Proprioceptive Augmentation With Illusory Kinaesthetic Sensation in Stroke Patients Improves Movement Quality in an Active Upper Limb Reach-and-Point Task

Abstract

Stroke patients often have difficulty completing motor tasks even after substantive rehabilitation. Poor recovery of motor function can often be linked to stroke-induced damage to motor pathways. However, stroke damage in pathways that impact effective integration of sensory feedback with motor control may represent an unappreciated obstacle to smooth motor coordination. In this study we investigated the effects of augmenting movement proprioception during a reaching task in six stroke patients as a proof of concept. We used a wearable neurorobotic proprioceptive feedback system to induce illusory kinaesthetic sensation by vibrating participants' upper arm muscles over active limb movements. Participants were instructed to extend their elbow to reach-and-point to targets of differing sizes at various distances, while illusion-inducing vibration (90 Hz), sham vibration (25 Hz), or no vibration was applied to the distal tendons of either their biceps brachii or their triceps brachii. To assess the impact of augmented kinaesthetic feedback on motor function we compared the results of vibrating the biceps or triceps during arm extension in the affected arm of stroke patients and able-bodied participants. We quantified performance across conditions and participants by tracking limb/hand kinematics with motion capture, and through Fitts' law analysis of reaching target acquisition. Kinematic analyses revealed that injecting 90 Hz illusory kinaesthetic sensation into the actively contracting (agonist) triceps muscle during reaching increased movement smoothness, movement directness, and elbow extension. Conversely, injecting 90 Hz illusory kinaesthetic sensation into the antagonistic biceps during reaching negatively impacted those same parameters. The Fitts' law analyses reflected similar effects with a trend toward increased throughput with triceps vibration during reaching. Across all analyses, able-bodied participants were largely unresponsive to illusory vibrational augmentation. These findings provide evidence that vibration-induced movement illusions delivered to the primary agonist muscle involved in active movement may be integrated into rehabilitative approaches to help promote functional motor recovery in stroke patients.

Keywords: Fitts' law; kinematics; reaching task; sensory-motor rehabilitation; stroke; vibration illusion.

Figure 1

Experimental setup.

Figure 2

Directness is measured by (A) the ratio between the length of the traveled trajectory and the length of a linear path between the starting and the final positions as well as the root mean square errors (RMSE) of the distance of the trajectory from the linear path projected in the (B) transverse plane and (C) sagittal plane. Higher values indicate a less direct trajectory. In each subplot medians and interquartile ranges [25th and 75th percentile (Q25, Q75)] with whiskers indicating the range of non-outlier values are shown for data aggregated across all participants and targets for both able-bodied and stroke patient participants when no vibration (white, NO VIB), 90 Hz vibration on the triceps (orange, TRI 90), 90 Hz vibration on the biceps (blue, BI 90), 25 Hz vibration on the triceps (yellow, TRI 25), and 25 Hz vibration on the biceps (light blue, BI 25) was applied. The statistical differences indicated (^*p < 0.05, ^**p < 0.001) refer to the main effect of the experimental condition.

Figure 3

Six representative movement trajectories projected in the transverse plane, each from a different (A) able-bodied and (B) stroke patient participant reaching a target set at 80% of the maximum reachable distance during three experimental conditions: no vibration (NO VIB), 90 Hz vibration applied to the triceps (TRI 90), and 90 Hz vibration applied to the biceps (BI 90). Dashed lines represent the ideal linear path while the solid lines represent the actual trajectory.

Figure 4

Example velocity profiles from (A) one able-bodied participant and (B) one stroke patient for the five experimental conditions while reaching a target at 80% of the maximum reachable distance. The dots represent the local peaks with a minimum prominence of 5 cm/s. The numbers in the right corner of each graph represent the traveled trajectory length (TL) and the smoothness index (Sm: -n. peaks/trajectory length). Higher (less negative) smoothness index values indicate a smoother movement.

Figure 5

(A) Movement smoothness was quantified as [-(number of velocity peaks/trajectory length)], where higher (less negative) values represent smoother movements. (B) Peak elbow angles were normalized and averaged across 12/12 able-bodied and 5/6 stroke patient participants. In each subplot medians and interquartile ranges [25th and 75th percentile (Q25, Q75)] with whiskers indicating the range of non-outlier values are shown for data aggregated across all participants and targets for both able-bodied and stroke patient participants when no vibration (white, NO VIB), 90 Hz vibration on the triceps (orange, TRI 90), 90 Hz vibration on the biceps (blue, BI 90), 25 Hz vibration on the triceps (yellow, TRI 25), and 25 Hz vibration on the biceps (light blue, BI 25) was applied. The statistical differences indicated (^*p < 0.05, ^**p < 0.001) refer to the main effect of the experimental condition.

Figure 6

Fitts' law parameters included (A) the expected over the prescribed target distance (De/D), (B) the normalized movement time, and (C) the ratio between the effective and prescribed index of difficulty (IDe/ID). In each subplot medians and interquartile ranges [25th and 75th percentile (Q25, Q75)] with whiskers indicating the range of non-outlier values are shown for data aggregated across all participants and targets for both able-bodied and stroke patient participants. The statistical differences indicated (^*p < 0.05, ^**p < 0.001) refer to the main effect of the experimental condition. (D) The relationship between the movement time and the effective index of difficulty averaged across all participants in the two groups, stroke patients and able-bodied participants, is also shown. The slopes of the lines in (D) represent the inverse of the throughput, which is expressed in bit/s.

Figure 7

Questionnaire results for perceived (A) fatigue and (B) movement accuracy during the task reported by both able-bodied (purple) and stroke patient participants (green). The mean value and the standard deviations were calculated across the ratings of the three repetitions of each experimental condition.