On the Improvement of the Trans-Tibial Amputee Musculoskeletal Model Scaling by Using a Virtual Marker Approach

Abstract

Modern marker-based motion capture techniques applied in clinical gait laboratories rely on the conventional gait analysis model as a standard tool for studying normal and pathological gait. It uses a Direct Kinematics (DK) method to calculate joint kinematics. However, current techniques are limited when used to analyse the biomechanics of amputee gait. This is because the residuum-socket interface is usually assumed to be rigid despite the considerable motion of the residual limb inside the socket which is known to occur. This relative motion is not captured in routine gait analysis due to the inability to place tracking markers inside the socket. In contrast to conventional models based on DK, musculoskeletal modelling software solve a global-optimisation for ‘pose estimation’ using an Inverse Kinematics (IK) method. Thus, while the residuum-socket interface should be treated explicitly as another joint in the system, this presents challenges when using traditional marker-based kinematic techniques. Previous studies of biomechanics in amputee gait have used various sensing technologies to directly record residuum-socket motion. However, such methods are complex and require costly or highly specialised equipment, these methods are therefore impractical for routine gait analyses (LaPrè et.al., Int J Numer Method Biomed Eng,34(4):29–36 (2018)). To address this limitation, we develop a whole-body musculoskeletal model including a residuum-socket joint with a least-squares global-optimisation method to study the kinematics of this interface, simultaneously with traditional joint angles during amputee walking. Experimental walking data from 4 subjects with unilateral trans-tibial amputations were used to demonstrate the utility of the virtual marker approach to scale a generic amputee musculoskeletal model during walking. The effects of conventional surface marker locations and the virtual marker approach to scale this generic musculoskeletal model were examined, and the resulting residuum-socket generalised motions for each model configuration were compared for all subjects.