Parkinson's Disease Handwriting Detection Using FCNN

EasyChair Preprint 7523

Abstract

Detection of Hypo-kinetic Rigid Syndrome handwriting patterns is productive for the diagnosis of Hypo-kinetic Rigid Syndrome. Nowadays, a little computer-aided method based on computer vision points out the Hypo-kinetic Rigid Syndrome handwriting. This paper proposed a computer vision-based recognition system to recognize the handwriting patterns of Hypo-kinetic Rigid Syndrome. We trained a fully convolutional neural network to classify the high-resolution images in the NITS of the Federal University of Uberlândia training set into two different classes. Our fully convolutional neural network is simple, accurate, efficient, and works on challenging computer vision tasks. The result of the analysis of the primary objectives is to classify between normal patient and Parkinson's disease patient is done very effectively than previous research. Numerical information about the above analysis, such as in terms of accuracy, recall, precision \& F1 score is done completely. Our suggested system can detect Parkinson's patients with an accuracy of 92.43%. Our study found that the Fully Convolutional Neural Network is more effective than other methods on all outcome measures to detect Parkinson's Disease. A likely explanation is that the proposed system used a different type of deep learning approach, which increases the accuracy.

Keyphrases: Fully Convolutional Neural Network, Handwriting analysis, Parkinson’s disease, computer vision

@booklet{EasyChair:7523,
  author    = {Minhazul Arefin and Kazi Mojammel Hossen and Rakib Hossen and Mohammed Nasir Uddin},
  title     = {Parkinson's Disease Handwriting Detection Using FCNN},
  howpublished = {EasyChair Preprint 7523},
  year      = {EasyChair, 2022}}