Novel sensors for adaptive neurorehabilitation systems

点击翻译按钮获取中文摘要

My research is concerned with developing technology that can help restore function after neurological injuries, such as spinal cord injury (SCI) or stroke. My current focus is on innovative signal and image processing solutions for monitoring and assessing the activity of the nervous system. These measures are needed as feedback control signals to create neurorehabilitation systems that can adapt to changing conditions both in the user’s body and in the external environment. In particular, implanted systems have used functional electrical stimulation (FES) of motor neural pathways to restore movement, but these devices currently rely on pre-defined patterns of stimulation and thus can only produce crude, fixed movements. The objective of the research program proposed here is to overcome this limitation by developing sensors that can provide the feedback signals needed to create closed-loop implanted FES systems. The intact nervous system achieves effective motor control only by relying on a combination of proprioceptive (limb position), tactile and visual information. My specific aims are therefore:1) To develop an implantable neural interface capable of monitoring the sensory information in peripheral nerves, such as tactile signals and limb position information. This interface will be based on multi-contact nerve cuff technology (cylindrical electrodes that wrap around a nerve, with recording contacts located on their inner surface). In one study, we will optimize the design of the electrodes and the signal processing algorithms, using first a novel computer model of a peripheral nerve, later followed by in vivo validation. In a second study, we will explore a new class of signal processing algorithms designed specifically to extract information from multi-contact nerve cuff recordings by exploiting spatiotemporal relationships in the electrical activity of the nerve. These projects are the first proposed attempt to tailor a multi-contact nerve cuff specifically to the monitoring of sensory activity. These devices have not yet been used for this purpose, but are a promising avenue because they combine suitability for chronic use in humans with improved selectivity compared to single-channel cuffs. 2) To develop a wearable sensor capable of providing visual information about the user’s environment and their interactions with it. This will make it possible to incorporate visual information into the control algorithms for closed-loop neuroprostheses. We will focus here on upper limb function. Computer vision technology will be used to analyze in real-time video from an ear-mounted wearable camera recording the user’s point of view. We will explore image processing and machine learning algorithms capable of parsing the visual information in the immediate neighbourhood of the hand with the goal of detecting environmental interactions (e.g. grasp attempt, success or failure, grip type used). No previous solution has been proposed to incorporate visual information into the control of FES. Our proposed study is the first to address this gap.Significance: The proposed research will produce novel sensor technologies that will be essential if we are to develop closed-loop neuroprosthetic systems capable of restoring natural movements to individuals with neurological injuries. Innovations in the natural science and engineering will include a new class of algorithms for processing neural signals, new computer vision techniques tailored to wearable cameras, and improved computer modeling of bioelectric activity. The research will ultimately lead to improved independence and quality of life after stroke and SCI, and reduce the economic impact of these conditions on the health care system.