Assessment of proprioception impairments in children with unilateral spastic cerebral palsy using a marker-less motion capture system

Prospective cohort study. Participants & Setting: 13 children with UsCP (6.8-19y, 8 females), university laboratory.

Our proprioception assessment used a Microsoft Kinect 2. The child was seated 1m from the front of the Kinect. During testing, the child was blindfolded. Children were asked to perform three different tasks. In each task, the experimenter first moved the child’s arm (model arm) to a raised position lateral to the trunk. The child would hold the model arm in position for three seconds. During this interval, the Kinect would capture xyz coordinates of each of the child’s upper body joints. In task 1 (ipsilateral remembered), after the child held the model arm in position for three seconds, the child would relax the arm to the side of the body, and then reposition the same arm (test arm), in the same position as the model arm had been placed. In task 2 (contralateral remembered), after the child held the model arm in position for three seconds, the child would relax the arm to the side of the body, and then position the contralateral arm (test arm) in a position that mirrored the model arm position. In task 3 (match), after the child held the model arm in position for three seconds, the child would maintain the position of the model arm, and position the contralateral arm (test arm) to mirror the position of the model arm. During each of these tasks, the Kinect would capture coordinates of the test arm after the child had positioned the arm. Both the more- and less-impaired arms were used, in turn, as the model and test arm for each task. Kinect files were processed offline in MATLAB. Differences in joint positions of the model vs. test arm were quantified. We measured unimanual hand movement speed using the Jebsen-Taylor Test of Hand Function (JTTHF) of each child. A blinded physical therapist also scored proprioception of each child using a clinical rating.

All children showed substantial deficits in proprioception. Proprioception scores were similar for each child for the three tasks, and were similar whether the more- or less-impaired arm was used as the model or test arm. There was not a correlation between JTTHF score and proprioception. Our platform could be easily used to test children with USCP, and provides quantified assessments of proprioception. Conclusions/Significance: Proprioception is an important component of motor control, but has been difficult in the past to quantify. We have developed a system that can quantify proprioception in children with USCP using low-cost technology. This tool can be used to further understand proprioception in children with CP.