A new classifier feature space for an improved Multiple Sclerosis lesion segmentation

Intensity based classification relies on contrast between tissue types adjacent in feature space and adequate signal compared to image noise. Contrast between brain tissue types in Multiple Sclerosis patients Magnetic Resonance Imaging is reduced due to the presence of lesions which intensity values overlap with healthy tissue, resulting in tissue misclassification. We propose a new, extended classifier feature space that is based in spatial locations, the intensity of which is abnormal when compared to the expected values in a healthy population in the same location. Segmentation results using our new extended feature space proves an improvement in both sensitivity and specificity in lesion classification.