TL;DR
This systematic review critically examines the methodological requirements for using inertial measurement units (IMUs) to accurately estimate lower limb joint kinematics, particularly for clinical applications. It highlights that while IMUs offer a promising, portable alternative to lab-based motion capture, their current implementation faces challenges related to sensor limitations, data processing, and validation, necessitating careful interpretation of results and further research into robust, subject-specific models and alternative validation methods.
What Problem Does This Paper Address?
Accurately measuring human movement, specifically lower limb joint kinematics, is crucial for understanding functional movements, assessing patient progress in rehabilitation, and diagnosing conditions. Traditional methods like 3D optical motion capture systems, while considered the gold standard, are expensive, confined to laboratory settings, and sensitive to marker placement. Inertial Measurement Units (IMUs) offer a compelling alternative due to their portability and potential for in-field, long-term monitoring. However, the widespread adoption of IMU-based kinematic estimation in clinical practice is hindered by a lack of clear understanding of the methodological requirements, including how to overcome inherent sensor limitations (noise, bias, integration drift) and ensure reliable, clinically relevant measurements. This paper aims to bridge this gap by systematically reviewing existing literature to evaluate these requirements and suggest future research directions.
Methods
This study employed a systematic literature review following the PRISMA guidelines. They searched three major databases (PubMed, Web of Science, EMBASE) for peer-reviewed articles and conference papers published in English. Inclusion criteria focused on studies that:
- Described the methodology for obtaining lower limb kinematic variables using IMUs.
- Validated their methods against a reference system (e.g., optical motion capture).
- Included methodologies applicable to outside-laboratory use.
Data extraction covered study characteristics (participants, activity, duration, joints, sensor placement, modalities), signal processing characteristics (pre-processing, prior information, assumptions, information fusion, drift compensation, initialization, calibration), and study results (accuracy against a reference system).
Datasets
Not discussed in the paper. The paper is a systematic review of existing literature and does not introduce new datasets or benchmarks.
Key Insights and Ideas
- Application Dependency is Key: Lower limb joint kinematics estimation methods are highly dependent on the specific application (e.g., gait, squat, running) and the desired output (e.g., sagittal plane motion vs. full 3D).
- Sensor Limitations Drive Methodology: IMU limitations like noise, bias, and integration drift necessitate the use of prior information and assumptions to compensate and achieve usable kinematic data.
- Prior Information is Crucial but Needs Careful Interpretation: Biomechanical knowledge (e.g., joint range of motion, segment lengths) and assumptions (e.g., segment rigidity, periodicity) are essential for IMU-based methods. However, these can be violated in real-world scenarios or specific patient populations, requiring careful consideration before clinical decisions are made.
- Validation Challenges: While optical motion capture is the common "golden standard," it has its own limitations (marker placement, soft tissue artifacts), particularly for estimating secondary joint kinematics (frontal and transversal planes). This suggests a need for alternative validation methods.
- Trend Towards Calibration-Free and Subject-Specific Models: There's a growing interest in methods that reduce reliance on precise calibration movements (which can be difficult for patients) and in developing subject-specific biomechanical models for improved accuracy.
Key Results
- Thirty-one studies were included in the review.
- Gait was the most commonly evaluated activity, followed by functional movements like sit-to-stand and squats.
- Sagittal knee joint movements were most extensively studied, with fewer studies investigating full 3D kinematics of the entire lower quadrant.
- Accuracy varies significantly, with sagittal plane RMSEs ranging from 1.3 to 11.22 degrees, and frontal/transversal RMSEs from 1 to 6.7 degrees. This variability underscores the application-dependent nature of the methods.
- Most studies used a combination of gyroscopes and accelerometers, with some incorporating magnetometers or other sensors.
- Common pre-processing steps include filtering (Butterworth filter being frequent) and calibration, while assumptions often involve segment rigidity, periodicity, and symmetry.
- Strap-down integration and complementary filters are common information fusion techniques, with Extended Kalman Filters (EKF) also being utilized.
Limitations and Future Work
- Limitations:
- The review highlights that current IMU-based kinematic estimates need careful interpretation due to the reliance on assumptions and prior information, which may not always hold true.
- The accuracy of IMU-based methods, especially for frontal and transversal plane movements, is still a concern, partly due to the limitations of optical motion capture as a reference.
- Many studies did not investigate long-term measurements (beyond 30 seconds) or disturbed movement patterns, limiting their applicability in real-world clinical scenarios.
- Future Work:
- Standardized Reporting: Adherence to ISB recommendations for reporting joint kinematics is crucial for clinical interpretability.
- Advanced Biomechanical Modeling: Developing more complex, subject-specific tibiofemoral joint models using IMUs could provide deeper insights into secondary joint kinematics, especially for ligament injuries.
- Improved Validation Methods: Exploring alternative validation techniques beyond optical motion capture (e.g., biplanar radiographic imaging) is needed to accurately assess frontal and transversal plane movements and validate IMU-based methods.
- Longer-Term and In-Field Measurements: Research should focus on increasing measurement durations and developing magnetometer-free methods for robust in-field applications.
- Disturbed Movement Patterns: Investigating IMU-based kinematic estimation for patient populations with movement disorders and developing calibration-free methods are critical for clinical translation.
- Combined Efforts: Collaboration between engineers and clinical experts is essential for developing practical, patient-specific IMU solutions.
Summary updated