At this key conference on medical image computing and computer assisted intervention, VRVis presented its research work on visualization techniques for interpretable AI in medicine as well as tumor segmentation on multimodal medical images using deep learning methods.
As the most important forum in the fields of medical image computing and computer assisted interventions, the MICCAI conference brought together the most innovative projects from top research and practice for the 23rd time from October 4 to 8, 2020. Due to the coronavirus, the conference, originally planned in Lima (Peru), was held virtually. As in previous years, the Biomedical Image Informatics group of VRVis, which specializes in the combination of visual computing and medical imaging, was represented with two very strong contributions.
Dimitrios Lenis and David Major presented their paper "Domain aware medical image classifier interpretation by counterfactual impact analysis" in the panel "Interpretability" on October 6th, in which they introduced the scientific community to their innovative approach for trustworthy decision visualization of AI systems for AI-based diagnostics in digital radiology; keyword: interpretable AI.
During the ML-CDS workshop (Multimodal Learning for Clinical Decision Support) on October 8th, Theresa Neubauer and Maria Wimmer also presented the paper "Soft Tissue Sarcoma Co-Segmentation in Combined MRI and PET/CT Data". This paper gives an overview of the research results of Theresa Neubauer's master thesis on segmentation of multimodal medical images using deep learning methods. The master thesis was written in the context of an FFG funded FEMtech internship at VRVis.
