DIVA: Diagnostic and Decision Support for Vertebral Body Fractures
Contact Persons
Matthias Keicher (matthias.keicher@tum.de)
Abstract
Osteoporosis is a widespread disease with significant socioeconomic consequences. These result to a large extent from fractures caused by osteoporosis that are detected too late. Fractures could often be detected in time if CT scans collected in routine clinical practice were analyzed for fractures. However, due to time constraints, this is often not the case. Automation can therefore add significant value to the healthcare system here. The aim of the research project described here is to develop a prototype for an assistance system in radiological diagnostics that automatically diagnoses and classifies vertebral fractures using artificial intelligence methods. In doing so, the prototype makes the decision of the neural network transparent by verbalizing and visualizing the inner decision basis of the network as a rationale. This system provides treating physicians for the first time with a comprehensive assistance solution that significantly improves the detection of incisional vertebral body fractures and provides and justifies diagnostic assessment relevant for further therapy and prevention with regard to etiology, classification and clinical relevance.
Keywords: Vertebral body fracture, osteoporosis, vision-language models, explainability, semantic features, structured report generation
Research Partners
Project members
This project is funded by the Federal Ministry of Education and Research (BMBF) as part of the KMU Innovativ: Medizintechnik program under the grant agreement 13GW0469.
