Detection of Parkinson's Disease Using Deep Learning: A Review
Rohit Kumar Singh, Swapnil Srivastava, Abhi Verma, Aditya Choudhary, Ishika Malik, Anmol Dhiman
Abstract
Parkinson's disease is a disorder that affects the nervous system and can worsens over time if left unmonitored. It can lead to problems associated with movements like shakiness, slow movement, etc. It can also affect non-motory actions like mood swings and disturbance in sleep patterns. Thus, early diagnosis of disease will help in delivering effective treatment to patients. However, traditional diagnostic methods like clinical observation and the Unified Parkinson's Disease Rating Scale (UPDRS) are often time consuming and sometimes expert's opinion driven. Recent progress in deep learning offers new ways to detect Parkinson's disease automatically and more accurately. The non-invasive methods like speech, handwriting, gait, and imaging data have aided in this direction. This review spans different implementations of DL-based PD detection by comparing performance metrics, datasets, and frameworks among them. Key challenges and potential future directions are also discussed at the concluding section to summarize the utility of the study.