Combining Image and Geometry Processing Techniques for the Quantitative Analysis of Muscle-Skeletal Diseases

For rheumatic diseases, it is fundamental to achieve an efficient medical evaluation of the patient's status and monitor the development of pathology. Acquiring and analyzing information on the pathology progression are important steps to customize the therapy and slow the disease's degeneration. This paper focuses on the localization of bone erosion sites, which are a typical symptom of rheumatic disease progression, from both morphological and tissue perspectives. To this end, we propose a geometry-based approach, which performs a geometric analysis of 3D segmented surfaces, and a texture-based approach, which analyses changes in the grey levels in a neighbour of the bone surface. These two approaches are integrated to define a more complete tool for the analysis and visualization of the input anatomical structures and the underlying pathology. The performances of the different methods are evaluated on the wrist district, acquired by a low-field magnetic resonance scanner.