Evaluation of Tremor Degree Using Triangulation Estimation for IMU Volumetric Method Based on Robust Mahalanobis Distance for Outlier Detection

Parkinson’s disease (PD) is a rapidly progressive neurodegenerative disorder whose core motor symptoms—tremor, bradykinesia, and rigidity—severely impair patients’ quality of life. Conventional clinical rating scales are inherently subjective and exhibit limited inter-rater reliability, underscoring the urgent need for objective and quantitative assessment tools. This article presents a novel proof-of-concept study that lays the foundation for objective tremor quantification based on inertial measurement unit (IMU) signals. The framework consists of three stages: first, a robust Mahalanobis distance algorithm based on the minimum covariance determinant (MCD) estimator eliminates multivariate outliers from raw IMU data; second, acceleration in the sensor coordinate system is converted to the geographic coordinate system via quaternion rotation, the gravitational component is removed, and 3D tremor trajectories are reconstructed using Simpson’s numerical integration; third, Delaunay triangulation partitions the reconstructed point cloud, constructs pentahedral voxels, and calculates the enclosed volume, which serves as an intuitive metric for tremor energy and spatial activity range. The method was evaluated using a BWT61CL IMU sensor (sampling frequency 100 Hz) to collect hand tremor data from four PD rest tremor (RT) patients and four matched healthy control (HC) participants under standardized rest conditions. To assess the ability of the method to distinguish pathological tremor from physiological motion, the control group was expanded to include four Essential tremor (ET) patients. This revealed a measurable reduction in tremor volume post-treatment compared to baseline, consistent with the expected direction of clinical observations. These preliminary findings suggest that the proposed IMU-based volume method, combining robust Mahalanobis distance and Delaunay triangulation, has potential for tracking the spatiotemporal evolution of tremor symptoms in PD, offering a proof-of-concept analytical framework for quantifying tremor spatial amplitude that warrants validation in larger cohorts.