A New Approach for Parkinson's Disease Imaging Diagnosis Using Digitized Spiral Drawing

Parkinson's disease is the most general neurodegenerative disease, upsetting notably motor functions of elderly persons. The diagnosis and monitoring of Parkinson's disease is a costly and inconvenient process even today, in particular, in developing parts of the world. The observable symptoms of Parkinson's disease at early stages include disorders in handwriting and repetitive tasks of spiral drawing. With the advancement of IT, it is easier to collect spiral drawing samples using digitized tablets. We proposed detailed analysis of static and dynamic spirals drawn by Parkinson's disease patients. For this, nearly all kinematic variables are taken out from data files generated for 25 patients and 15 healthy controls, using mathematical models and from PNG drawing files of 15 patients and 15 healthy controls. The second dataset of 30 persons is evaluated with HOG feature engineering and classified using ensemble and machine learning models. Results demonstrated nearly 91% classification accuracy to separate Parkinson's disease patients from healthy controls by applying feature engineering and three machine learning classifiers: Logistic regression, C-support vector classification and ensemble model random forest. This chapter confirms that digitized spiral drawings have a major impact on the classification of Parkinson's disease patients and healthy controls and hence can support future diagnosis and treatment of Parkinson's disease based on data files as well as digital drawings.