Hierarchical Mrf and Random Forest Segmentation of Ms Lesions and Healthy Tissues in Brain Mri

In this paper, we present an automatic hierarchical framework for the segmentation of a variety healthy tissues and lesions in brain MRI of patients with Multiple Sclerosis (MS). At the voxel level, lesion and tissue labels are estimated through a Markov Random Field (MRF) segmentation framework that leverages spatial prior probabilities for 9 healthy tissues through multi-atlas fusion (MALF). A random forest classifier then provides region level lesion refinement. Validation is performed on the data provided by the ISBI 2015 Longitudinal Multiple Sclerosis Lesion Segmentation Challenge.