A Model-Based Markerless Protocol for Clinical Gait Analysis Based on a Single RGB-Depth Camera: Concurrent Validation on Patients With Cerebral Palsy

• Published in: IEEE access Vol. 11; pp. 144377 - 144393
• Main Authors: Balta, Diletta, Figari, Giulio, Paolini, Gabriele, Pantzar-Castilla, Evelina, Riad, Jacques, Croce, Ugo Della, Cereatti, Andrea
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 2023; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Cameras; Cerebral Palsy; Gait; Gait recognition; Infrared analysis; Infrared cameras; Joints (anatomy); Kinematics; Knee; Legged locomotion; Markerless gait analysis; Motion capture; Musculoskeletal system; Orthopaedics; Ortopedi; Patients; Reliability; RGB-depth camera; sagittal lower-limb joint kinematics; Spatial temporal resolution; spatial-temporal parameters
• ISSN: 2169-3536; 2169-3536
• DOI: 10.1109/ACCESS.2023.3340622

Clinical gait analysis is a diagnostic tool often used for identifying and quantifying gait alterations in cerebral palsy (CP) patients. To date, 3D clinical gait analysis protocols based on motion capture systems featuring multiple infrared cameras and retroreflective markers to be attached to the subject's skin are considered the gold standard. However, the need for fully dedicated personnel and space in addition to the inconvenient requirement of multiple markers attached on the patient's body limit their use in the clinical practice. To shorten the time necessary to setup the patient and to limit his/her discomfort motion tracking performed using markerless technologies may offer a promising alternative to marker-based motion capture. This study aims at proposing and validating on 18 CP patients, an original markerless clinical gait analysis protocol based on a single RGB-D camera. Accuracy and reliability of the spatial-temporal parameters and sagittal lower limb joint kinematics were assessed based on a 3D marker-based clinical gait analysis protocol. The smallest percent mean absolute errors were obtained for stride duration (2%), followed by the step and stride length (2.2% and 2.5%, respectively) and by gait speed (3.1%). The average angular offset values between the two protocols were 8° for the ankle, 6° for the knee and 7° for the hip joint. The smallest root mean square error values were found for the knee joint kinematics (3.2°), followed by the hip (3.5°) and the ankle (4.5°). Both protocols showed a good-to-excellent reliability. Thus, this study demonstrated the technical validity of a markerless single-camera protocol for clinical gait analysis in CP population. The dataset containing markerless data from 10 CP patients along with the MATLAB codes have been made available.