Detection and delineation of multiple sclerosis lesions in gadolinium-enhanced 3D T1-weighted MRI data

Abstract

An automatic method for detecting and delineating Gd-enhanced lesions on T1-weighted magnetic resonance images in multiple sclerosis (MS) brains is described. In order to detect and limit the enhancements to the region defined by the brain mask, a combination of thresholding and mathematical morphological operations was implemented. A 3D connected component labeling algorithm is used for producing both the brain mask and labeling the enhanced lesions. False positives that arise from the enhancing vasculature and structures that do not exhibit a blood-brain barrier (BBB) were automatically detected and eliminated by spatially registering the Tl-weighted and the dual-echo affirmative images. Lesion enhancements were delineated using fuzzy connectedness. This technique is evaluated on MS patients with excellent results.