A Hybrid Framework for the Diagnosis of Parkinson’s Disease using Handwritten Drawings-Spiral and Wave

Abstract

Parkinson's disease is a progressive neurological disorder that significantly affects individuals worldwide. Early and accurate classification of the disease is crucial for timely intervention and improved patient outcomes. This study aims to develop an effective classification system using drawings of spirals and waves to discriminate between healthy individuals and those with Parkinson's disease, aiming to provide an early diagnostic method, leading to improved patient lifespan. The study utilizes two sets of drawings: spirals and waves. Data augmentation techniques are employed to increase the dataset size and enhance training data for deep neural networks. The Pyramid Histogram of Oriented Gradients (PHoG) algorithm is applied to compute shape descriptors from healthy and Parkinson's drawings. A Visual Geometry Group (VGG)-based deep learning model is used to extract significant features from the modified drawings, particularly from the fc6 and fc7 layers. Supervised classifiers, Support Vector Machine (SVM) and K-Nearest Neighbor (KNN), are employed individually and in combination to classify the extracted features. The results demonstrate that the fused features achieved the highest accuracy values: 98.6% for spiral drawings using SVM and 96.57% for wave drawings using KNN. These accuracy rates highlight the effectiveness of the proposed method in accurately classifying Parkinson's disease based on drawings of spirals and waves. The findings suggest that the proposed method has the potential to serve as a non-invasive and reliable tool for early diagnosis of Parkinson's disease. It can enable timely interventions and improved patient care.