The FIM section is continuing an effort to develop more robust, informative, and quantitative methods for mapping human brain function in both the activated state and during resting state. To summarize, the following were among the highlights, organized by research project corresponding to the listed post doc.? ? Rasmus Birn:? The Respiration Response Function:? Changes in the subjects breathing rate and depth can cause significant fMRI signal changes, which can decrease the detection power of functional activation, or lead to false positives, particularly in functional connectivity analyses. In this study, we derive a new respiration response function, which models the fMRI signal changes induced by variations in breathing. This new function fits the signal changes more accurately than previous models. ? ? Improvement of physiological noise correction:? Fluctuations in the fMRI signal resulting from the subjects heart beat and respiration are a dominant source of noise in fMRI, particularly at high field strengths. To reduce this noise, physiological noise correction techniques are typically employed. The success of these techniques, however, is reduced by subject motion. In this study we first evaluate the optimal order of conventional pre-processing steps image registration, slice-timing corrections, and physiological noise correction. We also present an improved physiological noise correction technique which incorporates information from the subjects motion. ? ? The effect of respiration variations on functional connectivity assessed by ICA:? Variations in the subjects respiration depth and rate can significantly affect estimates of functional connectivity derived from the correlation of signal fluctuations in the brain with a seed region. In this study, we investigate the effect of respiration variations on functional connectivity maps derived from independent component analysis (ICA) of resting-state data. We find that in most cases, ICA separates fluctuations in the default mode network from respiration-related signal changes, suggesting that ICA is more robust to respiration artifacts. ? ? Sources of functional under-connectivity in autism spectrum disorders:? Decreased correlations of BOLD fMRI time series between spatially remote regions of the brain have been observed in adolescents with autism spectrum disorders (ASD), supporting a model of under-connectivity in autism. This observed decrease in correlation, however, can arise from a number of possible mechanisms. In this study, we find that the reduced correlations between brain regions is not the result of trial-to-trial variability in performance or increased noise (such as subject motion). Furthermore, these differences in connectivity are greater when only rest periods are considered. These findings suggest that the disruption in functional connectivity in ASD is likely due to differences in task-unrelated neuronal fluctuations.? ? Kevin Murphy:? Previous studies of functionally connectivity with fMRI have found a set of brain regions that are negatively correlated with fluctuations in the default mode network of the brain. The interpretation is that the human brain is intrinsically organized into dynamic, anti-correlated functional networks. These studies, however, have all employed global signal regression as a pre-processing step. In this study, we show through both simulation and experiments that global signal regression can introduce anti-correlations, calling into question the previous interpretation of anti-correlated networks within the brain. ? ? Niko Kriegeskorte:? We have been pioneering the use of patten effect analysis for optimal extraction of information from brain activation images. Our focus has been on developing the novel approach of representational similarity analysis, which allows us (1) to combine evidence across brain space and experimental conditions to detect neuronal pattern information and (2) to relate results (a) between different modalities of brain-activity measurement, (b) between different species, and (c) between brain-activity data and computational models of brain information processing. This approach has been applied to comparing human and monkey data from hi-res fMRI and single-cell recordings, respectively. We investigated response patterns elicited by the same 92 photographs of isolated natural objects in inferotemporal (IT) cortex of both species. Within each species, we computed a matrix of response-pattern similarities (one similarity value for each pair of images). We found a striking match of the resulting similarity matrices for man and monkey. This finding suggests very similar categorical IT representations and provides some hope that data from single-cell recording and fMRI, for all their differences, may consistently reveal neuronal representations when subjected to massively multivariate analyses of response-pattern information.? ? Masaya Misaki:? Multivariate analysis can extract more information from the BOLD fMRI response. It has been indicated that spatial pattern of brain activation has more information than just one point of brain activation. To explore what aspects of BOLD signal have information about brain function further, we evaluated how much information the variables of BOLD fMRI response had. The variables of the BOLD response we investigated were spatial pattern of response amplitude fitted by the model of hemodynamic response function (HRF), spatial pattern of response amplitude in each time point from the onset of event, and spatiotemporal pattern of the brain response. We developed the multivariate mutual information analysis method for fMRI data and used the mutual information measure, which can capture any relationship between variables, to evaluate how much information the BOLD response has. We found spatiotemporal pattern of brain activation sometimes have more information about presented stimuli than spatial pattern of brain activation.? ? Dan Handwerker:? Functional MRI responses to the same tasks are known to vary greatly across individuals and populations. This variation affects the precision of group analysis studies and, if the variation is a systematic difference across populations, it can cause biased results. We quantitatively examined the effects of subject selection on a standard group analysis method. We showed less precision than is generally assumed and identified a new method, using binary significance maps from each volunteer, to increase group analysis precision. ? ? Anthony Boemio:? An approach to speech perception research has been to determine the mapping between local acoustic cues and perceived linguistic units such as phonemes, syllables, and words.We instead formulate speech perception as a problem in auditory pattern recognition. On this view, speech perception is characterized by the way the global spectrotemporal image (rather than local acoustic cues) are? mapped onto linguistic units. To discern this mapping, we decomposed American English sentences into sine wave speech, and then parametrically manipulated the position and morphology of the individual sinusoids. Both rigid and non-rigid body manipulations were performed.Intelligibility was measured to assess the effect of each manipulation. We found that for all manipulations, intelligibility decreased in a manner consistent with the pattern recognition framework. Specifically,intelligibility was well predicted by 1) the aggregate information contained in the speech signal, and 2) the relation of the global spectrotemporal image to auditory system anatomy and function such as the number of active channels and auditory cortical integration time constants.
Showing the most recent 10 out of 36 publications
