2025 ISAKOS Biennial Congress ePoster
Verification of the measurement accuracy of marker-less motion analysis in evaluating single-leg squat movements after anterior cruciate ligament (ACL) reconstruction
Ayaka Tanaka, MD, Suita, Osaka JAPAN
Tomoki Ohori, MD, PhD, Suita, Osaka, Asia JAPAN
Akira Tsujii, MD, PhD, Suita, Osaka JAPAN
Shuto Yamashita, MD, Suita, Osaka JAPAN
Toshitaka Tsunematsu, MD, Suita, Osaka JAPAN
Syunya Otani, MD, PhD, Tokyo JAPAN
Seira Sato, MD, PhD, Suita, Osaka JAPAN
Seiji Okada, MD, PhD, Prof., Suita, Osaka JAPAN
Osaka University Graduate School of Medicine, Suita, Osaka, JAPAN
FDA Status Not Applicable
Summary
This study found that while a markerless motion analysis app generally maintains accuracy within 5° of a marker-based system, it shows variability, especially for the left knee and hip under fast conditions, highlighting the need for careful camera placement and further research to validate its broader applicability.
ePosters will be available shortly before Congress
Abstract
Introduction
Motion analysis is crucial for assessing rehabilitation progress and knee function after anterior cruciate ligament (ACL) reconstruction. Marker-based systems are considered the gold standard for capturing accurate motion data, but they are costly and time-consuming to set up. While markerless systems are more accessible, their accuracy raises concerns. This study aimed to determine whether a markerless motion analysis app could achieve accuracy comparable to established marker-based systems.
Materials And Methods
The study involved a single patient who had undergone ACL reconstruction with a bone-patellar tendon-bone (BTB) graft. The task was a single-leg squat, performed six times on each leg under both slow and fast conditions. Data were collected using two systems. The first system was a marker-based optical system (OptiTrack, 360Hz) that recorded the 3D positional coordinates of reflective markers attached to the patient's body. Joint angles for ankle dorsiflexion, knee flexion, and hip flexion were calculated using Euler angles, providing a reference waveform. The second system was a markerless motion analysis app (SPLYZA Motion), which analyzed video footage (47.59Hz) captured from the patient's right side. The same joint angles were calculated, and the root mean square error (RMSE) between the waveforms from both systems was computed.
Results
The markerless app showed variability in accuracy compared to the marker-based system. Under slow conditions, RMSE values were 1.49° for the left ankle, 1.32° for the right ankle, 2.15° for the left knee, 3.36° for the right knee, 4.24° for the left hip, and 2.19° for the right hip. Under fast conditions, the RMSE values slightly increased, showing 3.09° for the left ankle, 1.31° for the right ankle, 5.33° for the left knee, 3.41° for the right knee, 7.23° for the left hip, and 2.37° for the right hip. These results indicate that the markerless app generally maintains accuracy within 5° of the marker-based system. However, a decrease in accuracy was noted for the left knee and hip (opposite side of the video analysis), particularly under fast conditions.
Discussion
This study suggests that markerless motion analysis systems can provide sufficient accuracy for many applications, with errors typically within 5° compared to marker-based systems. However, the increased RMSE observed for the left knee and hip under fast conditions highlights a limitation where data accuracy may be affected, particularly when the camera's field of view is obstructed or angles are suboptimal. To ensure reliable markerless motion analysis, careful placement of cameras, minimizing obstructions, and capturing clear views of the target joints are crucial. Despite these limitations, markerless systems serve as a valuable tool, especially when traditional systems are impractical. Further research with larger sample sizes and diverse movements will help validate these findings and explore the broader applicability of markerless systems in motion analysis.