A Novel Classification of Multiple Sclerosis Using Quantum Convolution Neural Network

Abstract

Multiple Sclerosis (MS) is a neurological disorder that affects the central nervous system (CNS), including the brain, spinal cord, and optic nerve. The aim was to enhance the efficacy of a quantum machine-learning algorithm in identifying MS and categorizing its advancement by creating advanced techniques.Detecting MS lesions has become increasingly difficult owing to the imbalanced nature of the dataset, which contains a disproportionately small number of lesion pixels.Subsequently, a novel feature selection ensemble (FS-Ensemble) technique was implemented, which employed four distinct filtering methods for selecting features, utilizing various statistical techniques, including methods such as the chi-square test, information gain, Minimum Redundancy Maximum Relevance, and RelieF.Subsequently, an innovative classification approach utilizing a Quantum Convolutional Neural Network (QCNN)was applied to identify the most crucial features.The findings demonstrated the efficacy of MRI in examining MS lesions, with a study involving 30 patients with MS and 100 healthy brain MRI scans showing accurate predictions of disease progression.In the realm of MS identification, QCNN exhibited exceptional performance, achieving an accuracy of 87.7% and sensitivity of 87.0%.Other notable metrics included specificity of 88.5%, precision of 88.7%, and AUC of 0.8775.Furthermore, studies suggest that employing non-shared parameters and more sophisticated filter designs can significantly enhance QCNN's efficacy of QCNNs.These findings contribute to the development of powerful quantum classifiers for multicategory image classification challenges.