Motor control of the lower limbs while walking with the TWIN exoskeleton operated by TWINActa in healthy subjects

Abstract

Lower limb exoskeletons are medical devices used in the rehabilitation to provide physical therapy to individuals with neurological disorders [1]. These robotic devices were originally designed and developed for spinal cord injury subjects, who do not have any residual walking abilities, and are therefore mainly based on an assistive approach. To successfully use these machines in subjects with residual capacities such as persons post stroke, the control system must be adapted to an assist-as- needed control scheme. This approach requires the subject active participation to stimulate the voluntary muscle contraction of the paretic limb and manage the somatosensory feedbacks subsequent to the action. For these reasons, we have developed TWINActa, a control suite for gait training following stroke, that provide asymmetric assistance to the user. The assistance-as-needed can be provided in the following modes: (i) support during the toe-off phase of the non-paretic leg; (ii) pelvic tilt damping during stance phase; (iii) stabilization of the knee joint of the supporting leg during the double support phases; and (iv) hip extension support of paretic leg during the stance phase. This study was aimed at verifying if motor control during a walking task performed with an exoskeleton, operated by TwinActa, is similar to the physiological control of walking in healthy subjects. Five healthy volunteers were asked to perform motor tasks (i.e., overground walking, stairs ascending and stair descending) wearing the TWIN exoskeleton [2] operated by TWINActa control system. During the task EMG signals were acquired from 12 muscles (10 placed on the lower limb and 2 on the upper limbs per side) and movement data were recorded from 2 IMUs placed on the shanks. Non-negative matrix factorization was carried out on the EMG recorded signal to identify muscle synergies [3]. The EMG activity patterns were also mapped onto the rostro-caudal location of ipsilateral motoneurons pools in the human spinal cord [4]. Based on the Kendall myotomal charts of segmental localization [5], we reconstructed the sacral (S1-S2) and lumbard (L2-L5) motor pool activation profiles. Pearson's correlation coefficient was used to evaluate the similarity degree of 1) muscle synergies and 2) sacral and lumbar motor pool activation profiles of walking with exoskeleton with respect to the normative reference (without device) provided by Eurobench project (Pepato scenario). Four muscle synergies accounted for a R2 > 0.85 for all participants during overground walking. A strong degree of similarity (correlation) of activation of sacral and lumbar motor pools with respect to the reference group was found (mean ± SD, 0.77 ± 0.06 and 0.78 ± 0.11, for sacral and lumbar spinal output respectively). Also the mean similarly of the muscle synergies activation profiles was strong (mean ± SD, 0.70 ± 0.14) with respect to the four physiological muscle synergies involved in gait control [6]. The four muscle synergies involved in the gait control [6] were found while walking with the exoskeleton operated by TWINActa in neurologically intact subjects. We found also a strong degree of similarity between the motor pools wearing the exoskeleton with respect to the normative reference of walking without the device. Since the spinal motor pools are associated with functional grouping of motoneurons of the lower limb muscles, our findings suggest that the Twin exoskeleton operated by TWINActa can be used as rehabilitation device to enhance the recovery of the physiological motor control in persons with residual ability. Further studies need to be conducted to verify this finding in subjects with neurological diseases.