Evaluation Of Kinematic Signal Noise During Resistance Exercise Motion Analysis

Abstract

In order to determine joint kinematics and kinetics about three orthogonal planes, a six-degree of freedom marker set, involving clusters of tracking markers mounted on a rigid plate, is necessary. Data collected using cluster markers for gait tasks appears valid, however, when applied to other tasks, such as resistance exercise, considerable noise exists when calculating angular velocity and net joint power. PURPOSE To ascertain the major source of error for a six-degree of freedom marker set and to determine the optimal noise reduction method. METHODS Data were collected during static (sitting) and dynamic (gait and squatting) trials. Dynamic trials were performed at normal, fast and slow velocities. Reflective markers on the joint landmarks determined joint axis of rotations and clusters of markers on the thigh, shank and foot determined the position of the segment during movement. Data were collected using an 8-camera opto-electric motion analysis system at 120Hz. From segment positions, joint angles were determined and differentiated to yield joint angular velocities. Digital filtering and moving average smoothing techniques were applied to coordinate and/or joint angular data to determine the optimal noise reduction method. RESULTS Coordinate data for the thigh marker during the static trial indicated non-uniform marker movement up to 0.004m (relative to other markers). Moving average smoothing failed to remove this non-uniform movement. Greater noise existed for slow dynamic trials compared to normal and fast dynamic trials. Noise existed independent of the task (gait versus squatting). Digital filtering (3–5Hz cutoff frequencies) failed to reduce noise. For gait (9-points) and squatting (11-points), the moving average technique optimally reduced noise. Increasing the number of averaging points resulted in large end-point errors. CONCLUSION The major source of error for a marker set involving clusters is from tracking of individual markers and their apparent position to other markers in the cluster. This error affects kinematics the most for tasks where change in segment position between two adjacent time points is small, such as during resistance exercise. For squatting, the optimal noise reduction method is the moving average technique applied to joint angular data, however, further evaluation is required to reduce the initial error as the smoothing technique may not be optimal for all tasks.