Reliability of tremor measurements using a multidimensional electromagnetic sensor system

Abstract

Objective: To investigate the reliability of repeated measurements of normal physiological tremor made with a multidimensional measurement system.

Equipment: Measurements of postural upper limb tremor at the hand were made in 10 normal subjects using a 3Space Fastrak^® (Polhemus, Inc.) which detects movement over six degrees-of-freedom (three of the measurement directions were examined).

Data collection: Reliability was assessed for two alternative upper limb postures (arm straight or bent) and data were collected on two days, twice at each session, to determine the repeatability between and within recording sessions.

Data analysis: The data were split into segments and subjected to autoregressive (AR) modelling. Three parameters (one for each of the measurement directions examined) were extracted from the models and used as variables for the reliability analysis.

Statistical tests: Variation within and between sessions was assessed by finding the median differences between efforts and days for each subject and then finding the overall median value and the corresponding 97.9% confidence intervals for each movement. This produced estimates of the population median value and indicated the precision of the estimates.

Results: All the confidence intervals encompassed the zero median difference point indicating that, in the population, this technique would produce repeatable results. For between-efforts comparisons there was some evidence that data collected for the bent arm posture were more repeatable than for a straight arm.

Conclusion: Normal physiological tremor can be measured reliably, within and between sessions, using the 3Space Fastrak system.