Novel framework for Detecting Multiple Sclerosis using Hybrid models

Abstract

The central nervous system (CNS) lesions that characterize the chronic inflammatory disease known as multiple sclerosis (MS) can cause significant physical or cognitive impairments as well as neurological problems. Segmentation of Magnetic Resonance Imaging (MRI) is one of the most important challenges in the process of detection and segmentation of MS. The process of segmentation of MRI means dividing the medical images into the disjoint part for making the process of analyzing and diagnosis too easy for clinicians. This paper presents a novel framework for the segmentation of the MRI to detect MS. The framework contains two stages; the first stage is about removing different types of noise such as rician and speckle noise; The framework uses different modified methods for removing rician and speckle noise such as uses Vibrational Mode Decomposition (VMD) along with Block-matching and 3D filtering (Bm3D)and (Conv-AE-Net, D-U-Net, Br-U-Net, DGan-Net, and DeRNet) for removing the rician noise. (DeRNet), (Dilated Convolution Auto encoder Denoising Network (Di-Conv-AE-Net), and Denoising Network (DGAN-Net) for removing the speckle noise. The second step in the framework is the segmentation of the most free-noise images using Residual U-net. The framework has achieved an accuracy of nearby 96.66 in dice score in the process of segmentation before removing noise and 96.89 after removing different types of noises.