Final Report Summary - E-WALK (Spinal cord rehabilitation enhanced by the use of data-driven and dynamic cortical state models)
Project summary:
Continuous electrical spinal cord stimulation (ESCS) can improve motor control after various neurological disorders. However, technical and conceptual limitations have restricted the development of more effective stimulation protocols. Moreover, stimulation protocols remain constant, irrespective of subject’s intentions. Here, we introduce a neuroprosthetic platform capable of spatially selective ESCS controlled by subject’s movement intentions decoded from motor cortex neuronal activity - a brain spinal interface (BSI). While our understanding of the organization of the brain has advanced dramatically in recent years, there have been few successes building a complete system to enable people in such states to regain the ability to interact with and control their environment. Two clinical approaches to treatment have been used: one where brain scientists and engineers have developed cortical implants to record and decode the intended movement for prosthetic control, and one where robotic assisted rehabilitation of the damaged area is driven by coordinated electrochemical stimulation. These two neural interface technologies paint the background of the proposed research. Expertise in the design of brain machine interfaces combined with the advanced spinal neuroprosthesis developed in the host laboratory open the intriguing possibility to merge both approaches, and pioneer a brain spinal interface (BSI) system capable of restoring movement in severely paralyzed subjects.
Results achieved in this project:
In an unprecedented and highly collaborative effort, we have designed, built, and evaluated the first brain-spinal interface in non-human primates to modulate locomotor circuits via brain-controlled epidural electrical stimulation. Three rhesus macaque monkey were implanted with (i) a 96-microelectrode Blackrock cortical array in the lower limb area of left MI, (ii) an 8-channel electromyogram (EMG) system into eight right leg muscles spanning four joints of the lower limb and (iii) a 16-electrode epidural electrical spinal cord stimulation (ESCS) array placed over the lumbar spinal cord. All three implants were equipped with modules for wireless data transfer which allowed us to simultaneously record wideband (30kHz) neuronal data and high fidelity EMG signals (2kHz) and initiate temporally and spatially selective ESCS protocols while the monkey walked freely on a horizontal treadmill at a speed of 1.5km/h. In real time, a linear discriminate analysis (LDA) algorithm predicted Foot Off and Foot Strike events based on 96 channels of multi-unit activity (MUA) and, upon prediction, initiated the Flexion and Extension ESCS protocols designed to increase flexion or extension of the right leg, respectively. Such synchronous phasic stimulation led to increased activity of flexor or extensor muscles during the swing or stance gait phases, respectively. In this way, we enhanced the locomotion without disrupting the natural rhythmic alternation of movements. We demonstrated a new method to manipulate spinal circuits in space and time that can be used for basic neuroscience research and translational medicine. Our results provide a substantial step for the development of ESCS-based neuroprostheses in order to reestablish locomotion in paralyzed individuals.
Additional detail. During recordings, EMG and neuronal signals were collected, synchronized and digitized by the Neural Signal Processor, which streamed the recordings over the local Ethernet network. The control computer ran a custom made C++ application which collected the recordings, performed the prediction of the gait states using the LDA algorithm and, upon prediction, sent the commands to initiate the stimulation protocols. The commands were sent via Bluetooth to the stimulation programmer attached on monkey’s back. The programmer relayed the commands to the implanted stimulator via infrared connection, which then sent the pulses to the ESCS array to stimulate the spinal cord. Prediction-to-stimulation delay was estimated to be 192ms. LDA prediction algorithm provided Foot Off and Foot Strike probabilities given the 96 channels of MUA history. It was calibrated using Foot Off and Foot Strike events of the right leg extracted from EMGs, and 96 channels of MUA taken from 4-5 minutes of recordings during monkey’s free walking in the absence of ESCS. Flexion or Extension ESCS protocols were initiated when Foot Off or Foot Strike probabilities passed 0.8 respectively. Subsequent analysis revealed that 88% of Flexion and Extension stimulations were initiated within 150ms of the exact Foot Off and Foot Strike events reconstructed from motion tracking.
Socioeconomic impact:
Neuromotor disease and insult negatively affect the lives of millions of people worldwide. In some cases, such as severe spinal cord injury (SCI), a completely functional central nervous system is abruptly disconnected from the completely functional body. Currently, there is no therapy capable of promoting recovery after complete spinal cord injury. While our understanding of the organization of the brain has advanced dramatically in recent years, there have been few successes building a complete system to enable people in such states to regain the ability to interact with and control their environment.
Continuous electrical spinal cord stimulation (ESCS) can improve motor control after various neurological disorders. However, technical and conceptual limitations have restricted the development of more effective stimulation protocols. Moreover, stimulation protocols remain constant, irrespective of subject’s intentions. Here, we introduce a neuroprosthetic platform capable of spatially selective ESCS controlled by subject’s movement intentions decoded from motor cortex neuronal activity - a brain spinal interface (BSI). While our understanding of the organization of the brain has advanced dramatically in recent years, there have been few successes building a complete system to enable people in such states to regain the ability to interact with and control their environment. Two clinical approaches to treatment have been used: one where brain scientists and engineers have developed cortical implants to record and decode the intended movement for prosthetic control, and one where robotic assisted rehabilitation of the damaged area is driven by coordinated electrochemical stimulation. These two neural interface technologies paint the background of the proposed research. Expertise in the design of brain machine interfaces combined with the advanced spinal neuroprosthesis developed in the host laboratory open the intriguing possibility to merge both approaches, and pioneer a brain spinal interface (BSI) system capable of restoring movement in severely paralyzed subjects.
Results achieved in this project:
In an unprecedented and highly collaborative effort, we have designed, built, and evaluated the first brain-spinal interface in non-human primates to modulate locomotor circuits via brain-controlled epidural electrical stimulation. Three rhesus macaque monkey were implanted with (i) a 96-microelectrode Blackrock cortical array in the lower limb area of left MI, (ii) an 8-channel electromyogram (EMG) system into eight right leg muscles spanning four joints of the lower limb and (iii) a 16-electrode epidural electrical spinal cord stimulation (ESCS) array placed over the lumbar spinal cord. All three implants were equipped with modules for wireless data transfer which allowed us to simultaneously record wideband (30kHz) neuronal data and high fidelity EMG signals (2kHz) and initiate temporally and spatially selective ESCS protocols while the monkey walked freely on a horizontal treadmill at a speed of 1.5km/h. In real time, a linear discriminate analysis (LDA) algorithm predicted Foot Off and Foot Strike events based on 96 channels of multi-unit activity (MUA) and, upon prediction, initiated the Flexion and Extension ESCS protocols designed to increase flexion or extension of the right leg, respectively. Such synchronous phasic stimulation led to increased activity of flexor or extensor muscles during the swing or stance gait phases, respectively. In this way, we enhanced the locomotion without disrupting the natural rhythmic alternation of movements. We demonstrated a new method to manipulate spinal circuits in space and time that can be used for basic neuroscience research and translational medicine. Our results provide a substantial step for the development of ESCS-based neuroprostheses in order to reestablish locomotion in paralyzed individuals.
Additional detail. During recordings, EMG and neuronal signals were collected, synchronized and digitized by the Neural Signal Processor, which streamed the recordings over the local Ethernet network. The control computer ran a custom made C++ application which collected the recordings, performed the prediction of the gait states using the LDA algorithm and, upon prediction, sent the commands to initiate the stimulation protocols. The commands were sent via Bluetooth to the stimulation programmer attached on monkey’s back. The programmer relayed the commands to the implanted stimulator via infrared connection, which then sent the pulses to the ESCS array to stimulate the spinal cord. Prediction-to-stimulation delay was estimated to be 192ms. LDA prediction algorithm provided Foot Off and Foot Strike probabilities given the 96 channels of MUA history. It was calibrated using Foot Off and Foot Strike events of the right leg extracted from EMGs, and 96 channels of MUA taken from 4-5 minutes of recordings during monkey’s free walking in the absence of ESCS. Flexion or Extension ESCS protocols were initiated when Foot Off or Foot Strike probabilities passed 0.8 respectively. Subsequent analysis revealed that 88% of Flexion and Extension stimulations were initiated within 150ms of the exact Foot Off and Foot Strike events reconstructed from motion tracking.
Socioeconomic impact:
Neuromotor disease and insult negatively affect the lives of millions of people worldwide. In some cases, such as severe spinal cord injury (SCI), a completely functional central nervous system is abruptly disconnected from the completely functional body. Currently, there is no therapy capable of promoting recovery after complete spinal cord injury. While our understanding of the organization of the brain has advanced dramatically in recent years, there have been few successes building a complete system to enable people in such states to regain the ability to interact with and control their environment.
