Methods toward improved lower extremity rehabilitation

Abstract

Ambulation is a very important part of everyday life and its absence has a detrimental effect on an individual's quality of life. While much is understood about the neurobiological systems involved in locomotion through detailed anatomical connectivity and lesion studies, it is not well understood how neurons across different regions of the nervous system share information and coordinate their firing activity to achieve ambulation. Moreover, while it is clear that understanding the processes involved in healthy ambulation are essential to understanding how diseases affect an individual's ability to walk, diseases such as stroke tend to "take out" large portions of the underlying system. Until technologies are developed to allow restoration of damaged neural tissue back to its original state, physical therapy (which aims to restore function by establishing new motor-cortical connections among the remaining neurons) remains the most viable option for patients. The aim of this thesis is to elucidate some of the underlying neurobiological mechanisms of walking and to develop tools for rehabilitation robotics that allow finer quantification of patient improvement. To elucidate the neural mechanisms of locomotion, we studied how task relevant information (e.g. positions, velocities, and forces) modulate single unit neural activity from hindlimb/trunk region of the rat motor cortex during adaptations to robot-applied elastic loads and closed-loop brain-machine-interface (BMI) control during treadmill locomotion. Using the Point Process-Generalized Linear Model (PP-GLM) statistical framework we systematically tested parametric and non-parametric point process models of increased complexity for 573 individual neurons recorded over multiple days in six animals. The developed statistical model captures within gait-cycle modulation, load-specific modulation, and intrinsic neural dynamics. Our proposed model accurately describes the firing statistics of 98.5% (563/573) of all the recorded units and allows characterization of the neural receptive fields associated with gait phase and loading force. Understanding how these receptive fields change during training and with experience will be central to developing rehabilitation strategies that optimize motor adaptations and motor learning. The methods utilized for this analysis were developed into an open source neural Spike Train Analysis Toolbox (nSTAT) for Matlab (Mathworks, Natick MA). Systematic analyses have demonstrated the effectiveness of physical therapy, but have been unable to determine which approaches tend to be most effective in restoring function. This is likely due to the multitude of approaches, diseases, and assessment scales used. To address this issue, we develop an extension of the Force Field Adaptation Paradigm, originally developed to quantitatively assess upper extremity motor adaptation, to the lower extremity. The algorithm is implemented on the Lokomat (Hocoma HG) lower extremity gait orthosis and is currently being utilized to assess short-term motor adaptation in 40 healthy adult subjects (ClinicalTrials.gov NCT01361867). Establishing an understanding of how healthy adults' motor systems adapt to external perturbations will be important to understanding how the adaptive mechanisms involved in gait integrate information and how this process is altered by disease.