Investigation of muscle activity during loaded human gait using signal processing of multi-channel surface EMG and IMU

Abstract

Human gait is a complex process resulting from contraction of various muscle groups with different sizes. With the loss of a lower limb, amputees use passive prosthetics to replace the lost limb and regain function. Operating a prosthetic leg, requires more metabolic energy expenditure and greater pressure on the residual limb. In order to understand the muscle activity during human gait, a set of loads were used to model the amputated gait on normal subjects. The loads comprised of sandbags with weights of 5, 10, and 15 lbs. Using 10 Inertial Measurement Units (IMU) alongside 20 Electromyography (EMG) sensors, physiological and kinetic signals were recorded with non-invasive sensors placed on the lower body. Trials were comprised of recording gait from 8 voluntary subjects, and this data was analyzed in the following steps. First, the data was pre-processed using signal processing techniques and, the steps were extracted using a local extrema detection technique from IMU signal as time stamps. Next, to have a numerical measure for the ease of analysis, several features extracted from the EMG signal for each step. The distribution of the features extracted from the signals while subjects performed gait in different states were compared. The results were obtained using students' t-test and the hypothesis of having the same distribution was rejected with a p-value of less than 0.005. The results revealed that the muscles on the intact limb had more activity and sensitivity as a result of compensation for the loaded leg. Vastus Medialis, Vastus Lateralis and Biceps Femoris for the left leg provided escalation in activity according to the features for 100% of the subjects, even with the addition of the smallest load (5 lbs.). Results of this study will determine the sensitivity of muscles to deviation from normal gait and, fewer number of inputs will be used to calibrate and control an active prosthetic limb. This will reduce the complexity and increase the speed of computation.