MRC DiMeN Doctoral Training Partnerships: Inference of the neural drive to muscles using tensor decompositions and single-trial decoding algorithms

The aim of this project is to elucidate how the human brain and spinal cord interact with the musculoskeletal system to generate upper-limb movement. The post-holder will develop a mathematical modelling framework to analyse neurophysiological (electromyography – EMG) and behavioural signals (kinematics - kinetics) recorded while human subjects performs a variety of postural and dynamic tasks. The analysis framework will be based on our past work on the extraction of muscle synergies from EMG signals but will also be informed by our recent findings on the physiological properties of the intermuscular and corticospinal pathways involved in motor control.
The project will consist of two stages. During the first stage, the post-holder will first extend the current muscle synergy framework to consider the spectral properties of EMG signals and assess their role in motor coordination. This will lead to the development of new machine learning algorithms for muscle synergy extraction. Second, the new algorithms will be applied to EMG recordings from the upper limb during static motor tasks to probe muscle covariations in the frequency domain. This analysis will provide novel insights about the neural motor drives that activate the upper limb muscles and will investigate whether the muscle input is primarily shared or specific to each muscle.
During the second stage, the above framework will be applied to EMG recordings from more complex dynamic motor tasks to assess whether the neural basis of motor coordination underlying static tasks can predict muscle activation patterns during dynamic movement. The goal is to provide a compact yet accurate and complete characterization of upper limb neuro-muscular patterns of activity that will serve as a benchmark for healthy upper limb movement and against which clinical populations can be tested. Moreover, the identified patterns of muscle activity could be used to drive neuro-prosthetic devices that will be able to effectively move artificial limbs or robotic rehabilitation devices aiming to restore proper motor function.