Thomas M. Talavage - US grants

Description (provided by applicant): As the usage of functional magnetic resonance imaging (fMRI) for the purpose of tracking neurological activity increases, the spectrum of experimental and research applications of this technology is destined to diversify. For fMRI to achieve its potential, it is essential for the interaction between the subject and the stimuli to become more sophisticated. A user-interface device that enhances the spectrum of responses a subject may make and that may be recorded and analyzed will (1) simplify the procedure of measuring and interpreting response times, (2) permit complex responses to be studied, and (3) make IMRI more useful from a clinical standpoint. It is the intent of this research project to develop a set of user-interface equipment to increase the breadth of fMRI research. Current devices used for complex subject interaction with a presented stimulus or virtual environment in fMRI experiments suffer from two short-comings that may affect both subject safety and the quality of the acquired data. The first short-coming of these devices (including commercially available "MRI-compatible" devices) is that they typically contain significant quantities of electrical components that require shielding for proper operation in the high magnetic field environment. The second short-coming, primarily affecting image quality, is that typical user-input devices require non-trivial physical movements on the part of the subject to achieve a particular task. This research project will develop a compact hand-controller device for use in fMRI experiments that (1) has essentially no interaction with the magnetic fields required for imaging, and (2) permits a significant number of inputs through which the subject, using limited physical movement, may interact with a stimulus. The construction of the physical device will require novel engineering solutions to provide robust operation of both binary and analog inputs. It is anticipated that it will be possible to construct a refined version of the hand-controller and associated software such that this device may be made readily accessible to the fMRI research community. The availability of a hand-controller that limits possible interference with the imaging process, is easy to use, and provides a significant variety of subject responses (particularly when hand-controllers are utilized in both hands) will increase the ease and accuracy of future experiments requiring subject feedback and interaction.

This award provides support for an REU Site at Purdue University for up to three years. Ten students will be recruited from throughout the U.S. for a ten-week summer experience in research related to biomedical engineering. The program will serve to encourage undergraduates to pursue advanced degrees in engineering and thus impact the nations need to increase the number of U.S. citizens and permanent residents engaged in research-related careers. It is also anticipated that this REU Site will have a Broader Impact on the nations need to have broader participation in the research enterprise of citizens and permanent residents from demographic groups traditionally underrepresented in science and engineering.

[unreadable] DESCRIPTION (provided by applicant): Functional magnetic resonance imaging (fMRI) using echo-planar imaging (EPI) results in generation of a systematic confound-acoustic noise ("acoustic imaging noise")-that can interfere with the fMRI measurement of cortical activity related to a desired (acoustic) stimulus. Acoustic imaging noise is unavoidably perceived by a hearing subject, leading to neuronal activity throughout the auditory pathway, including cortex. Perception of acoustic imaging noise can alter sensory perception of the desired stimulus-an alteration that will affect physiologic activity and distort measurement of the hemodynamic response (HDR) to the desired stimulus. Methods, initially directed at auditory fMRI, are proposed to develop correction algorithms that may be used to correct distortions of the "ideal" HDR (that measured in the absence of confound) associated with systematic confounds. Significant benefit to the fMRI community arises from the process by which the correction algorithm is developed, as the process may be applied to other systematic confounds known to be associated with acquisition of fMRI data. The specific aims of this research are summarized as follows: Aim 1: Obtain "ideal" measurements of the HDR to the systematic confound of acoustic imaging noise for blipped-EPI at 1.5T. Aim 2: Obtain non-ideal measurements of HDRs to a desired (acoustic) stimulus, parameterized by variations in the systematic confound (acoustic imaging noise), allowing quantification of interactions between the HDRs. Aim 3: Develop correction algorithms to compensate for distortions present in non-ideal HDR measurements to a desired (acoustic) stimulus, such that resulting HDR estimates resemble those which would be obtained in absence of the systematic confound (acoustic imaging noise). Relevance: Improvements in localization accuracy and detection sensitivity for fMRI experiments conducted in the presence of repeated confounds (e.g., acoustic noise associated with imaging) will allow more precise discrimination of cortical responses to stimuli that may differ only in a subtle fashion. These benefits will provide greater confidence in location and extent of detected activity, significantly impacting human brain mapping, but also directly benefiting efforts at pre-surgical mapping. [unreadable] [unreadable]

[unreadable] DESCRIPTION (provided by applicant): The project proposed herein will develop a flexible analytical framework for functional magnetic resonance imaging (fMRI) data that will identify multi-voxel regions of activation through statistical evaluation of hemodynamic response model fits to the observed blood-oxygenation level-dependent effect responses. The proposed framework will allow the incorporation of multiple hemodynamic response models that may be used to evaluate fMRI responses both on a single-trial and averaged (blocked) basis. In addition to an expected improvement in the identification of "true" fMRI activation, this framework is expected to be capable of reducing the background noise level in summary images by breaking from the tradition of analyzing each individual voxel for its statistical significance and subsequently grouping "significant" voxels into regions of activation. Finally, this framework will also allow us to identify regions that are co-activated, or at least exhibit similar temporal structure in response to the stimulation, independent of the amplitude of the individual responses, thus contributing a new tool to studies of functional connectivity. Aim: Develop a novel framework for fMRI data analysis that uses a priori knowledge of both the experimental paradigm and hemodynamic response to detect activation on a regional basis. Achievement of the specific aim of this proposal will provide a powerful tool that may greatly enhance the future results obtained by the functional MRI community. Development of an analytical procedure that focuses on (potentially disjoint) multi-voxel regions of interest as the fundamental activation unit to be detected will increase the probability of a detected voxel being a true detection, enhancing the usefulness of fMRI in furthering our understanding of both normal and disordered cortical activity. [unreadable] [unreadable] [unreadable]

This subproject is one of many research subprojects utilizing the resources provided by a Center grant funded by NIH/NCRR. The subproject and investigator (PI) may have received primary funding from another NIH source, and thus could be represented in other CRISP entries. The institution listed is for the Center, which is not necessarily the institution for the investigator. Cochlear implants are used in profoundly deaf individuals to partially restore the sense of hearing. These implants electrically stimulate the cochlea in response to auditory stimuli measured at the implant's microphone. To date, the algorithms used to convert the auditory stimuli to the pattern of electrical stimuli applied over the length of the cochlea have been quite simple and have not been optimized for word recognition. We are interested in developing parametric stimulation algorithms and optimizing the parameters to minimize the perceptual difference between the neural activation patterns (neural firing probability over time and position in the cochlea) of normal hearing listeners and individuals using cochlear implants. For reasons of practicality, it is necessary to use a simulator to compute the neural activation patterns. This simulator is a Monte Carlo simulation, making it well suited to parallel computing. However, given the iterative nature of the optimizations we wish to perform, many such simulations must be made. Thus, access to a computer cluster would be quite advantageous.