Christine Lucas Tardif1, John B. Richardson2,
Claude Lepage1, D Louis Collins1, Alan C. Evans1,
G Bruce Pike1
1McConnell Brain Imaging Centre,
Montreal Neurological Institute, Montreal, QC, Canada; 2Department
of Neuropathology, Montreal Neurological Institute/Hospital, Montreal, QC,
Canada
Cortical
grey matter (GM) multiple sclerosis lesions are difficult to segment in MR
images due to poor contrast with normal appearing GM and spatial variation in
healthy GM. We propose using an observer-independent profile-based method for
cortical parcellation to detect cortical lesions. Following tissue
classification and surface extraction, profiles are extended from the white
matter surface to the cortical surface. The cortex is parcellated according
to profile intensity and shape using a kmeans classifier with squared
Euclidean distance. The method is tested on high-resolution MR data from a
fixed postmortem MS brain, and validated using myelin basic protein
immunohistochemistry.