Problem description
Hand and wrist radiographs are among the most common X-ray examinations, and every emergency department orders them, usually after a fall. Yet radiology departments are under growing pressure, and not every hospital has musculoskeletal (MSK) expertise around the clock. The images are also difficult to read, with many small bones that partly overlap. Acute non-displaced fractures are easily missed, and so are accompanying ligament injuries. Treatment and follow-up often depend on precise measurements of alignment and joint spaces, which take time to do by hand and vary between readers.
Project summary
MSKAssist is an AI system that significantly improves the analysis of hand and wrist radiographs by combining multiple diagnostic components into a single product that integrates seamlessly into the radiology workflow. The system combines the most important and most common tasks for assessing fractures, joint abnormalities and rare abnormalities such as bone tumours. MSKAssist builds on previously published research and addresses an important clinical need for standardized musculoskeletal assessment. It provides explainable AI decision-making through intuitive visualizations and automatically generates a structured report. This reduces diagnostic variability, eases the radiological workload and improves the quality of patient care. The RTC AI contributes to the development of the core system and its integration into the clinical workflow.
Funding
This project is funded by the NWO Demonstrator.
References
Sources from the images in the project thumbnail (upper left, counter clock-wise): Laurens van Baardewijket al, Louis A. Gilula and Ileana Chesaru, David Cuete, Henk Jan van der Woude and Robin Smithuis
