A novel multi-atlas and multi-channel (MAMC) approach for multiple sclerosis lesion segmentation in brain MRI

This paper presents a novel approach for automatic segmentation of MS lesion including both of number and volume. The novelty includes the combination of the multiplicative intrinsic component optimization algorithm (Li et al. in Magn Reson Imaging 32:913–923, 2014) in bias field correction and normal tissue segmentation simultaneously, and the development of a multi-atlas and multi-channel (MAMC) segmentation approach. The first research focus is the classification of brain tissue into white matter, cerebrospinal fluid and gray matter in T1-w image and FLAIR image. The second research focus is the segmentation of MS lesion in white matter region using atlas. In label fusion, the coefficient as a specific weight is assigned to target label image based on the correlation function between atlases. This novel MAMC approach is evaluated by 20 training cases obtained from Medical Image Computing and Computer Aided Intervention Society 2008 MS Lesions Segmentation Challenge. The numerical results are presented in terms of accuracy, specificity and absolute volume difference. A comparison of MAMC approach and other conventional approaches is presented in terms of the true positive rate and the positive predictive value. Furthermore, the total lesion volume is calculated and compared with expert delineation. It can be seen that the MAMC approach is able to acquire a larger mean value of the Dice similarity coefficient than the other conventional approaches do. Therefore, this novel approach is an added value for the clinical evaluation of MS patients.