Novel Features for Diagnosis of Parkinson’s Disease From off-Line Archimedean Spiral Images

Abstract

Parkinson’s Disease (PD) is difficult to diagnose and is commonly a diagnosis of exclusion. A common early symptom of PD is handwriting and/or drawing difficulty. Most of the early systems rely on on-line handwritten / hand-drawn data which need specialized equipments to capture. Such costly systems may not be available where infrastructural facilities are limited. So we intend to devise a low cost system for the same purpose. Towards the goal, in this paper we present novel distance based features to diagnose Parkinson’s disease from off-line hand drawn Archimedean Spiral. We have tested our algorithm on a benchmark database PaHaW. Performance of our system is compared with that of some existing systems. Experimental results suggest that proposed feature works good and is better than existing systems.