Evaluation of methods based on conventional videography for detection of gait events

Abstract

The detection of gait events plays an important role in gait analysis. The initial foot contact (IC) and the end of foot contact or foot off (FO) represent the start of stance and swing, respectively. They are commonly used to compute parameters that offer information for the characterization of gait. To determine these events, kinematic methods that implement optical systems, including conventional video cameras, have been suggested. The key objective of this study was to evaluate the performance of two event detection methods, one visual (VM) and one semiautomatic (SAM), which use conventional video recordings, for the visual detection of IC and FO, during level ground and stair walking. The methods were tested in 10 healthy subjects, wearing confortable shoes and walking at a self-selected speed. The accuracy of the proposed methods was compared with force plate data (gold standard). Furthermore, the VM and the SAM were compared with an alternative kinematic method: The Velocity-Algorithm (VA), previously described in the literature. The results of the level ground test showed that the VM and the SAM detected IC and FO with an absolute mean difference (AMD) lower than 30 ms. For the stair climbing test, it was found that the VM and the SAM detected IC and FO with an AMD lower than 55 ms. In general terms, the VM and the SAM produced differences within the ranges reported in the literature for both terrains. These results provide preliminary evidence supporting the use of both methods for gait event detection on level walking as well as stair climbing.