Deep Multimodality Image-Guided System for Assisting Neurosurgery (DeepIGN)

DeepIGN is a comprehensive AI-supported image-guided neurosurgery (IGN) system that integrates pre-operative MRI data, real-time intra-operative ultrasound (iUS), and positional instrument tracking to assist neurosurgeons. The project focuses on challenging aspects of medical image processing, including brain tumor segmentation, multimodal registration of images, and the explainability of Artificial Intelligence (AI) and deep learning models. It features an open-source deep learning framework (DeepSeg) that automatically and accurately delineates brain tumors (gliomas). To compensate for tissue deformation during surgery, it uses a fast, automated approach (iRegNet) for registering pre-operative MRI to intra-operative ultrasound volumes. Furthermore, it integrates an Explainable AI framework (NeuroXAI) that provides understandable visual explanations for the AI models' predictions, significantly increasing medical experts' trust.