To-the-point (TTP) is a novel approach for analyzing compositional MR imaging data. By representing tibial and femoral cartilage T2 values as a dense point cloud, our approach can leverage the data's inherent sparsity while maintaining local geometric properties, leading to improved feature extraction and faster image processing times. Experiments on the whole OAI T2 dataset show strong performance in an OA diagnosis task 82.44% sens, 82.59% spec, with extracted features even identifying patients who would become diagnosed with OA 1 to 2 years in the future.
This abstract and the presentation materials are available to members only; a login is required.