Functional electrical stimulation (FES) can improve the gait of stroke patients by stimulating the peroneal nerve in the swing phase of the affected leg, causing dorsiflexion of the foot that allows the toes to clear the ground. A sensor can trigger the electrical stimulation automatically during the stroke gait. We previously used a heel sensor system, which detects the contact pressure of the heel, in FES to correct foot drop gait. However, the heel sensor has disadvantages in cosmetics and durability. Therefore, we have replaced the heel sensor with an acceleration sensor that can detect the swing phase based on the acceleration speed of the affected leg, using a machine learning technique (Neural Network). We have used a signal for heel contact in a gait using the heel sensor before training with the Neural Network. The accuracy of the Neural Network detector was compared with a swing phase detector based on the heel sensor. The Neural Network detector was able to detect similarly the swing phase in the heel sensor. The largest difference in timing of the swing phase was less than 60 milliseconds in normal subjects and 80 milliseconds in stroke patients. We were able to correct foot drop gait using FES with an acceleration sensor and Neural Network detector. The present results indicate that an acceleration sensor positioned on the thigh, which is cosmetically preferable to systems in which the sensor is farther from the entry point of the electrodes, is useful for correction of stroke gait using FES.
ASJC Scopus subject areas
- Biochemistry, Genetics and Molecular Biology(all)