Changes in Muscle Co-activation in Spinal Cord Injured Individuals After Body-Weight Supported Treadmill Training 1 Sabrina S.M. Lee, 2Katherine Pauhl, 1Erika Harder, 2Tania Lam, 1James M. Wakeling1 1 Simon Fraser University, Burnaby, BC, Canada 2 University of British Columbia, BC, Canada E-mail: [email protected] INTRODUCTION Locomotor training with body weight supported treadmill training (BWSTT) is a commonly used rehabilitation tool for retraining individuals with incomplete spinal cord injury (SCI) [1]. It is often augmented with robotic devices to assist leg movements [2], however, there is recent evidence that resistance to leg movements, instead of assistive forces, results in increased flexor muscle activity through enhanced afferent feedback [3]. As sufficient foot clearance can be problematic for SCI individuals, the timing of muscle activation is crucial for normal gait. Electromyography (EMG) is often used to measure muscle activity, in addition to kinematics and kinetic measurements to evaluate gait function. Recent methods of using wavelet analysis to resolve EMG signals into frequency components provide information beyond the standard timing and amplitude measures [4,5]. The purpose of this study is to investigate the influence of BWSTT on co-activation between antagonistic muscles using wavelet analysis. METHODS We tested individuals (n=7, 42.0±14.3 yrs, 1.75±0.12 m, 79.7±22.6 kg) who had incurred an incomplete spinal cord injury due to a nonprogressive lesion above the thoracic level of T10, at least 12 months prior to recruiting to the study. Subjects were tested pre- and post- training where they walked on a treadmill while in body weight supported harness (the minimum support necessary) at their self-selected comfortable and maximum speeds. EMG data were collected bilaterally from the tibialis anterior (TA), medial gastrocnemius (MG), rectus femoris (RF), and biceps femoris (BF) (Delsys, Boston, MA). Force sensitive resistors were used to detect foot contact and toe-off. After the initial visit, subjects participated in a BWSTT program where they walked on a treadmill with a harness that supported a percentage of their body weight, while the Lokomat device applied velocitydependent moment against the hip and knee joints in the sagittal plane that was scaled to their maximum hip and knee flexor voluntary contraction (Lokomat: Hocoma, AG, Volketswil, Switzerland). All subjects underwent 45 minutes walking session, three days a week, for a total of 36 sessions. Age-, gender-, and height-matched healthy individuals (n=10, 40.0±12.8yrs, 1.75±0.09m, 74.5±13.6kg) were also tested at the same % body weight support and velocities as their matched SCI. Data for approximately 50 strides per subject from the preand post-training tests were analysed. EMG signals were resolved into time-frequency components using wavelet analysis [4,5]. The mean intensity at each wavelet domain (10 wavelets ranging from 19.29 Hz to 395.4 Hz) was calculated to generate an intensity spectrum. The correlation coefficient, r, was calculated for the EMG intensities between pairs of antagonistic muscles at each frequency band to generate a correlation spectrum. The intensity spectra and correlation spectra were analysed using principal component analysis [5]. We calculated an EMG-normalcy score as the resultant distance between the principal component loading scores for each antagonistic muscle pair and the mean principal component loading scores for the control subjects. ANOVA was performed to determine the effect of training, antagonistic muscle pair, speed, gait phase, and subject (random) on the mean correlation coefficient with Tukey post-hoc tests when appropriate. RESULTS AND DISCUSSION For the TA-MG muscles, the correlation coefficient of pre-training SCI was significantly greater than that of healthy controls (p = 0.02) and tended to be greater than that of post-training SCI (Fig. 1). BWSTT decreased the correlation between the MGTA muscles especially at the mid-range frequencies (Fig. 1). Correlation and coordination of the RF and BF also improved after BWSTT (Fig. 1), although not as substantially as the more proximal muscles. The EMG normalcy-score, calculated from the PCI and PCII loading scores (Fig. 2) were significantly less after training for MG-TA (p = 0.03), indicating that muscle function improved during training. Correlation R2 0.2 MG-TA 0 -0.2 RF-BF 0.2 0 -0.2 100 200 300 Frequency (Hz) 400 Figure 1: Correlation spectra of EMG intensities between muscles of healthy controls (blue), pretraining SCI (black), and post-training SCI (red) of the antagonistic pairs of muscles: 1) medial gastrocnemius (MG) and tibialis anterior (TA) and 2) rectus femoris (RF) and biceps femoris (BF). PCII Loading score MG-TA -1.0 1.5 - 1.0 - 0. 5 RF-BF Body weight supported treadmill training with resistance at the hip and knee joints appears to decrease the co-activation of the MG-TA muscles during gait. We are currently analyzing a set of similar data for a group of SCI individuals that underwent BWSTT with assistive forces applied to the hip and knee joints for comparison. Using wavelet analysis, we can calculate the intensity and correlation between antagonistic muscles over a range of frequencies. This EMG method considers the variation in frequency components in the EMG signal and is therefore more informative of muscle activity than traditional methods. This is useful in developing and assessing training interventions as specific frequencies can be targeted. Application of this EMG analysis to quantify and evaluate muscle dysfunction and coordination in SCI offers new insights into the fundamental mechanisms behind SCI impaired gait and into the effectiveness of rehabilitation treatments. CONCLUSIONS Body weight supported treadmill training with resistance at the hip and knee joints is an effective rehabilitation training that improves co-activation of antagonistic muscles. By employing advanced EMG processing techniques such as wavelet analysis and principal component analysis, information about the frequency content of the EMG signal can be obtained which is more informative that traditional methods and provides insight into muscle dysfunction and coordination in SCI individuals. -1.0 1.0 REFERENCES \ 1.0 -1.0 -2.0 PCI Loading score Figure 2: First and second principal component (PCI and PCII) loading scores of intensity and correlation spectra of healthy controls (blue), pretraining SCI (black), and post-training SCI (red) of the antagonistic pairs of muscles: 1) medial gastrocnemius (MG) and tibialis anterior (TA) and 2) rectus femoris (RF) and biceps femoris (BF). 1. Barbeu H, et al. Curr Opin Neurolo 14, 735-740, 2001. 2. Behrman AL, et al. Phys Ther 85, 1356-1371, 2005. 3. Lam T, et al. Neurorehabil Neural Repair 22, 438-446, 2008. 4. Von Tscharner V. J. Electromyo Kines 10, 433445, 2000. 5. Wakeling JM. Gait and Posture 25, 580-589, 2007.
© Copyright 2026 Paperzz