Robotic Assessment of Wrist Proprioception During Kinaesthetic Perturbations: A Neuroergonomic Approach

Abstract

Position sense refers to an aspect of proprioception crucial for motor control and learning. The onset of neurological diseases can damage such sensory afference, with consequent motor disorders dramatically reducing the associated recovery process. In regular clinical practice, assessment of proprioceptive deficits is run by means of clinical scales which do not provide quantitative measurements. However, existing robotic solutions usually do not involve multi-joint movements but are mostly applied to a single proximal or distal joint. The present work provides a testing paradigm for assessing proprioception during coordinated multi-joint distal movements and in presence of kinaesthetic perturbations: we evaluated healthy subjects' ability to match proprioceptive targets along two of the three wrist's degrees of freedom, flexion/extension and abduction/adduction. By introducing rotations along the pronation/supination axis not involved in the matching task, we tested two experimental conditions, which differed in terms of the temporal imposition of the external perturbation: in the first one, the disturbance was provided after the presentation of the proprioceptive target, while in the second one, the rotation of the pronation/ supination axis was imposed during the proprioceptive target presentation. We investigated if (i) the amplitude of the perturbation along the pronation/supination would lead to proprioceptive miscalibration; (ii) the encoding of proprioceptive target, would be influenced by the presentation sequence between the target itself and the rotational disturbance. Eighteen participants were tested by means of a haptic neuroergonomic wrist device: our findings provided evidence that the order of disturbance presentation does not alter proprioceptive acuity. Yet, a further effect has been noticed: proprioception is highly anisotropic and dependent on perturbation amplitude. Unexpectedly, the configuration of the forearm highly influences sensory feedbacks, and significantly alters subjects' performance in matching the proprioceptive targets, defining portions of the wrist workspace where kinaesthetic and proprioceptive acuity are more sensitive. This finding may suggest solutions and applications in multiple fields: from general haptics where, knowing how wrist configuration influences proprioception, might suggest new neuroergonomic solutions in device design, to clinical evaluation after neurological damage, where accurately assessing proprioceptive deficits can dramatically complement regular therapy for a better prediction of the recovery path.

Keywords: biomechanics; motor control; multi-joint; proprioception; robotic assessment; static perturbation.

Figure 1

(A) The WristBot device. (B) Experimental setup. The subject is comfortably seated on a chair with the right forearm fixed on the WristBot robotic device while holding its handle. In the contralateral hand the subject holds the button to press during the proprioceptive “Matching Phase.” The subject wears a mask over his eyes to perform the experiment based only on his proprioceptive feedback. (C,D) are represented the temporal sequences for the two JPM conditions: JPMUPJPM_UP and JPM_PPPM. From the initial position, the wrist joint is passively moved towards the proprioceptive target (passive reaching) and then maintained for 3 s. An auditory cue marks that the proprioceptive target is reached. After returning to the resting position participants are asked to match the target, as accurate as possible (Matching Phase) by pressing the button with the contralateral hand. Another auditory cue signals to the subject the start of the Passive Matching Phase in which it is required to stop the robot once the same movement amplitude has been perceived. In different temporal moments, depending on the condition experienced, a perturbation is given (angular rotation along the PS axis of a certain random amplitude). This is evidenced by the red arrow in the figure. Orange dot represents the device end-effector position, while the black dot represents the proprioceptive target position.

Figure 2

Comparison between the two experimental conditions (JPM_UP vs. JPM_PP) for the (A) Error Bias and for the (B) Matching Error outcomes for AA (light blue) and FE (orange) DoFs. Each gray point represents the average result for a single subject. The mean result across the population is reported as light blue point for AA and orange point for FE joint. The line through the origin (equality line) is represented by a black line; if the subject performance stays above this line the error is higher for the JPM_PP task, vice versa if it stays under the line.

Figure 3

Probability density distributions for the Error Bias of the two DoFs AA (A,B) and FE (C,D) in both the JPM_UP (first column) and JPM_PP (second column) conditions. Colored lines show the mean distribution for the specific perturbation denoted in the legend. The vertical dotted line highlights the error equal to zero, a distribution shifted to the left indicates error undershooting, while a distribution shifted to the right represents a tendency of target overshooting.

Figure 4

Outcome measures relative to the two DoFs: AA on the top and FE on the bottom for JPM_UP (gray) and JPM_PP (magenta) conditions. First column represents the Error Bias (A). Second column is relative to the Matching Error (B). On the x-axis is evidenced the amount of angular perturbation provided along the PS axis (−45°, −20°, −5°, 0°, 5°, 20°, 45°) during the experiment.