| Abstrakt: |
Current technologies are unable to identify the simultaneous contraction of deep and superficial muscles when the subject is in motion. In this contribution, we propose a method to identify the co-contraction patterns of four muscles of the abdominal wall, namely, the rectus abdominis (RA), obliquus externus (OE), obliquus internus (OI), and transversus abdominis (TRA), with only two pairs of surface electrodes. Surface electromyography (EMG) signals are acquired from two bipolar leads placed on the lower abdomen of the volunteers. Following the extraction of features, a principal component analysis (PCA) is conducted to optimize the data representation, and a random forest classifier is employed to classify the co-contraction patterns. Our method achieves up to 86.30% accuracy, which demonstrates the possibility of identifying 14 co-contraction patterns of four different muscles of the abdominal wall, either surface or deep muscles. This method does not need to be adapted to a new patient, which is better suited to the physiotherapist’s practice. Moreover, it opens a field of research regarding the role of deep muscles in motion either during exercises or daily life tasks as well as in pathologies with complex etiology. EMG research will benefit from this method, which provides a better understanding of muscle co-contractions, but also reduces the number of sensors needed to acquire relevant information, while remaining noninvasive. The clinical interest lies in the improvement of the physiotherapeutic management. Indeed, a better knowledge of the patient’s co-contraction patterns makes it easier to adapt the physiotherapist’s treatment plan to the patient’s needs. |
