Verification of Markerless Gait Analysis: Multi-Camera and Single-Camera Approaches in Comparison to Marker-Based Gait Analysis

Abstract

Background and Objectives: This study aimed to verify the validity of markerless gait analysis using both single-camera markerless system (S-ML) and multi-camera markerless system (M-ML) approaches by comparing them with a gold-standard marker-based system (MB). Materials and Methods: Sixteen healthy adults walked at their gait speed, and their gaits were simultaneously analyzed using three systems: S-ML, M-ML, and MB. Intraclass correlation coefficients were used to assess the reliability of the spatiotemporal parameters, and the root mean squared error (RMSE) was calculated to quantify the kinematic differences relative to the MB systems. Results: Both S-ML and M-ML demonstrated good-to-excellent reliability in spatiotemporal parameters, including step length, stance time, and gait speed, whereas stride length and swing time measured by S-ML revealed moderate reliability. In terms of joint kinematics, S-ML demonstrated a performance comparable to that of M-ML, particularly at the hip and knee in the sagittal plane. For certain parameters, such as knee abduction/adduction in the frontal plane, the S-ML demonstrated lower RMSE values than M-ML. In contrast, the ankle joint angles estimated using S-ML exhibited reduced agreement. Conclusions: In conclusion, markerless gait analysis can serve as an alternative to conventional gait analysis. However, certain parameters need to be improved.