce misdiagnosis rates in classifying paranasal anomalies.Publicationtype: Conference Paper27 - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Access status: without files , Weakly supervised Medulloblastoma tumor classification using domain specific patch-level feature extractionMedulloblastoma (MB) is the most common embryonal tumour of the brain. In order to decide on an optimal therapy, laborious inspection of histopathological tissue slides by neuropathologists is necessary. Digital pathology with the support of deep learning methods can help to improve the clinical workflow. Due to the high resolution of histopathological images, previous work on MB classification involved manual selection of patches, making it a time consuming task. In order to leverage only slide labels for histopathology image classification, weakly supervised approaches first encode small patches into feature vectors using an ImageNet pretrained encoder based on convolutional neural networks. The representations of patches are further utilized to train a data-efficient attention-based learning method. Due to the domain shift between natural images and histopathology images, the encoder is not optimal for feature extraction for MB classification. In this study, we adapt weakly supervised learning for MB classification and examine different histopathological specific encoder architectures and weights for the MB classification task. The results show that ResNet encoders pretrained with histopathology images lead to better MB classification results compared to encoders pretrained on ImageNet. The best performing method uses a ResNet50 architecture, pretrained on histopathology images and achieves an area under the receiver operating curve (AUROC) value of 71.89%, improving the baseline model by 2%.Publicationtype: Conference Paper10 - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Access status: without files , Leveraging the Mahalanobis Distance to Enhance Unsupervised Brain MRI Anomaly Detection(Springer, 2024); ; ; ; Unsupervised Anomaly Detection (UAD) methods rely on healthy data distributions to identify anomalies as outliers. In brain MRI, a common approach is reconstruction-based UAD, where generative models reconstruct healthy brain MRIs, and anomalies are detected as deviations between input and reconstruction. However, this method is sensitive to imperfect reconstructions, leading to false positives that impede the segmentation. To address this limitation, we construct multiple reconstructions with probabilistic diffusion models. We then analyze the resulting distribution of these reconstructions using the Mahalanobis distance to identify anomalies as outliers. By leveraging information about normal variations and covariance of individual pixels within this distribution, we effectively refine anomaly scoring, leading to improved segmentation. Our experimental results demonstrate substantial performance improvements across various data sets. Specifically, compared to relying solely on single reconstructions, our approach achieves relative improvements of 15.9%, 35.4%, 48.0%, and 4.7% in terms of AUPRC for the BRATS21, ATLAS, MSLUB and WMH data sets, respectively.Publicationtype: Conference Paper30 - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Access status: without files , Quantitative evaluation of activation maps for weakly-supervised lung nodule segmentation(SPIE, 2024-01); ; ; ; The manual assessment of chest radiographs by radiologists is a time-consuming and error-prone process that relies on the availability of trained professionals. Deep learning methods have the potential to alleviate the workload of radiologists in pathology detection and diagnosis. However, one major drawback of deep learning methods is their lack of explainable decision-making, which is crucial in computer-aided diagnosis. To address this issue, activation maps of the underlying convolutional neural networks (CNN) are frequently used to indicate the regions of focus for the network during predictions. However, often, an evaluation of these activation maps concerning the actual predicted pathology is missing. In this study, we quantitatively evaluate the usage of activation maps for segmenting pulmonary nodules in chest radiographs. We compare transformer-based, CNN-based, and hybrid architectures using different visualization methods. Our results show that although high performance can be achieved in the classification task across all models, the activation masks show little correlation with the actual position of the nodules.Publicationtype: Conference Paper12