Kristine Marie Mosier DMD, PhD, Presenter: Nothing to Disclose

Yang Wang MD, Abstract Co-Author: Nothing to Disclose

Robert Scheidt PhD, Abstract Co-Author: Nothing to Disclose

Santiago Acosta MS, Abstract Co-Author: Nothing to Disclose

Ferdinando A Mussa-Ivaldi PhD, Abstract Co-Author: Nothing to Disclose

Brain-machine interfaces (BMI) are systems that link the output of firing neurons in the cerebral cortex with a non-biological device. The purpose of these systems is to provide a means to “thought-control” artificial limbs or other apparatus for paralyzed patients. Most systems developed have been tested in primates and only recently has approval been obtained for human application. In learning to control the artificial device, the brain must learn to map new motor commands into the artificial system. However, determining the cortical changes that occur in this re-mapping in humans with implanted systems is not feasible using current fMRI methods. We have developed a simple fMRI paradigm to simulate the learning challenges in BMI that can be applied to healthy subjects as well as other patient populations (e.g. stroke, Parkinson’s, etc).

Seventeen healthy adult subjects were examined in a behavioral study and three subjects were imaged on a Siemens 3T Trio or GE 1.5T Signa MR system using standard BOLD techniques. The subjects wore a MR-compatible cyberglove (DataGlove, 5DT, San Jose, CA) on the dominant hand. The 16 output signals from the glove were transformed into a two-dimensional (x,y) position of a cursor that was displayed to the subject on a computer screen. A total of 50 targets were presented to which the subject was required to align the cursor.

Activation was observed in the primary motor, sensory, premotor, and parietal cortices. Impulse response functions for different cortical areas showed either phasic or tonic patterns of activity that differed with the phase of learning (early vs late).

The results of this study demonstrate subjects can learn to remap motor commands into a novel coordinate system, and that differential modulation in premotor and parietal cortices is associated with learning of this task. These results provide valuable information on training strategies for BMIs, but additionally provide insight into motor recovery following stroke or following deep brain stimulation for Parkinson’s.

Mosier, K, Wang, Y, Scheidt, R, Acosta, S, Mussa-Ivaldi, F, Radiology and the Cyborg: A Novel Functional Imaging Paradigm as a Testbed for Brain-Machine Interfaces.  Radiological Society of North America 2004 Scientific Assembly and Annual Meeting, November 28 - December 3, 2004 ,Chicago IL. http://archive.rsna.org/2004/4416242.html