Feasibility of a direct brain-to-muscle upper-limb neuroprosthesis

点击翻译按钮获取中文摘要

DESCRIPTION (provided by applicant): Implanted neuromuscular stimulation systems capable of restoring full arm and hand motion have now been implemented in people paralyzed below the neck due to spinal cord injury. Intracortical microelectrode arrays are also now being tested in other paralyzed individuals for chronically recording neural activity in the motor cortex. By combining these two technologies, we can potentially develop complete systems that will bypass damage in the spinal cord and restore natural movement by thought to people with spinal cord injuries. However, if the brain signals are decoded into kinematic aspects of movements (e.g. limb position, velocity, joint angles etc.), then we still have to develop additional technology to convert those kinematic commands into the muscle stimulation patterns needed to generate the desired movement. This is not a trivial task. In this study, we will evaluate the alternative approach of retraining the bain to control the muscle stimulators directly, therefore bypassing the challenging and still open problem of how best to translate one's intended movement into the muscle activation levels that will generate those movements. Several studies have shown that local field potentials and the firing rates of motor cortical neurons correlate with recorded muscle activity in able- bodied animals. However, in this study we are specifically testing if motor cortex can be retrained to command the muscle activation levels needed to generate the desired motion of a paralyzed limb where only a subset of the normal muscles can be stimulated and muscles have atrophied after paralysis. Since intracortical recording electrodes are not always capable of detecting firin of individual neurons as the body encapsulates the electrodes over time, we will also compare the effectiveness of using firing rates of the recorded neurons to control the muscle stimulators versus using the more robustly recorded local field potentials to control muscle stimulators. The novel decoding methods we will use in this study can be applied clinically to paralyzed individuals and not just able-bodied animals. In aims 1-2, monkeys will be trained to use the firing rates of ensembles of motor and premotor cortical neurons to control the activation levels of subsets of muscles in a real-time musculoskeletal simulation of a paralyzed arm. The monkeys will be given juice rewards when they successfully move the simulated arm to different targets. In aim1 the monkeys will control movements of the model arm in the horizontal plane by controlling the activation levels of six muscles. In aim 2 external forces will be applied to the model arm. The animals will have to alter the muscle activation levels (by alternating their neural firing) in order to still get the arm to the targets. Aim 3 will determine if the same level of arm control seen in aims 1 & 2 can also be achieved using only local field potentials recorded from intracortical microelectrode arrays.