Automated segmentation method of white matter and gray matter regions with multiple sclerosis lesions in MR images

• DOI: 10.1007/s12194-010-0106-x
• Publication Date: 2010-09-29T04:33:56Z
• Authors: Taiki Magome; Hidetaka Arimura; Shingo Kakeda; Daisuke Yamamoto; Yasuo Kawata; Yasuo Yamashita; Yoshiharu Higashida; Fukai Toyofuku; Masafumi Ohki; Yukunori Korogi
• Keywords: Adult; Male; Multiple Sclerosis; Adolescent; Brain; Middle Aged; Magnetic Resonance Imaging; Automation; Young Adult; 03 medical and health sciences; 0302 clinical medicine; Image Processing, Computer-Assisted; Humans; Female; Diagnosis, Computer-Assisted

Our purpose in this study was to develop an automated method for segmentation of white matter (WM) and gray matter (GM) regions with multiple sclerosis (MS) lesions in magnetic resonance (MR) images. The brain parenchymal (BP) region was derived from a histogram analysis for a T1-weighted image. The WM regions were segmented by addition of MS candidate regions, which were detected by our computer-aided detection system for the MS lesions, and subtraction of a basal ganglia and thalamus template from "tentative" WM regions. The GM regions were obtained by subtraction of the WM regions from the BP region. We applied our proposed method to T1-weighted, T2-weighted, and fluid-attenuated inversion-recovery images acquired from 7 MS patients and 7 control subjects on a 3.0 T MRI system. The average similarity indices between the specific regions obtained by our method and by neuroradiologists for the BP and WM regions were 95.5 ± 1.2 and 85.2 ± 4.3%, respectively, for MS patients. Moreover, they were 95.0 ± 2.0 and 85.9 ± 3.4%, respectively, for the control subjects. The proposed method might be feasible for segmentation of WM and GM regions in MS patients.