MS Lesion Segmentation for Single and Multichannel MRI Images Using MICO Technique

摘要

We Implement multiplicative intrinsic component optimization technique to single and multichannel MRI images for segmenting multiple sclerosis lesions. This technique solves the problem of intensity inhomogeneity by estimating and correcting bias field. In segmentation it has more advantages over other MS lesion segmentation techniques in terms of precision. In multichannel segmentation, distinct weights are given for different channels to regulate the effect of each channel output. We use T1-w and FLAIR images in the multichannel segmentation of MS lesions. we provide higher weight to FLAIR images as they regulate in enhancing the segmented lesions. we were able to solve the problem of intensity inhomogeneity through bias field estimation and correction and followed by lesion segmentation. Our technique showed better and promising results when compared to ground truth images.