AI WALKUP: A Computer-Vision Approach to Quantifying MDS-UPDRS in Parkinson's Disease

Xiang Xiang,Zihan Zhang,Jing Ma,Yao Deng

2024-04-02

Abstract:Parkinson's Disease (PD) is the second most common neurodegenerative disorder. The existing assessment method for PD is usually the Movement Disorder Society - Unified Parkinson's Disease Rating Scale (MDS-UPDRS) to assess the severity of various types of motor symptoms and disease progression. However, manual assessment suffers from high subjectivity, lack of consistency, and high cost and low efficiency of manual communication. We want to use a computer vision based solution to capture human pose images based on a camera, reconstruct and perform motion analysis using algorithms, and extract the features of the amount of motion through feature engineering. The proposed approach can be deployed on different smartphones, and the video recording and artificial intelligence analysis can be done quickly and easily through our APP.

Computer Vision and Pattern Recognition,Artificial Intelligence,Image and Video Processing,Signal Processing

What problem does this paper attempt to address?

This paper attempts to solve the problems of subjectivity, lack of consistency, high cost and low efficiency in the evaluation process of Parkinson's Disease (PD). Specifically, traditional evaluation methods rely on the Movement Disorder Society - Unified Parkinson’s Disease Rating Scale (MDS - UPDRS), and this method has the following problems in practical applications:

1. **Highly subjective**: There may be large differences in scores among different evaluators.
2. **Lack of consistency**: For the same patient, the results may be inconsistent when evaluated at different times or by different doctors.
3. **High cost and low efficiency**: Manual evaluation requires the participation of professionals, increasing the burden on medical resources and being less efficient.

To solve these problems, the author proposes a computer - vision - based method. It captures action videos of patients through the smartphone camera and uses algorithms for motion analysis and feature extraction. This method can achieve an objective and quantitative description of the motion state of Parkinson's disease patients, thus assisting doctors in evaluating the condition more accurately.

### Main objectives

- **Automated evaluation**: Use computer vision technology to automatically quantify six key motor function indicators in MDS - UPDRS (such as tremors, finger tapping, hand movements, etc.), reducing the influence of human factors.
- **Improve evaluation efficiency**: Quickly record and analyze videos through smartphone applications, simplify the data collection process, and reduce the burden on patients and medical staff.
- **Standardize evaluation**: Ensure the consistency and objectivity of evaluation, reduce individual subjective evaluation bias, and standardize the assessment of the severity of Parkinson's disease.

### Specific methods

1. **Data collection**: Use the smartphone camera to record specific action videos of patients.
2. **Pose estimation**: Use tools such as MediaPipe Pose and MediaPipe Hand for human key - point detection to obtain 3D pose information.
3. **Feature extraction**: Design six motion signals to evaluate different MDS - UPDRS items, and use the tsfresh library to extract time - series features from these signals.
4. **Visualization and analysis**: Visualize the extracted features to help doctors and patients understand the changes in the motion state.

Through this method, researchers hope to significantly improve the efficiency and reliability of Parkinson's disease evaluation without affecting the accuracy of the evaluation.