Classification and analysis of rhythmic electromyographic activity in lower limb muscles in individuals with spinal cord injury elicited by epidural electrical stimulation

Abstract

During the last decades, independent animal studies on spinal control have led to the now wide-accepted idea that specific spinal neural networks, known as central pattern generators (CPG), are responsible for basic rhythmic activity underlying different types of locomotion. However, the existence of CPGs has remained controversial in the human spinal cord and species-specific adaptations to bipedal locomotion are unsolved. Epidural electrical stimulation (EES), a therapy tool used to relieve chronic, intractable pain, can have an additional effect in individuals with severe spinal cord injuries (SCI): when applied over the lumbar spinal cord, i.e., the spinal segments innervating the lower limbs, the stimulation can induce rhythmic and interdependent activities in the lower limb muscles despite invariant stimulation parameters. These activities can be recorded using surface-electromyography (EMG). Some of the rhythmic activities involve reciprocity between flexor and extensor muscles of one leg as well as between left and right legs, and are referred to as “locomotor-like activity.”The objective of the present thesis was to develop and validate algorithms to automatically identify and extract segments of rhythmic EMG activity recorded from single or multiple lower limb muscles.To this aim, I exploited in my thesis a database of several hours of EMG recordings previously collected from seven individuals with motor-complete SCI (neurological levels: C4-Th10). All individuals participated in a clinical program using EES of the lumbar spinal cord to control severe lower-limb spasticity and pain. The EMG recordings were acquired during the testing phases of EES, when EES was applied with various active electrode combinations, stimulation frequencies and amplitudes to determine the individually most effective parameter combination for spasticity control. EMG activity was recorded from the quadriceps and hamstrings muscle groups, tibialis anterior and the triceps surae muscle group bilaterally. Algorithms operating on a time scale of several seconds of EMG recordings were developed to describe the specific behavior of rhythmic EMG activities occurring in single muscle groups as well as to investigate the relationship between rhythmic activities occurring simultaneously in several muscle groups of one or both legs.Applying the algorithms to the database showed tibialis anterior was rhythmically active in 251 sections of the 997 total tested sections. In comparison, quadriceps showed rhythmic behaviour in 201 sections, hamstrings in 222 sections, and triceps surae in 208 sections. In terms of duration, tibialis anterior was involved in 71% of all rhythmic activity, which makes it the most rhythmically active muscle compared to quadriceps (58%), hamstrings (57%) and triceps surae (57%). A statistical analysis suggested Tibialis Anterior was significantly more rhythmic than expected, while the Hamstring was significantly less rhythmic than expected. Additionally, some cases of reciprocity between mutually rhythmic tibialis anterior muscles of the left and right side were punctually observed. Such analysis can provide insight into the capacity of the human lumbar spinal cord network to coordinate flexor-extensor and left-right activity. Further, it allows to solidify the relationship between applied EES parameters and the generation of locomotor-like activities in paralyzed lower limb muscles of individuals with motor-complete SCI. In fact, the algorithms provided clear cases of rhythmic behaviour in all four studied muscles of the database, under various EES parameters; a complete list registering every case, for every muscle, was produced. The algorithms developed in this thesis will hence help advance the current knowledge of locomotor circuits within the human lumbar spinal cord. Such knowledge is crucial for understanding the neural control of locomotion at the level of the lumbar spinal cord and conceiving innovative neurorehabilitation paradigms for individuals with SCI.