Automated detection of Parkinson's disease using improved linknet-ghostnet model based on handwriting images

Parkinson's disease (PD), is a neural disorder that damages movement control, which is reflected by different non-motor and motor symptoms. PD is caused by the weakening of neurons that produce dopamine in the brain, and it includes symptoms like bradykinesia (delay in movements), stiffness, and tremors. People frequently suffer from loss of motor skills when the illness worsens, which has a big influence on everyday tasks like writing. Micrographia is a disorder marked by very tiny, cramped handwriting and is one of the symptoms of PD. As a reflection of the disease's wider motor impairments, patients may observe that their handwriting gets harder to read and control. Detecting Parkinson's disease via handwriting images is one of the major research areas in the medical field. This research proposes an automated PD detection approach with handwriting images using an improved hybrid classification model. Primarily, a modified Wiener filter is employed for pre-processing the handwriting image. Then, modified PHOG, Deep features and Shape features are extracted. Finally, detection is performed using hybrid Improved LinkNet and Ghostnet models, termed (ILN-GNet), whose outcomes indicate if the individual is healthy or affected. From the analysis, a higher precision of 0.99 is achieved by the ILN-GNet, while existing methods attained low precision. Thus, these innovations significantly enhance early diagnosis and monitoring, enabling timely interventions before the disease progresses. Moreover, the proposed approach can contribute to remote healthcare solutions, by providing a scalable, and efficient tool for PD diagnosis.