Evaluation of kinematic-based algorithms for gait event detection in subjects with transtibial amputation walking on different terrains

Abstract

Several kinematic-based algorithms have shown accuracy for gait event detection in unimpaired and pathological gait. However, their validation in subjects with lower limb amputation is still limited. The aim of this paper was to evaluate three kinematic-based algorithms for detection of gait events in subjects with unilateral transtibial amputation walking on different terrains. The Coordinate-Based Algorithm, the Velocity-Based Algorithm and the High-Pass Filtered Algorithm (HPA) were implemented for detection of Initial Contact and Foot Off for the intact and prosthetic limbs of twelve subjects with unilateral transtibial amputation walking on level ground and up and down a slope. True Error (TE) was calculated as the time difference between the algorithms and the force platform in detecting the events. A post-processing of HPA was proposed to further understand its performance for amputee gait. The results showed that mean TE was smaller than 40 ms for all algorithms in all conditions (events, terrains and limbs). The performance of the algorithms was significantly affected (p<0.01) by the terrain and the limb under analysis. The post-processing of HPA showed that the signal processing used was not always valid for amputee gait. To conclude, if an accuracy of 40 ms is acceptable for the application, then all three algorithms can be used for event detection in amputee gait. Otherwise, considering the results of this study, an evaluation of the algorithms in pathological gait walking on the terrain of interest is recommended