The research themes of the section involve the continuing effort to develop more robust, informative, and quantitative methods for mapping human brain function. The specific research goals of the section include the following: A) Development of a more comprehensive and detailed understanding of the neuronal correlates of fMRI signal changes; B) Development of methods for extraction of quantitative and clinically relevant physiologic information from the fMRI time series signal; C) Development of processing methods to assess brain activation information content or pattern formation associated with activation; D) Development of integrated paradigm design, behavioral, and processing methods allowing improvement of the quality and usability of information obtain from fMRI time series data; E) Development of methods to assess and use ?resting state? fMRI data (both the baseline fluctuations in the time series and the baseline anatomic images). Section investigators and collaborators made several significant findings. To summarize, the following were among the highlights. 1. We discovered, using MEG (Magnetoencephalography), neuronal correlates of processes involved with perceptual decision-making (i.e. deciding for instance, if a scrambled, noisy picture is a face or a house). This is a beginning towards the better understanding of precisely where and when the appropriate calculations (based on previous experience and a collection of incomplete sensory information) in the brain are made when a perceptual decision is made. 2. We discovered that changes in respiration volume per unit time contribute to ?resting state? BOLD contrast fluctuations. In this study, we identified the resting state ?default? network from regions that showed a decrease in signal during cognitive tasks. We also found that these regions showed a high degree of correlation in their resting state signal. In addition, other regions outside of this ?default network? appeared to show a high degree of correlation in the time series signal. It was hypothesized that these regions were not related to correlated neuronal activity but, instead, were artifacts. During these studies we monitored respiration rate and volume. We found that by regressing out the respiration volume per unit time curve from the time series, the spatial extent of the resting state temporal correlation overlapped more precisely the regions identified as showing a reduction during an array of cognitive tasks. This important study sheds light on contributions to the ?resting state? correlations in the time series and demonstrates a method by which we can correct for them. We are currently investigating the mechanism by which these respiration volume changes cause changes in BOLD signal over time. Dr. Rasmus Birn is primarily carrying out these studies and has submitted a manuscript to NeuroImage relating these findings. 3. We have demonstrated that reduction in voxel volume, coupled with the use of a high sensitivity array of RF coils, actually increases signal to noise and functional contrast to noise in regions that suffer from image artifacts related to magnetic field inhomogeneities. A paper, submitted by Dr. Patrick Bellgowan, is currently in press in NeuroImage, which relates these findings. 4. We have demonstrated that the magnitude of the BOLD response amplitude and linearity is a function of stimulus duty cycle and the ?off? duration of the paradigm. By comparisons with modeled neuronal and BOLD responses, we have determined the relative contribution of neuronal and hemodynamic effects to this observation. This study was published, with Dr. Rasmus Birn as first author, in NeuroImage. 5. We have further developed and refined a multivariate method for ?information? extraction from the fMRI time series signal. Not only does this technique have the potential for being more sensitive in many contexts than standard univariate processing techniques, but also has the potential for revealing new networks of activation at a smaller spatial scale than previously studied with fMRI. The primary person involved with this project is Dr. Niko Kriegeskorte.
Showing the most recent 10 out of 36 publications
