A Deep Learning Method to Detect Parkinson’s Disease from MRI Slices

Neurodegeneration starts with the deterioration of the substantia nigra in Parkinson’s disease, progresses over time and causes preciseness, automatism, and agility in the patient’s movement. Parkinson’s disease ought to be treated with extraordinary significance in terms of diagnosis and treatment. Currently, the computer-aided prognosis is getting widespread in Parkinson’s disease. Although the prognosis method for Parkinson’s disease consists of considering medical records and neurological examination, there are more than a few research totally based on medial pictures to detect Parkinson’s disease in the literature. In this study, a supervised classification-based deep learning method has been proposed to distinguish Parkinson’s patients from a control group consisting of healthy individuals, aimed at detecting degeneration in the brain using MR images. Accordingly, the classification performance of the proposed method is 90.36% for the accuracy metric and 90.51% for the area metric under the ROC. Furthermore, precision, sensitivity, and F1 scores are computed to be 90.08%, 90.52%, and 90.25%, respectively.