TL;DR
This paper presents a novel, automated MATLAB-based system for processing motion capture data of human knee joint kinematics. It addresses the limitations of manual, time-consuming commercial software by introducing a "Patch Program" to effectively handle missing or noisy data, enabling efficient batch processing for accurate kinematic analysis.
What Problem Does This Paper Address?
Measuring human knee joint kinematics is crucial for understanding biomechanics, diagnosing conditions, and guiding rehabilitation after procedures like Total Knee Arthroplasty (TKA). Traditional methods often rely on commercial software that requires extensive manual processing for each subject and movement, making it time-consuming and labor-intensive, especially for large datasets. Furthermore, motion capture data can be prone to missing or noisy signals due to marker occlusion or sensor issues, which can significantly impact the accuracy of kinematic analysis. This paper aims to overcome these challenges by developing an automated and robust data processing pipeline.
Methods
The core of this work is a custom MATLAB program designed for efficient and automated processing of motion capture data. Key methodological components include:
- Data Acquisition: Motion capture data was collected from 29 healthy volunteers performing squatting and stair-climbing movements using the Optotrak Certus system.
- Local Coordinate System Establishment: Anatomical points on the femur and tibia were identified and used to establish local coordinate systems for each bone. This is a standard approach in biomechanics for defining joint angles.
- "Patch Program" for Data Repair: A novel "Patch Program" was developed to address missing or noisy data. This program uses cubic spline interpolation to reasonably compensate for gaps and anomalies, ensuring more complete motion data. It can repair data when one marker is missing but discards data if two or more markers are missing.
- Kinematic Parameter Calculation: The program calculates knee joint kinematics, specifically flexion/extension, abduction/adduction, and internal/external rotation, by applying coordinate transformations between the established local coordinate systems of the femur and tibia.
- Universal Data Processing Program: A general-purpose program was also developed to analyze the relative movement of various human body components, which can be modified for specific analyses.
Datasets
The study utilized motion capture data from 29 healthy volunteers performing squatting and stair-climbing movements. The data collected included 3D coordinate data of markers placed on the thighs and calves, as well as anatomical feature points. No new benchmark datasets were introduced; rather, existing motion capture data was processed.
Key Insights and Ideas
- Automation is Key: The development of an automated MATLAB program significantly reduces the time and effort required for motion capture data processing compared to manual methods in commercial software.
- Robustness through Data Repair: The "Patch Program" is a critical innovation, enabling the analysis of datasets that would otherwise be discarded due to missing or noisy marker data. This improves the completeness and reliability of the kinematic analysis.
- Versatility of the Approach: The developed programs are designed to be adaptable, allowing for the analysis of various body segments and specific kinematic analyses, making them broadly applicable in biomechanics research.
- Clinical Relevance: The ability to accurately measure knee kinematics has direct implications for assessing human body function, guiding rehabilitation strategies, and evaluating surgical outcomes.
Key Results
The developed MATLAB program was successfully applied to process motion capture data from 29 healthy volunteers during squatting and stair-climbing. The program generated kinematic data, including flexion/extension, abduction/adduction, and internal/external rotation of the knee joint. For example, the maximum internal/external rotation during squatting was found to be up to 30.5 degrees, and 14 degrees during stair climbing. The results were compared with existing literature and found to be consistent, validating the reliability of the developed methods. The "Patch Program" demonstrated its effectiveness in interpolating missing data points, resulting in smoother and more complete motion curves.
Limitations and Future Work
- Marker Data Loss: The "Patch Program" can only repair data when one marker is missing; if two or more markers are missing, the data must be discarded. This limits the robustness of the repair for severe data loss.
- MATLAB Processing Speed: While automated, the paper notes that MATLAB's pre-processing speed is not as fast as some commercial software.
- Anatomical Feature Point Determination: The accuracy of establishing the coordinate systems for the femur and tibia relies on the precise determination of anatomical feature points, which can be influenced by the position of these points.
- Future Work: The authors suggest increasing the sample size and continuously improving the program to obtain even more accurate kinematics data. They also highlight the potential for further research into error analysis of marker points on skin surface to reduce measurement errors. The developed technologies are envisioned for use in judging abnormal body functions, aiding rehabilitation, and evaluating surgical effectiveness.
Summary updated