Low Amplitude Hand Rest Tremor Assessment in Parkinson’s Disease Based on Linear and Nonlinear Methods

The assessment of rest tremor can help following up the progress of Parkinson's disease (PD). Several studies have found that low amplitude tremor may not be detected accurately. This research aims to fill this need in the area. Rest tremor was measured by an accelerometer positioned on the hand of 17 people with PD during the OFF-medication period (i.e., without drug effect). Each participant performed the hand rest task as defined in the Unified Parkinson's Disease Rating Scale (MDS-UPDRS). The estimated MDS-UPDRS scores for hand tremor assessment was between 0 and 2 for all individuals, confirming the presence of low amplitude tremor. The tremulous activity was decomposed by Empirical Mode Decomposition (EMD) into Intrinsic Mode Functions (IMFs). The power spectral density estimated by Adaptive Sine Multitaper was employed for the identification of a physical meaning for specific IMFs. It was found that the second IMF (IMF2) represents the low amplitude rest tremor oscillating between 3 and 8 Hz. The Root Mean Square (RMS) and the Approximate Entropy (ApEn) were calculated for IMF2. The Spearman's rank correlation coefficient was used to estimate the correlation between RMS and ApEn. The results showed that these features can be used for the quantification of low amplitude rest tremor, and that they have a strong negative correlation (p = -0.81), as usually observed in moderate to intense tremulous activity. These findings may be employed to detect and monitor the progress of tremor, and thus for the adjustment of drug and surgical based treatments.