Reference in-vitro dataset for inertial-sensor-to-bone alignment applied to the tibiofemoral joint

Ive Weygers, Manon Kok, Thomas Seel, Darshan Shah, Orçun Taylan, Lennart Scheys, Hans Hallez, Kurt ClaeysScientific 2021 · DOI

Abstract

Skin-attached inertial sensors are increasingly used for kinematic analysis. However, their ability to measure outside-lab can only be exploited after correctly aligning the sensor axes with the underlying anatomical axes. Emerging model-based inertial-sensor-to-bone alignment methods relate inertial measurements with a model of the joint to overcome calibration movements and sensor placement assumptions. It is unclear how good such alignment methods can identify the anatomical axes. Any misalignment results in kinematic cross-talk errors, which makes model validation and the interpretation of the resulting kinematics measurements challenging. This study provides an anatomically correct ground-truth reference dataset from dynamic motions on a cadaver. In contrast with existing references, this enables a true model evaluation that overcomes influences from soft-tissue artifacts, orientation and manual palpation errors. This dataset comprises extensive dynamic movements that are recorded with multimodal measurements including trajectories of optical and virtual (via computed tomography) anatomical markers, reference kinematics, inertial measurements, transformation matrices and visualization tools. The dataset can be used either as a ground-truth reference or to advance research in inertial-sensor-to-bone-alignment.

Introduction

In biomechanics, skin-attached inertial measurement units (IMUs) are essential for tracking movement outside of laboratory settings. However, IMUs measure their own orientation rather than the orientation of the underlying bone. To derive clinically relevant data, such as knee flexion angles, researchers must align sensor axes with anatomical axes. Current validation methods are often compromised by Soft Tissue Artifacts (STA), where skin movement relative to the bone introduces significant noise. This study presents a "gold standard" reference dataset that bypasses the "skin problem" by rigidly attaching sensors directly to the bone.

Methods and Experimental Setup

The researchers employed a multimodal approach using a fresh-frozen cadaveric lower limb. To ensure a rigid connection, bone-pins were used to attach both optical marker clusters and wireless IMUs directly to the femur and tibia.

Experimental setup overview: (a) bone-pinned optical markers, (b) 3D bone model, (c) inertial sensor attachment, and (d) coordinate system definitions.

A volumetric CT scan was performed with the pins in place, allowing for the identification of precise osseous anatomical landmarks and the creation of "virtual" markers within a 3D bone model. The limb was then moved through 53 dynamic trials by physiotherapists, varying in plane, speed, and range of motion (60° to 110° flexion).

Anatomical Alignment and Dataset

The dataset provides the necessary transformation matrices between the IMU internal axes, optical marker clusters, and CT-defined anatomical axes. By registering CT scans to the optical capture system, the authors can track internal anatomical points, such as the knee joint center, as virtual markers.

Detailed view of the knee joint model showing osseous anatomical landmarks (e.g., FLE, FME) and the associated coordinate systems.

The comprehensive dataset includes:

  • Raw Data: 100Hz IMU measurements and optical marker trajectories.
  • CT Data: 3D surface bone models (.stl) and 19 anatomical landmark coordinates.
  • Reference Kinematics: Ground truth joint angles calculated via Grood & Suntay definitions.

Key Results and Insights

Technical validation confirmed the high precision of the dataset. The temporal synchronization mismatch between optical and inertial systems was minimal (0.01 to 0.02 seconds). Furthermore, the misalignment between IMUs and bone-pinned optical clusters remained extremely low (0.98° ± 0.55° for the femur), confirming the rigidity of the setup throughout the 321,073 recorded samples. The data successfully captured complex biomechanical phenomena, such as the "screw-home" mechanism of the knee, with internal rotations reaching 22.86° during flexion.

Conclusion

This dataset eliminates the "skin gap" error, providing a platform where the only source of error is the alignment algorithm or the sensor itself. While limited by the use of a single cadaveric specimen, it serves as a critical resource for validating "plug-and-play" or model-based alignment methods that aim to function without specific user calibration movements.

Summary updated

Abstract

Skin-attached inertial sensors are increasingly used for kinematic analysis. However, their ability to measure outside-lab can only be exploited after correctly aligning the sensor axes with the underlying anatomical axes. Emerging model-based inertial-sensor-to-bone alignment methods relate inertial measurements with a model of the joint to overcome calibration movements and sensor placement assumptions. It is unclear how good such alignment methods can identify the anatomical axes. Any misalignment results in kinematic cross-talk errors, which makes model validation and the interpretation of the resulting kinematics measurements challenging. This study provides an anatomically correct ground-truth reference dataset from dynamic motions on a cadaver. In contrast with existing references, this enables a true model evaluation that overcomes influences from soft-tissue artifacts, orientation and manual palpation errors. This dataset comprises extensive dynamic movements that are recorded with multimodal measurements including trajectories of optical and virtual (via computed tomography) anatomical markers, reference kinematics, inertial measurements, transformation matrices and visualization tools. The dataset can be used either as a ground-truth reference or to advance research in inertial-sensor-to-bone-alignment.
BibTeX
@inproceedings{weygers2021reference,
  author = {Ive Weygers and Manon Kok and Thomas Seel and Darshan Shah and Orçun Taylan and Lennart Scheys and Hans Hallez and Kurt Claeys},
  title = {{Reference in-vitro dataset for inertial-sensor-to-bone alignment applied to the tibiofemoral joint}},
  booktitle = {Scientific},
  year = {2021},
  doi = {10.1038/s41597-021-00995-8},
}
Figure