Handwriting Analysis for Early Detection ofAlzheimer's Disease

Neurodegenerative diseases like Alzheimer's and Parkinson's impair the cognitive and motor abilities of the patient, which brings memory loss and confusion. As handwriting involves the functioning of the brain and motor control, it is affected. Alteration in handwriting is one of the first signs of Alzheimer's disease (AD). The handwriting gets shaky, due to loss of muscle control, confusion, and forgetfulness. The symptoms get progressively worse. The handwriting becomes illegible and phonological spelling mistakes become inevitable. In this study we are using the process of extracting features, which is to be used as a parameter for diagnosis. Deep learning technology is chosen as the technical tool to identify and classify common features of handwriting of patients with AD. Variational auto encoder (VAE), a deep unsupervised learning technique, has been applied, which is used to compress the input data and then to reconstruct it keeping the targeted output the same as the input. This study is aimed at successfully extracting distinguishable characteristics in the handwritten samples with the help of image segmentation and VAE reconstruction, which can be used as a diagnostic tool for early detection of AD.