|
Brain
Machine Interfaces
|
Research A fundamental challenge in Neuroscience is understanding how the brain encodes information. With a proper paradigm it is possible to use brain signals to control external devices such as prosthetic limbs. A critical step to realizing this goal will be to "close the loop" such that the external device provides feedback to the brain via tactile or visual stimuli. With such feedback Brain Machine Interfaces will enable artificial devices to be explictly represented in the brain as part of the body, augmenting it in virtual space in ways that never thought possible. Several challenges remain. First, the number of implanted electrodes must be increased to cover more regions in the brain. In addition, mixed signal VLSI technology must be developed to reduce the size of the circuitry required for sampling and analysis of brain signals.A custom VLSI "neurochip" for decoding brain signals is currently being developed at Duke as part of an overall brain machine interface that may one day be used to restore motor function to patients who are paralyzed or affected by a neurological deficit. The technological revolution of the modern era is changing how we acquire and conceive of biological data. This is due to an explosion in the amount of data that can be acquired with an ever increasing array of modern tools.The meaningful reduction of data (in the analytic sense) is becoming an increasing priority, and taking on increased urgency. We are developing new signal processing strategies to look at the full spatio-temporal characteristics of human brain data obtained tools like the recording of brain waves (EEG) or the imaging of brain function using functional MRI. An important goal is to translate these signal processing tools developed in the laboratory, to the clinical setting where the problems of data analysis have reached a crisis state. In recent years, one of the most exciting areas of modern science has developed ---functional imaging of the human brain. The inventions of positron emission tomography (PET), and more recently, functional magnetic resonance imaging (fMRI) have opened up new vistas on human brain research. We may now view the structure and the function of the normal living human brain non-invasively. As a result, for the first time in human history we don't merely think about thinking, we can now measure it and map it. These tools for visualizing the thinking brain represent the cutting-edge of modern engineering and physics, and together with experimental questions developed by cognitive scientists and human neuroscientists, represent a stunning advance in brain research. A critical deficiency of these exciting new technologies is a paucity of appropriate image analysis tools to fully realize the vast potential of image data acquired from these revolutionary imaging systems. The goal of this research is to advance the methodologies for the processing and registration of functional brain images. These new tools will enhance our ability to evaluate of the statistical significance of (possibly) active regions of interest within experimentally acquired images so as to account for local spatial correlations, or to jointly model simultaneous or nearly simultaneous activations at distinct locations in the brain. The statistical methodology developed in this program has the potential for revolutionizing the analysis not only of functional brain images, but a much broader class of medical and non-medical images as well. Ensemble
Recording and Processing
|
