Physiologically corrected coupled motion during gait analysis using a model-based approach.

Gait analysis is used in daily clinics for patients' evaluation and follow-up. Stereophotogrammetric devices are the most used tool to perform these analyses. Although these devices are accurate results must be analyzed carefully due to relatively poor reproducibility. One of the major issues is related to skin displacement artifacts. Motion representation is recognized reliable for the main plane of motion displacement, but secondary motions, or combined, are less reliable because of the above artifacts. Model-based approach (MBA) combining accurate joint kinematics and motion data was previously developed based on a double-step registration method. This study presents an extensive validation of this MBA method by comparing results with a conventional motion representation model. Thirty five healthy subjects participated to this study. Gait motion data were obtained from a stereophotogrammetric system. Plug-in Gait model (PiG) and MBA were applied to raw data, results were then compared. Range-of-motion, were computed for pelvis, hip, knee and ankle joints. Differences between PiG and MBA were then computed. Paired-sample t-tests were used to compare both methods. Normalized root-mean square errors were also computed. Shapes of the curves were compared using coefficient of multiple correlations. The MBA and PiG approaches shows similar results for the main plane of motion displacement but statistically significative discrepancies appear for the combined motions. MBA appear to be usable in applications (such as musculoskeletal modeling) requesting better approximations of the joints-of-interest thanks to the integration of validated joint mechanisms.