Task-Specific Tremor Quantification in a Clinical Setting for Parkinson’s Disease

Abstract

Objective analysis of tremors in Parkinson’s disease (PD) using wearable sensors has been studied, however, there is a lack of studies addressing the discrimination between postural and hand rest tremors. This paper proposed a new method to study rest and postural tremors, exploring amplitude and bandlimited signal energy as extracted features, and their correlations with Unified Parkinson’s Disease Rating Scale (UPDRS). Eleven people with PD executed rest and postural tasks with inertial measurement units (IMUs) placed on their hands (IMU1) and forearms (IMU2). Each IMU contained a triaxial accelerometer (A), gyroscope (G), and magnetometer (M). The features were extracted from the resultant, X (anterior–posterior), Y (medial–lateral) and Z (vertical) components. IMU1-A yielded the highest mean correlation with the UPDRS scores (IMU1: r(A) = 0.72, p = 0.013) for the rest task, while there was no difference of results for the accelerometer (IMU1: r(A) = 0.85, p = 0.001) and gyroscope (IMU1: r(G) = 0.85, p = 0.003) in the postural task. The statistical analysis confirmed that postural and rest tremors can be studied by the components of the following uniaxial sensors: IMU1-A-Z (p = 0.986), IMU1-G-Y (p = 0.118), IMU1-M-Z (p = 0.544) and IMU2-A-Z (p = 0.198). The features and their correlation with UPDRS were sensitive to the severity of wrist tremor and successfully discriminated rest from postural tremors. The results suggest that a single IMU placed at the back of the hand is sufficient to study hand rest and postural tremors.