SCIRTS Postdoctoral Fellowship for Robert Flint in Support of: Brain machine interface for grasp combining kinetics and kinematics

Restoring lost hand function after spinal cord injury is a major public health goal and a compelling scientific and engineering challeange. It also represents a potentially enormous change in the quality of life for people who live with spinal cord injury. I propose a scientific study that addresses a gap in our current knowledge: how does the brain command the fingers to grasp an object? In a noninjured person, the hands are controlled by the brain’s commands, executed by the muscles. Following a spinal cord injury, the commands can no longer reach their targets, but they are still being generated. I believe the best way to understand these commands, and to restore lost hand function, is to record brain signals directly. Then, the desired movements that a person wants to make can either control a prosthesis, or be used to stimulate the paralyzed muscles in the person’s own hand to make it move again. This method can restore lost function in a way that seems natural, and does not require extensive effort by the user. A key to understanding the brain signals that control hand function is to examine what happens when a person grasps an object. Previous studies of the human motor system tell us that moving the fingers through empty space is very different from exerting force on an object, but we don’t yet know how the brain signals change to bring about this difference. I am proposing to build a bridge between the brain and an external device, sometimes called a brain-machine interface. This interface can help us learn how the transition from movement to force takes place. I plan to work with hospital patients who are being monitored with electrocorticography (ECoG). ECoG records signals from a sheet of flat metal discs, embedded in soft plastic. These are placed on the surface of the brain during a surgical procedure. I have worked with ECoG patients in the past, to understand the relationship between brain signals and grasp. I propose that ECoG data contain the information necessary to understand how the brain’s commands are changing when the fingers change from movement to force. In the study, ECoG patients will follow along with a hand grasp behavior as it is demonstrated for them by a virtual hand on a computer screen. The behavior will include both movement and force. While the patient is following along, I will record their movements and their ECoG signals, to learn how the two are related. Using this information, the patients will then control a robotic hand directly with their brain signals. Their brain signals will make the robot move, and then exert force just as they did with their own hands. I will ask a second group of patients to imagine (but not perform) the hand grasp that is being demonstrated. Afterwards, they will also attempt to control the robot to create movement and force. Constructing a decoder in this way is potentially useful for people after spinal cord injury, who can not make example movements.