No abstract

Inertial sensing and computer vision are promising alternatives to traditional optical motion tracking, but until now these data sources have been explored either in isolation or fused via unconstrained optimization, which may not take full advantage of their complementary strengths. By adding physiological plausibility and dynamical robustness to a proposed solution, biomechanical modeling may enable better fusion than unconstrained optimization. To test this hypothesis, we fused video and inertial sensing data via dynamic optimization with a nine degree-of-freedom model and investigated when this approach outperforms video-only, inertial-sensing-only, and unconstrained-fusion methods. We used both experimental and synthetic data that mimicked different ranges of video and inertial measurement unit (IMU) data noise. Fusion with a dynamically constrained model improved estimation of lower-extremity kinematics by a mean ± std root-mean-square error of 6.0° ± 1.2° over the video-only approach and estimation of joint centers by 4.5 ± 2.8 cm over the IMU-only approach. It consistently outperformed single-modality approaches across different noise profiles. When the quality of video data was high and that of inertial data was low, dynamically constrained fusion improved joint kinematics by 3.7° ± 1.2° and joint centers by 1.9 ± 0.5 cm over unconstrained fusion, while unconstrained fusion was advantageous by 3.0° ± 1.4° and 1.2 ± 0.7 cm in the opposite scenario. These findings indicate that complementary modalities and techniques can improve motion tracking by clinically meaningful margins and that data quality and computational complexity must be considered when selecting the most appropriate method for a particular application.

Inexpensive wearable sensors are expected to transform both research and clinical practice by monitoring patient movement outside of the laboratory and helping personalize the treatment of mobility impairments [1]. To meet these expectations, wearable sensors need to be benchmarked against clinical standards, be robust to placement errors by non-experts, and provide reliable data over long periods of time. Inertial sensing remains the only wearable technology that has been comprehensively characterized and benchmarked against gold-standard biomechanical measurements, but it is sensitive to both drift and placement error [2] and does not provide estimations of muscle activity, which are relevant to numerous mobility impairments. Here we characterize capacitive touch sensing [3] as a gait rehabilitation monitoring technology for the first time, finding that it captures clinically relevant biomechanical measures with the fidelity of laboratory tools. We also show that a circumferential lower-limb capacitive sensing sleeve is more effective than electromyography and musculoskeletal simulations at detecting therapeutically relevant gait modifications used to prevent osteoarthritis progression. Finally, we show that our capacitive sensing approach is robust to placement errors and measurement drift over a 6-hour trial, both of which are insignificant to tracking adherence to therapeutic gait prescriptions. Our results indicate that capacitive sensing wearables could make rehabilitation monitoring outside laboratory environments more feasible and could be used synergistically with other emerging wearable technologies to provide real-time feedback to patients during daily life [4]. We expect this foundational study of capacitive sensing for rehabilitation monitoring to be translatable to other parts of the body and applicable to a wide range of mobility-related pathologies and emerging human-in-the-loop wearable health technologies [5–7].