Automatic segmentation of multiple sclerosis lesions in multispectral MR images using kernel fuzzy c-means clustering

Magnetic resonance (MR) images can be used to detect lesions in the brains of multiple sclerosis (MS) patients and is essential for diagnosing the disease and monitoring its progression. An automatic method is presented for segmentation of MS lesions in multispectral MR images. Firstly a PD-w image is subtracted from its corresponding T1-w image to get an image in which the cerebral spinal fluid (CSF) is enhanced. Then based on kernel fuzzy c-means (KFCM) algorithm, the enhanced image and the corresponding T2-w image are segmented respectively to extract the CSF region and the CSF combining MS lesions region. A raw MS lesions image is obtained by subtracting the CSF region from CSF combining MS region. By applying median filter and thresholding to the raw image, the MS lesions are detected finally. Results are quantitatively evaluated on BrainWeb images using Dice similarity coefficient (DSC). Finally, the potential of the method as well as its limitations are discussed.