Detecting statistically significant changes in cartilage thickness with sub-voxel precision

Introduction: MRI-based quantification of the cartilage thickness is a robust and well validated technique for the assessment of cartilage degradation in osteoarthritis [1], and for the analysis of cartilage deformations after exercise [2]. Nowadays changes in cartilage thickness are evaluated by comparing averaged thickness over regions defined on an anatomical basis [1]. However, the combination of high-field MRI, sub-voxel segmentation techniques and the use of registration methods allow an accurate voxel-based detection of changes in cartilage thickness. The aim of this work was to establish a method to track statistically significant changes in the cartilage thickness in longitudinally acquired datasets based on the measurement of the errors in thickness.