Automatic Prediction of the Conversion of Clinically Isolated Syndrome to Multiple Sclerosis Using Deep Learning.

Multiple sclerosis (MS) is a chronic inflammatory disease of the central nervous. Disability can be prevented by early detection of lesions. Deep learning techniques, such as convolutional neural networks (CNN), can learn patterns on brain magnetic resonance image (MRI) so as to predict the conversion of clinically isolated syndrome (CIS) to definite multiple sclerosis. The aim of this paper is to develop a method to automatically detect the conversion. The proposed algorithm is an improved CNN which uses LeNet architecture coding in Python and Keras library. It consists of different convolutional layers which learn the patterns of input images by using convolutional filters. ReLU activation function and max-pooling are used to reduce the dimensions of images for efficient and fast processing. A detailed investigation of automatic prediction algorithm is performed on the MRI images of 21 patients. The 21 patients were scanned at onset of CIS and a year later (of whom, 11 converted to MS and 10 did not convert to MS). The proposed deep learning algorithm predicted the presence of MS with an accuracy of 83.3% and 100% in two experiments. In the first experiment, 5 out of 6 patients were predicted correctly. In the second experiment, 6 out of 6 patients were predicted correctly. The experiments have proved that the proposed method is an automated system which can predict the disease accurately and quickly, contributing to the prevention and alleviation of the disability caused by the disease.