A Semi-automatic Method for Segmentation of Multiple Sclerosis Lesions on Dual-Echo Magnetic Resonance Images

The identification and segmentation of focal hyperintense lesions on magnetic resonance images (MRI) are essential steps in the assessment of disease burden in multiple sclerosis (MS) patients. Manual lesion segmentation is considered to be the gold standard, although it is time-consuming and has poor intra- and inter-operator reproducibility. Here, we present a segmentation method based on dual-echo MR images initialized by manual identification of lesions and a priori information. The classification technique is based on a region growing approach with a final segmentation refinement step. The results have revealed high similarity between the segmentation performed with this method and that performed manually by an expert operator, as well as a low misclassification of lesions. Moreover, the time required for segmentation is drastically reduced.