Performance Analysis of Different Machine Learning Classifiers in Detection of Parkinson’s Disease from Hand-Drawn Images Using Histogram of Oriented Gradients

The Parkinson’s disease is a neurodegenerative disorder that has affected millions of people and found mostly in aged people. It occurs due to loss of dopaminergic neurons in substantia nigra part which is found in thalamic region of human brain. Diagnosis of Parkinson’s disease is very much costly. Most of the research works that have been performed to detect Parkinson’s disease are based on speech utterances, kinematic features, pen-based features, etc. In this paper, we aim to simplify the process for early detection of Parkinson’s disease by relying only on hand-drawn figures taken from the disease-affected patients. Histogram of oriented gradients features has been extracted from different types of images, which act as an input to various machine learning classifiers such as k-nearest neighbour, random forest, support vector machine, Naive Bayes, multi-layer perceptron, and the performance of different classifiers is shown. Experimental result analysis shows that with the available training datasets, Dataset 1 and Dataset 2 have achieved 74.7% and 96.8% of accuracy, respectively.