The XAIS project creates interactive medical image segmentation methods using explainable AI systems and XR visualization that advise medical experts on image-based diagnosis and clinical treatment planning. The aim is to integrate the user expertise more concretely into the AI capabilities by enabling an interactive dialogue between the actors. This interactive AI system will be based on probabilistic approximate Bayesian Deep Learning methods. This will be integrated with the XR visualization and interaction for efficient feedback to facilitate the AI system for the volumetric segmentation of tissues. For the XR research a Human-Centered Design approach will be used. Controlled user experiments with medical expert users will be carried out. The expected results will make the medical image analysis methods more accurate and reliable, increasing the medical experts' confidence in the segmentation results and leading to savings in analysis time and avoiding possibly costly mistakes.

2022

2024