Unsupervised Multimodal Magnetic Resonance Images Segmentation and Multiple Sclerosis Lesions Extraction based on Edge and Texture Features
Segmentation of Multiple Sclerosis (MS) lesions is a crucial part of MS diagnosis and therapy. Segmentation of lesions is usually performed manually, exposing this process to human errors. Thus, exploiting automatic and semi-automatic methods is of interest. In this paper, a new method is proposed to segment MS lesions from multichannel MRI data (T1-W and T2-W). For this purpose, statistical features of spatial domain and wavelet coefficients of frequency domain are extracted for each pixel of skull-stripped images to form a feature vector. An unsupervised clustering algorithm is applied to group pixels and extracts lesions. Experimental results demonstrate that the proposed method is better than other state of art and contemporary methods of segmentation in terms of Dice metric, specificity, false-positive-rate, and Jaccard metric.
Multiple sclerosis, Segmentation, Multichannel Magnetic Resonance Imaging (MRI), Wavelet, Energy, Entropy