THE FRANCIS LAB
Brain-machine interfaces for movement and touch
We build closed-loop interfaces that decode what the brain intends to do and use microstimulation to feed sensation back to it.
About the lab
We record from populations of neurons in sensorimotor cortex and thalamus and decode their activity with machine learning. The decoder's output runs robotic arms and hands. We also work in the other direction, designing microstimulation of peripheral nerves, the somatosensory thalamus and somatosensory cortex so that the cortical activity it evokes looks like the activity from a real touch. Down the road this should let a person feel a prosthetic limb as well as move it.
A second line of work asks how reward and motivation, and the effort a movement takes, change activity in motor and somatosensory cortex, and how an interface could use those signals to adjust itself. We draw on neurophysiology, control theory, reinforcement learning and deep learning, whatever the problem needs.
We're in the Department of Biomedical Engineering in the Cullen College of Engineering at the University of Houston, and our current work is supported by the National Institutes of Health.
Current work
Touch through microstimulation
We stimulate the somatosensory thalamus and record what that does in cortex. Then we keep adjusting the stimulation until the cortical response looks like the response to natural touch.
Interfaces that learn from reward
Reward expectation shows up in motor cortex activity, which gives an interface a feedback signal it could use to adjust itself. Using spiking activity and local field potentials together, we predicted reward outcome with 97 percent accuracy.
Mathematical modeling and simulation
Models let us test an idea before it reaches an animal. We built spiking models of sensorimotor cortex that learn to reach with a virtual arm and can drive a robotic one. Other models predict the S1 field potentials evoked by thalamic microstimulation, or choose stimulation patterns by adaptive inverse control. Deep networks are the newest addition.
Join the lab
We're looking for a Ph.D. student to work on somatosensory neuroprosthetics and closed-loop neural control. If you're an M.S. or undergraduate student at UH and want a research project, ask us.
Francis Lab
Department of Biomedical Engineering
Cullen College of Engineering
University of Houston, Houston, Texas
Contact
jtfrancis@uh.edu
713-743-0985