Detection of Parkinson Disease in Brain MRI using Convolutional Neural Network

Abstract

Parkinson Disease (PD) is one of the most critical progressive neurological diseases which mainly affects the motor system. The accurate diagnosis of PD has been a challenge to date, mainly due to the close relevance of PD to other neurological diseases. These close characteristics are the reasons that cause 25% inaccurate manual diagnosis of PD. In this paper, we present a Convolutional Neural Network (CNN) based automatic diagnosis system which accurately classifies PD and healthy control (HC). Parkinson's Progression Markers Initiative (PPMI) provides publically available benchmark T2-weighted Magnetic Resonance Imaging (MRI) for both PD and HC. The mid-brain slices of 500, T2-weighted MRI are selected and aligned using image registration technique. The performance of the proposed technique is evaluated using accuracy, sensitivity, specificity and AUC (Area Under Curve). The detailed comparison in the result section shows that the CNN archived a better performance from 3%–9% in terms of accuracy, sensitivity, specificity, and AUC when compared to the some existing techniques.