Quantifying evolving processes in multimodal 3D medical images

Quantitative measurements of changes in evolving brain pathology, such as multiple sclerosis lesions and brain tumors, are important for clinicians to perform pertinent diagnoses and to help in patient follow-up. Lesions or tumors can vary over time in size, shape, location and composition because of natural pathological processes or the effect of a drug treatment or therapy. In the past, people have used as a quantitative measurement the change in total or regional lesion/tumor volume. In this paper we propose a new model to quantify changes in evolving processes in multimodal 3D medical images. We believe this model reflects changes in pathology more accurately because it simultaneously takes into account information in multiple imaging modalities and the location of lesion/tumor voxels. We demonstrate the effectiveness of this model with experiments on synthetic lesion data.