The Diagnosis of Parkinson's disease Based on Human Brain Imaging

Parkinson's disease (PD) is a neurodegenerative condition in which the lowered levels of the neurotransmitter “dopamine” and a sort of neurotransmission changes in the basal ganglia are supposed to play key roles in movement problems and symptoms. In recent years, the structural brain networks (SBNs)-based diagnosis of PD has been sought-after by many researchers due to an uptick in the prevalence of PD and the development of innovative brain imaging techniques. This study aims to diagnose PD by analyzing the data obtained from “resting-state functional magnetic resonance imaging” (rs-fMRI). The proposed algorithm explores the differences between “healthy individuals” and “PD patients” based on statistical data and cross-correlation assessments. In this algorithm, after pre-processing the images and creating the time series of the 16 brain regions, the features of these regions are extracted and the best attribute for each stage is chosen by analysis of variance (ANOVA) at p<0.05. The feature matrix consists of a group of features picked from time-series statistical data and cross-correlation analyses between the affected PD brain’s regions. In this study, the simulation results indicate a specificity of 97% and a sensitivity of 100% in diagnosing PD using the artificial neural network (ANN) classifier.