2017 Volume 55 Issue 5 Pages 193-204
The incidence of limb paralysis due to cerebrovascular disorders is increasing. In the affected patients, returning to normal life is a concern that needs to be addressed. Therefore an effective treatment for recovery of motor function is required. Rehabilitation to encourage patients to perform a variety of exercises is required. One of the effective methods for recovery of motor function is repetitive facilitation of exercises. This method requires that a therapist assists a patient on a one-to-one basis. The lack of qualified personnel to assist individuals and the increased burden on the medical personnel in recent years have made rehabilitation using repetitive facilitation of exercise difficult. Therefore, it is important that we use repetitive facilitation exercises and training equipment in combination to facilitate high-quality rehabilitation for a large number of patients with paralysis. Such training equipment should allow stable performance of a variety of exercises. The purpose of this study was to develop an arm exercise function recovery system based on surface electrical stimulation using multi-channel electrodes for a patient to undergo rehabilitation without assistance at home. As the training equipment, we used functional electrical stimulation (FES) with surface electrodes, which is a non-invasive method. Stimulating a specific muscle is difficult using FES with surface electrodes. Therefore, we traced an electrode pattern including many electrode points to achieve appropriate stimulation. In the case of multi-channel surface electrodes, we considered that there should be as many electrode patterns as possible. In this report, we describe an example of the method to determine the electrode pattern that can evoke finger and hand positioning using a multi-channel electrode selector “FES selector device”, by which the electrode pattern can be changed freely. We used the electrode pattern that had the highest estimated joint precision obtained by estimating joint angles using an artificial neural network. We then performed clustering of finger and hand positions based on the electrode pattern. This cluster was then used as the cluster that correlates the actual finger and hand positions with the electrode pattern. We performed clustering of the electrode pattern based on this cluster of finger and hand positions. In addition, we demonstrate in this report an example of searching the electrode pattern by clustering using the FES selector device and an artificial neural network. From this search, we identified the electrode patterns that evoked exercise, and found around 16 to 24 clusters of finger and hand positions. These findings suggest that a variety of exercises can be evoked without assistance. Based on the clustering result, we made a table of the required positions and the corresponding electrode patterns. Using this table, a patient can easily find the pattern appropriate to the desired position.
