2019 Volume 57 Issue 6 Pages 215-223
Early detection of muscle strain injury caused by excessive stress is important to prevent progression of injury. However, slight injuries are difficult to detect with current measurement technologies. Our study relates the electrical impedance to the injury-induced change in the tissue composition, and proposes a portable and reproducible detection system of injured muscle. The proposed system scans the exciting and sensing positions using multiple electrodes placed on the body surface, and obtains multi-dimensional potential data related to the impedance distribution. Any injury is then discriminated in the data. To ensure a robust discrimination without measurements of injured muscle, the discrimination parameters were optimized using the data computed by a potential simulation model representing the muscle strain injury together with the possible noise and measurement data obtained from normal samples. In the experiment, we applied an AC voltage of 10kHz in a circuit and scanned the surface potential with 16 electrodes. The injury-discrimination performance was then assessed using a phantom representing the electrical characteristics of the leg muscle. In phantoms representing normal tissues and three-stage injured samples, wefirst investigated the noise level of the electrode position that maximized the discrimination rate. The maximum discrimination rate of the proposed system was 88.3%for a Gaussian noise with a 0.24mm standard deviation. Applying the noise settings that maximized the discrimination rate, we then evaluated the discrimination rate of injured tissue at locations not included in the learning data. The system delivered similar discrimination performance at different injured locations compared with the small injured sample. Therefore, the proposed method is expected to detect slight muscle strain injuries using a simple hardware setup.
