Start of funding 01.07.2025

Longitudinal analysis of spatially aligned Radiology images using Multimodal Large language Models

Prof. Dr. Daniel Rückert
Technische Universität München
School of Medicine and Health - Klinikum rechts der Isar

Prof. Dr. Akshay Chaudhari
Stanford University
Stanford Center for Artificial Intelligence in Medicine and Imaging (AIMI)



Advances in artificial intelligence have led to the emergence of large-scale deep learning models capable of jointly analyzing medical images and clinical text. This project introduces a novel framework that integrates deformable image registration with vision-language models to enhance the performance of downstream radiological tasks. The proposed system will spatially align longitudinal MRI and CT scans, encode volumetric imaging features into a multimodal vision-language representation, and generate clinically meaningful interpretations. This integration enables precise monitoring of disease progression, automated quantification of tumor dynamics, and assessment of therapeutic response over time. Through systematic design and evaluation, this work aims to demonstrate the feasibility and utility of registration-informed, AI-driven analysis of longitudinal medical imaging, thereby advancing the state of the art in automated radiological interpretation.