An Effective Hand Pose Estimation Based Evaluation Method in Assessing Parkinson’s Finger Tap Movements

Parkinson’s disease (PD) a neurodegenerative disorder that affects motor function and significantly impacts the quality of life. Accurate and objective assessment of motor impairments, such as finger tapping movements, is crucial for diagnosis and monitoring disease progression. Traditional methods for assessing finger tapping movements in PD have limitations in terms of subjectivity, limited metrics, invasiveness, and limited monitoring scope. In this study, we propose an effective assessment method that combines hand pose estimation, feature extraction, and score generation to evaluate finger tapping movements in PD patients. The method utilizes non-contact hand pose estimation techniques to accurately capture hand movements. Relevant motion parameters, such as finger opening amplitude and tapping speed, are extracted using feature extraction algorithms. Then, the extracted parameters are used as features and input into a trained TCN classification network to generate an assessment score reflecting finger tapping performance. Experimental results demonstrate that the proposed method achieves an accuracy of 84.75% in evaluating finger tapping actions in PD patients, can effectively assessing finger tapping movements.