The MRI Scanner as a Diagnostic: Image-less Active Sampling
This work addresses accessibility issues in MRI-based diagnostics by potentially reducing magnetic field strength and acquisition times, though it is incremental as it builds on existing undersampling and ML techniques.
The paper tackles the challenge of making MRI more accessible for point-of-care disease identification by proposing an ML-based framework that uses reinforcement learning to actively sample k-space data at the patient level, directly inferring disease without image reconstruction; it achieves diagnostic performance comparable to ML-based methods using fully sampled data for meniscus tear detection in knee MRI.
Despite the high diagnostic accuracy of Magnetic Resonance Imaging (MRI), using MRI as a Point-of-Care (POC) disease identification tool poses significant accessibility challenges due to the use of high magnetic field strength and lengthy acquisition times. We ask a simple question: Can we dynamically optimise acquired samples, at the patient level, according to an (automated) downstream decision task, while discounting image reconstruction? We propose an ML-based framework that learns an active sampling strategy, via reinforcement learning, at a patient-level to directly infer disease from undersampled k-space. We validate our approach by inferring Meniscus Tear in undersampled knee MRI data, where we achieve diagnostic performance comparable with ML-based diagnosis, using fully sampled k-space data. We analyse task-specific sampling policies, showcasing the adaptability of our active sampling approach. The introduced frugal sampling strategies have the potential to reduce high field strength requirements that in turn strengthen the viability of MRI-based POC disease identification and associated preliminary screening tools.