Method for image-based clinical decision support for axial spondyloarthritis using artificial intelligence

A method performed by a computer device for generating information for assisting in the diagnosis of axial spondyloarthritis, the method comprising:

• preprocessing an input magnetic resonance imaging (MRI) image by performing normalization and histogram matching to standardize the MRI image;
• extracting, for each of a plurality of slices of the preprocessed MRI image, a first slice feature representing visual characteristics of the MRI image by applying a first feature extraction model;
• extracting a second slice feature representing quantitative characteristics of bone marrow edema (BME) by applying a second feature extraction model to BME data associated with the preprocessed MRI image;
• generating slice-level integrated feature vectors by combining, on a slice-by-slice basis, the first slice feature and the second slice feature, applying attention-based importance weights to the respective integrated feature vectors, and generating a patient-level final integrated feature vector by summing the weighted vectors;
• and
• generating diagnostic support information based on the patient-level final integrated feature vector for use by a clinician in making a diagnostic decision.