muscle forces

IN VIVO ACHILLES TENDON FORCES DURING CYCLING DERIVED FROM 3D ULTRASOUND-BASED MEASURES OF TENDON STRAIN

Introduction and Objectives: Traditional motion analysis provides limited insight into muscle and tendon forces during movement. This study used B-mode ultrasound, in combination with measured joint angles and scaled musculoskeletal models, to provide subject-specific estimates of in vivo Achilles tendon (AT) force. Previous studies have used ultrasound images, tracked in 3D space, to estimate AT strains during walking, running, and jumping [1,2]. Our approach extends this work in one novel way. Specifically, we characterized AT stiffness on a subject-specific basis by recording subjects’ ankle moments and AT strains during a series of isometric tests. We then used these data to estimate AT force during movement from in vivo measurements of tendon strain. To demonstrate this approach, we report AT forces measured during cycling. Cycling offers a unique paradigm for studying AT mechanics. First, because the crank trajectory is constrained, joint angles and muscle-tendon unit (MTU) lengths of the gastrocnemius (MG, LG) and soleus (SOL) are also constrained. By varying crank load, subjects’ ankle moments can be altered without imposing changes in MTU lengths. For this study, 10 competitive cyclists were tested at 4 different crank loads while pedaling at 80 rpm. Based on published EMG recordings (e.g., [3]) and on in vivo tendon force buckle data from one subject [4], we hypothesized that the cyclists’ AT forces would increase systematically with crank load. Methods: We coupled B-mode ultrasound with motion capture, EMG, and pedal forces to estimate in vivo AT forces non-invasively during cycling and during a series of isometric ankle plantarflexion tests. Marker trajectories were tracked using an optical motion capture system. Joint angles and MTU lengths were calculated based on scaled musculoskeletal models [5] using OpenSim [6]. A 50 mm linear-array B-mode ultrasound probe was secured over the distal muscle-tendon junction (MTJ) of the MG and was tracked using rigid-body clusters of LEDs. AT lengths were calculated as the distance from a calcaneus marker to the 3D coordinates of the MG MTJ. Subject-specific AT force-strain curves were obtained from isometric tests using ultrasound to track the MTJ, markers to track both the ultrasound probe and the AT insertion, and a strain gauge to measure the net ankle torques generated by each of the subjects at ankle angles of -10° dorsiflexion, 0°, +10° plantarflexion, and +20° plantarflexion. AT strain during cycling was converted to AT force using each subject’s force-strain relation. Subject-specific tendon slack lengths were calculated as the mean tendon length at 310° over all pedal cycles, based on examination of the AT length changes and on published data showing that this position in the pedal cycle precedes tendon loading across multiple pedalling conditions [4]. Results: Peak AT forces during cycling ranged from 1320 to 2160 N ± 400 N (mean± SD) and increased systematically with load (p<0.001; Fig. 1A/B). At the highest load, the peak AT forces represented, on average, 50 to 70 % of the combined MG, LG, and SOL muscles’ maximum isometric force-generating capacity, as estimated from the muscles’ scaled volumes [7], the muscles’ scaled optimal fiber lengths [5], and a specific tension of 20-30 N/cm2. Peak AT forces occurred midway through the pedaling downstroke, at about 80°, which is consistent with the AT forces directly measured from one subject [4] and with patterns of EMG during cycling [3]. Peak AT strains during cycling were uncoupled from the MG MTU strains and ranged from 3 to 5 % across the different loads examined, measured at the MG MTJ. Conclusion: Our results are consistent with published data from a single subject in which AT force was measured using an implanted tendon buckle [8]; however, our results were obtained non-invasively using ultrasound and motion capture. These methods substantially augment the experimental tools available to study muscle-tendon dynamics during movement. References: [1]Lichtwark and Wilson, 2005, J Exp Biol, 208(24), 4715-4725. [2]Lichtwark et al., 2007, J Biomech, 40(1), 157-164. [3]Wakeling and Horn, 2009, J Neurophysiol, 101(2), 843-854. [4]Gregor et al., 1987, Int J Sports Med, 8(S1), S9-S14. [5]Arnold et al., 2010, Ann Biomed Eng, 38(2), 269-279. [6]Delp et al., 2007, IEEE Trans Bio Med Eng, 54(11), 1940-50. [7]Handsfield et al., 2014, J Biomech, 47(3),631-638. [8]Gregor et al. 1991, J Biomech, 24(5), 287-297
Listed In: Biomechanics, Sports Science


Muscle force prediction of the lower limb compared to surface EMG at different walking speeds in individual healthy subjects.

BACKROUND: Recent developments in modelling have made it easier to use muscle force predictions to augment clinical gait analysis and enhance clinical decision making. OpenSim claims to provide a straight forward, standardised pipeline (SimTrack) to predict muscle forces implemented in routine processing. This project aims to test SimTrack’s potential in the context of clinical gait analysis by developing a standardised protocol which compares predicted muscle forces with surface EMG at a range of walking speeds. METHODS: 10 healthy participants walked at 3 different speeds (comfortable, ±20%). Kinematics, kinetics and surface EMG of the lower limb were captured. Joint angles and ground reaction forces serve as inputs to predict muscle forces using computed muscle control (CMC) within SimTrack. Predicted muscle forces were compared with EMG to validate the model outputs. RESULTS: Agreement between force prediction and EMG varies between muscles. Some muscles show a general agreement and similar variation with walking speed, others show large unexpected differences between CMC outputs and observed EMG. DISCUSSION: These results suggest that this protocol is running in general. For most walking speeds, CMC muscle forces can be predicted within a timeframe appropriate for clinical purposes. However using the default settings, the model predictions do not agree with EMG measurements. Furthermore, during pilot testing of quicker walking speeds (up to +40%) CMC crashed due the chosen musculoskeletal model being too weak. These findings suggest the need of either different generic parameters or subject specific parameters to obtain valid results. Work is continuing to identify these.
Listed In: Biomechanics, Gait, Other