Multiple Sclerosis Classification Using Deep Learning Techniques

Abstract

The diagnosis of Multiple sclerosis with different types is a big challenge for the doctor and takes more time in real life. We develop two deep learning techniques in order to classify the MS type. The MS has four types: MS-axial, control-axial, MS-sagittal, and control-sagittal. After that, we apply many preprocessing steps to the dataset in order to make it suitable to feed to the classification process like convert the target class label to numeric. We used four evaluation metrics to compare deep learning models: VGG19 and VGG16: recall, f1-score, accuracy, and precision. The results showed that the VGG19 gave better results compared with the VGG19 model in terms of four evaluation metrics of accuracy = 98.6%. The results indicated that we can rely on VGG19 in the classification process for many MS types.