Showing papers by "Andrea Schenk published in 2022"

Improving automatic liver tumor segmentation in late-phase MRI using multi-model training and 3D convolutional neural networks

Abstract: Abstract Automatic liver tumor segmentation can facilitate the planning of liver interventions. For diagnosis of hepatocellular carcinoma, dynamic contrast-enhanced MRI (DCE-MRI) can yield a higher sensitivity than contrast-enhanced CT. However, most studies on automatic liver lesion segmentation have focused on CT. In this study, we present a deep learning-based approach for liver tumor segmentation in the late hepatocellular phase of DCE-MRI, using an anisotropic 3D U-Net architecture and a multi-model training strategy. The 3D architecture improves the segmentation performance compared to a previous study using a 2D U-Net (mean Dice 0.70 vs. 0.65). A further significant improvement is achieved by a multi-model training approach (0.74), which is close to the inter-rater agreement (0.78). A qualitative expert rating of the automatically generated contours confirms the benefit of the multi-model training strategy, with 66 % of contours rated as good or very good, compared to only 43 % when performing a single training. The lesion detection performance with a mean F1-score of 0.59 is inferior to human raters (0.76). Overall, this study shows that correctly detected liver lesions in late-phase DCE-MRI data can be automatically segmented with high accuracy, but the detection, in particular of smaller lesions, can still be improved.

Evaluating Soft Organ-Shaped Tangibles for Medical Virtual Reality

Abstract: Connecting digital information with the physical is one of the essential ideas of tangible user interfaces. The design of the physical representation is important especially for specialised domains like surgery planning, because surgeons rely heavily on their tactile senses. Therefore, this research work investigates the effect of a soft and a hard 3D model as an interaction device for virtual reality surgical planning. A user study with 13 surgeons reveals a clear preference for the softer, more realistic material and a significantly higher haptic user experience for the soft model compared to the hard one. These results advocate for stressing material aspects along with the interaction design in domains with an inherently high focus on tactile aspects.

Efficacy of Oncolytic Herpes Simplex Virus T-VEC Combined with BET Inhibitors as an Innovative Therapy Approach for NUT Carcinoma

Abstract: Simple Summary Since T-VEC is already approved for treatment of melanoma, its promising efficacy shown here also for NUT carcinoma (NC) cell lines may create a rapid transition to individual treatments as well as clinical trials in NC patients. The idea of combining T-VEC immunotherapy with BET inhibitors is strengthened by the assumption that the initial rapid response of NC to BET inhibitor therapy and the additional direct tumor cell lysis triggered by virotherapeutics may be able to effectively stabilize or even shrink the tumor cell mass to bridge the time gap until the durable immune response, induced by immunovirotherapy, can lead to complete tumor remission. This would signify a real breakthrough for patients suffering from this extremely aggressive tumor, whose average survival time is currently in the range of only six months. Abstract NUT carcinoma (NC) is an extremely aggressive tumor and current treatment regimens offer patients a median survival of six months only. This article reports on the first in vitro studies using immunovirotherapy as a promising therapy option for NC and its feasible combination with BET inhibitors (iBET). Using NC cell lines harboring the BRD4-NUT fusion protein, the cytotoxicity of oncolytic virus talimogene laherparepvec (T-VEC) and the iBET compounds BI894999 and GSK525762 were assessed in vitro in monotherapeutic and combinatorial approaches. Viral replication, marker gene expression, cell proliferation, and IFN-β dependence of T-VEC efficiency were monitored. T-VEC efficiently infected and replicated in NC cell lines and showed strong cytotoxic effects. This implication could be enhanced by iBET treatment following viral infection. Viral replication was not impaired by iBET treatment. In addition, it was shown that pretreatment of NC cells with IFN-β does impede the replication as well as the cytotoxicity of T-VEC. T-VEC was found to show great potential for patients suffering from NC. Of note, when applied in combination with iBETs, a reinforcing influence was observed, leading to an even stronger anti-tumor effect. These findings suggest combining virotherapy with diverse molecular therapeutics for the treatment of NC.

Improving deep learning based liver vessel segmentation using automated connectivity analysis

Abstract: Segmenting full vessel systems of the human liver is important for many applications in liver surgery and intervention planning. While methods exist for training DNNs for vessel segmentation, no method we know of efficiently extracts the vessel graph without modifying the DNN architecture. We demonstrate a fully automatic method for extracting and separating vessel graphs directly from the output of a segmentation model by applying a modified algorithm for vessel connectivity analysis. This method significantly improves the centerline sensitivity of reconstructed graphs on the IRCAD dataset and achieves similar scores for splitting vessel systems as the recently published TopNet approach.

Hepatic artery segmentation with 3D convolutional neural networks

Abstract: The segmentation of liver vessels is a crucial task for liver surgical planning. In selective internal radiation therapy, a catheter has to be placed into the hepatic artery, injecting radioactive beads to internally destroy tumor tissue. Based on a set of 146 abdominal CT datasets with expert segmentations, we trained three-level 3D U-Nets with loss-sensitive re-weighting. They are evaluated with respect to different measures including the Dice coefficient and the mutual skeleton coverage. The best model incorporates a masked loss for the liver area, which achieves a mean Dice coefficient of 0.56, a sensitivity of 0.69 and a precision of 0.66.

Computer-assisted image-based risk analysis and planning in lung surgery - a review

Abstract: In this paper, we give an overview on current trends in computer-assisted image-based methods for risk analysis and planning in lung surgery and present our own developments with a focus on computed tomography (CT) based algorithms and applications. The methods combine heuristic, knowledge based image processing algorithms for segmentation, quantification and visualization based on CT images of the lung. Impact for lung surgery is discussed regarding risk assessment, quantitative assessment of resection strategies, and surgical guiding. In perspective, we discuss the role of deep-learning based AI methods for further improvements.