Wearable Loop Sensor for Bilateral Knee Flexion Monitoring

Abstract

We have previously reported wearable loop sensors that can accurately monitor knee flexion with unique merits over the state of the art. However, validation to date has been limited to single-leg configurations, discrete flexion angles, and in vitro (phantom-based) experiments. In this work, we take a major step forward to explore the bilateral monitoring of knee flexion angles, in a continuous manner, in vivo. The manuscript provides the theoretical framework of bilateral sensor operation and reports a detailed error analysis that has not been previously reported for wearable loop sensors. This includes the flatness of calibration curves that limits resolution at small angles (such as during walking) as well as the presence of motional electromotive force (EMF) noise at high angular velocities (such as during running). A novel fabrication method for flexible and mechanically robust loops is also introduced. Electromagnetic simulations and phantom-based experimental studies optimize the setup and evaluate feasibility. Proof-of-concept in vivo validation is then conducted for a human subject performing three activities (walking, brisk walking, and running), each lasting 30 s and repeated three times. The results demonstrate a promising root mean square error (RMSE) of less than 3° in most cases.

Keywords: Faraday’s law; bioelectromagnetic; e-textiles; electromotive force; joint flexion; wearable sensor.

Figure 1

Proposed wearable loop sensors for monitoring bilateral knee flexion angles by capturing θ_f1 and θ_f2 from two receiving loops, Rx1 and Rx2, respectively.

Figure 2

Simulation setup for the wearable loop sensor placed on cylindrical limb model for bilateral knee flexion angle monitoring.

Figure 3

Process flow for manufacturing the polymer-embedded e-thread-based loops.

Figure 4

(a) Polymer-embedded e-thread-based loop sensor, (b) phantom-based experimental setup for bilateral knee flexion angle monitoring.

Figure 5

Experimental setup used for bilateral flexion angle monitoring on a human subject.

Figure 6

Simulation and phantom experimental results: (a) flexion curves of leg 1 (|S₂₁|vs. θ_f1) for different θ_f2, (b) flexion curves of leg 2 (|S₄₃|vs. θ_f2) for different θ_f1.

Figure 7

Calibration results from slow flexion and extension: (a) left leg, (b) right leg.

Figure 8

Representative real-time comparisons between the flexion angles captured by the sensor and MoCap system for 20 s during (a) walking, (b) brisk walking, and (c) running for the left leg; and (d) walking, (e) brisk walking, and (f) running for the right leg.

Figure 9

Representative interval RMSE and curve fitting result for 20 s during (a) walking, (b) brisk walking, and (c) running for the left leg; and (d) walking, (e) brisk walking and (f) running for the right leg.