To improve the quality of MRI-based diagnosis with trustworthy AI methods. For this, we aim to optimize the full chain from image acquisition prescription, to image analysis and introduction of AI supported acquisition and diagnosis in clinical practice. This can greatly improve the accuracy of diagnosis, while reducing the costs. At first we focus on neurological and MSK diseases.
Within the overall aim there are specific work packages that interact and together build towards the overall aim:
- develop smart, adaptive, MR imaging protocols for precision diagnosis
- develop end-to-end deep-learning based MR image reconstruction
- develop trustworthy AI for integrated diagnostics of brain tumors
- develop trustworthy AI methods for improved diagnosis of bone and soft-tissue lesions on MRI
- develop an approach that ensures acceptance of AI technology in daily clinical radiology setting.
The smart adaptive protocols together with the deep-learning based MR image reconstruction will form the basis for clinical MR scanning in the future. Key ingredients for these improvements come from the clinical expertise and experiences we have, and further develop, in the AI assisted diagnostic approaches for brain tumors and bone/soft-tissue lesions. Specifically, in these approaches image-based disease biomarkers will be created, which can be a subsequent target for the protocol adaptation. By focusing on two clinical domains, we enhance generalizability of the developed methods and ensure broad applicability across disease domains.
Conversely the clinical work packages start from the state-of-the-art acquisition methods to improve diagnostic value and identify key descriptors of disease. Throughout all work packages there is a continued focus on the applicability and acceptance of the developed methodology in terms of explainability, accuracy and reliability. There will be continuous interaction among the PhD students involved which will benefit all work packages in this lab.
In addition to the participants from BIGR that are listed below, Marion Smits, Edwin Oei (Erasmus MC), Juan Hernandez-Tamames (Erasmus MC), Jan-Jaap Visser (Erasmus MC), and Sandra Sülz (Erasmus University) are principal investigators in this project.