A Method for Identifying Parkinson’s Disease from Spiral Images (Graphs) Created by Hand

A chronic brain ailment known as Parkinson’s disease (PD) affects millions of people around the globe. The breakdown or death of brain cells responsible for producing dopamine, a neurotransmitter that regulates movement, causes this condition. Issues with mobility, equilibrium, and posture are brought on by PD, which develops as a result of this. To lessen the disease’s course and boost patients’ quality of life, early diagnosis is essential. A cosine annealing scheduler and deep transfer learning are presented in this study as a method for handwriting-based prediction. It uses handwriting samples from people with and without Parkinson’s disease, found in the NIATS collection. With the use of spiral sketching and CNN, this research hopes to provide a PD detection method. The main concept is to determine whether a person is healthy or has Parkinson’s disease based on their spiral drawings. Spiral drawings made by individuals in good health resemble conventional spiral forms. People with Parkinson’s disease have sluggish movement and poor hand-brain coordination, therefore their spiral drawings seem deformed because they stray greatly from the correct curve. This research presents the results of using a Convolutional Neural Network to diagnose Parkinson’s disease; the classification accuracy achieved was 83.6%.