سامانه همایش های دانشگاه آزاد اسلامی واحد نجف آباد

صفحه اصلی / اولین همایش بین المللی تحول خدمات سلامت: تلفیق مراقبت، درمان، بازتوانی و امیدبخشی

Early Diagnosis of Multiple Sclerosis Using Magnetic Resonance Imaging Based on LASSO Feature Selection and Convolutional Neural Networks

نویسندگان :

Alireza Zarei¹ Ebrahim Daneshifar²

1- دانشگاه بین المللی امام رضا علیه السلام 2- دانشگاه بین المللی امام رضا علیه السلام

کلمات کلیدی :

Multiple Sclerosis،Early Diagnosis،Magnetic Resonance Imaging،Deep Learning،Neural Networks Computer،Image Processing Computer-Assisted

چکیده :

ackground: Multiple Sclerosis (MS) is a progressive autoimmune disease of the central nervous system, and its early diagnosis plays a vital role in controlling the disease progression (1, 2). Manual investigation of lesions in Magnetic Resonance Imaging (MRI) by physicians is time-consuming and prone to human error (3). The aim of this study is to present an intelligent deep learning-based approach to automate and accelerate the early diagnosis of MS (4, 5). Method: In this study, the standard BraTS dataset containing MRI images of patients was utilized. Following image pre-processing, a combination of the LASSO method and the fractal model was employed for optimal feature extraction (including brightness intensity, edges, and texture) and dimensionality reduction. Ultimately, the classification of images into three groups (suspected MS, relapsing-remitting, and progressive) was performed using a Convolutional Neural Network (CNN) based on a four-layer Deep Belief Network. Finding: Simulation results demonstrated that the proposed model is capable of identifying and classifying MS-related lesions with an accuracy of 98.68%. The combined application of LASSO and the fractal model resulted in the elimination of redundant features, prevention of network overfitting, and a substantial enhancement in the classifier’s performance compared to traditional methods. Conclusion: The proposed intelligent system, by leveraging deep learning and optimal feature selection methods, can serve as a Computer-Aided Diagnosis (CAD) tool. In addition to reducing human error, it can significantly assist radiologists and neurologists in a faster and more accurate diagnosis of MS.