Currently, sEMG-based pattern recognition is a crucial and promising control method for prosthetic limbs. A 1D convolutional recurrent neural network classification model for recognizing online finger and wrist movements in real time was proposed to address the issue that the classification recognition rate and time delay cannot be considered simultaneously. This model could effectively combine the advantages of the convolutional neural network and recurrent neural network. Offline experiments were used to verify the recognition performance of 20 movements, and a comparative analysis was conducted with CNN and LSTM classification models. Online experiments via the self-developed sEMG signal pattern recognition system were established to examine real-time recognition performance and time delay. Experiment results demonstrated that the average recognition accuracy of the 1D-CNN-RNN classification model achieved 98.96% in offline recognition, which is significantly higher than that of the CNN and LSTM (85.43% and 96.88%, respectively, p < 0.01). In the online experiments, the average accuracy of the real-time recognition of the 1D-CNN-RNN reaches 91% ± 5%, and the average delay reaches 153 ms. The proposed 1D-CNN-RNN classification model illustrates higher performances in real-time recognition accuracy and shorter time delay with no obvious sense of delay in the human body, which is expected to be an efficient control for dexterous prostheses.

No abstract

The root mean square (RMS) of the surface electromyography (sEMG) signal can respond to neuromuscular function, which displays a positive correlation with muscle force and muscle tension under positive and passive conditions, respectively. The purpose of this study was to investigate the changes in muscle force and tension after multilevel surgical treatments, functional selective posterior rhizotomy (FSPR) and tibial anterior muscle transfer surgery, and evaluate their clinical effect in children with spastic cerebral palsy (SCP) during walking. Children with diplegia (n = 13) and hemiplegia (n = 3) with ages from 4 to 18 years participated in this study. They were requested to walk barefoot at a self-selected speed on a 15-m-long lane. The patient's joints' range of motion (ROM) and sEMG signal of six major muscles were assessed before and after the multilevel surgeries. The gait cycle was divided into seven phases, and muscle activation state can be divided into positive and passive conditions during gait cycle. For each phase, the RMS of the sEMG signal amplitude was calculated and also normalized by a linear envelope (10-ms running RMS window). The muscle tension of the gastrocnemius decreased significantly during the loading response, initial swing, and terminal swing (p &lt; 0.05), which helped the knee joint to get the maximum extension when the heel is on the ground and made the heel land smoothly. The muscle force of the gastrocnemius increased significantly (p &lt; 0.05) during the mid-stance, terminal stance, and pre-swing, which could generate the driving force for the human body to move forward. The muscle tension of the biceps femoris and semitendinosus decreased significantly (p &lt; 0.05) during the terminal stance, pre-swing, and initial swing. The decreased muscle tension could relieve the burden of the knee flexion when the knee joint was passively flexed. At the terminal swing, the muscle force of the tibial anterior increased significantly (p &lt; 0.05), which could improve the ankle dorsiflexion ability and prevent foot drop and push forward. Thus, the neuromuscular function of cerebral palsy during walking can be evaluated by the muscle activation state and the RMS of the sEMG signal, which showed that multilevel surgical treatments are feasible and effective to treat SCP.