# Sensor-to-Bone Calibration with the Fusion of IMU and Bi-Plane X-rays

## TL;DR

This paper introduces a new method to calibrate IMUs using bi-plane X-rays and a fusion box, improving accuracy in measuring bone orientation for patients with locomotor issues.

## Contribution

A novel sensor-to-bone calibration method using bi-plane X-rays and a fusion box is proposed for more accurate kinematic measurements in pathological populations.

## Key findings

- The fusion of bi-plane X-rays and IMUs is feasible for measuring anatomical axes orientation.
- Sensor-to-bone calibration showed higher validity for pelvic tilt compared to conventional methods.
- Similar validity was observed for other degrees of freedom between the two calibration methods.

## Abstract

Inertial measurement units (IMUs) need sensor-to-segment calibration to measure human kinematics. Multiple methods exist, but, when assessing populations with locomotor function pathologies, multiple limitations arise, including holding postures (limited by joint pain and stiffness), performing specific tasks (limited by lack of selectivity) or hypothesis on limb alignment (limited by bone deformity and joint stiffness). We propose a sensor-to-bone calibration based on bi-plane X-rays and a specifically designed fusion box to measure IMU orientation with respect to underlying bones. Eight patients undergoing total hip arthroplasty with bi-plane X-rays in their clinical pathway participated in the study. Patients underwent bi-plane X-rays with fusion box and skin markers followed by a gait analysis with IMUs and a marker-based method. The validity of the pelvis, thigh and hip kinematics measured with a conventional sensor-to-segment calibration and with the sensor-to-bone calibration were compared. Results showed (1) the feasibility of the fusion of bi-plane X-rays and IMUs in measuring the orientation of anatomical axes, and (2) higher validity of the sensor-to-bone calibration for the pelvic tilt and similar validity for other degrees of freedom. The main strength of this novel calibration is to remove conventional hypotheses on joint and segment orientations that are frequently violated in pathological populations.