Spatio-Temporal Segmentation of Rheumatoid Arthritis Lesions in Serial MR Images of Joints

Abstract

Recent innovations in drug therapies in rheumatoid arthritis (RA) have made it highly desirable to obtain sensitive biomarkers of disease progression that can be used to quantify the performance of candidate disease modifying drugs. We present a spatio-temporal analysis technique to automatically quantify small changes in a bone in in-vivo serial MR images from an experimental model of RA. The technique integrates the time-domain information across all the time points by building a 5-dimensional feature space (3 spatial dimensions, 1 intensity dimension, and 1 temporal dimension) from the serial MR images after rigid image registration. The feature space is then delineated by the mean shift algorithm to give high-intensity bone lesions as 4D segmentations. We detected significant temporal changes in bone lesion volume in 5 out of 7 identified candidate bone lesion regions, and significant difference in bone lesion volume between male and female subjects in 1 out of 7 candidate bone lesion regions. We quantitatively compared this technique with a previous method using simulated and real MR images, and histology of the subjects. We found that this technique was more sensitive to small bone lesion changes than a previous method.