Showing papers by "Bertrand Thirion published in 2006"

Dealing with the shortcomings of spatial normalization: multi-subject parcellation of fMRI datasets.

Abstract: The analysis of functional magnetic resonance imaging (fMRI) data recorded on several subjects resorts to the so-called spatial normalization in a common reference space. This normalization is usually carried out on a voxel-by-voxel basis, assuming that after coregistration of the functional images with an anatomical template image in the Talairach reference system, a correct voxel-based inference can be carried out across subjects. Shortcomings of such approaches are often dealt with by spatially smoothing the data to increase the overlap between subject-specific activated regions. This procedure, however, cannot adapt to each anatomo-functional subject configuration. We introduce a novel technique for intra-subject parcellation based on spectral clustering that delineates homogeneous and connected regions. We also propose a hierarchical method to derive group parcels that are spatially coherent across subjects and functionally homogeneous. We show that we can obtain groups (or cliques) of parcels that well summarize inter-subject activations. We also show that the spatial relaxation embedded in our procedure improves the sensitivity of random-effect analysis.

Combined permutation test and mixed‐effect model for group average analysis in fMRI

Abstract: In group average analyses, we generalize the classical one-sample t test to account for heterogeneous within-subject uncertainties associated with the estimated effects. Our test statistic is defined as the maximum likelihood ratio corresponding to a Gaussian mixed-effect model. The test's significance level is calibrated using the same sign permutation framework as in Holmes et al., allowing for exact specificity control under a mild symmetry assumption about the subjects' distribution. Because our likelihood ratio test does not rely on homoscedasticity, it is potentially more sensitive than both the standard t test and its permutation-based version. We present results from the Functional Imaging Analysis Contest 2005 dataset to support this claim.

Anatomo-Functional Description of the Brain : A Probabilistic Approach

Abstract: The acquisition of brain images in fMRI yields rich topographic information about the functional structure of the brain. However, these descriptions are limited by strong inter-subject variability. A recent approach to represent the gross functional architecture across the population as seen in fMRI consists in automatically defining accross-subjects brain parcels. This technique yields large-scale inter-subject correspondences while allowing some spatial relaxation in the alignment of the brains. We address here the open question of an optimal parameterization (number of parcels) of brain parcellations using information theoretic criteria and cross-validation. Moreover, a finer analysis of variance components enables us to better characterize intra- and inter-subject variability sources in parcellation models.

Robust statistics for nonparametric group analysis in fMRI

Abstract: In order to deal with inhomogeneous groups of subjects in fMRI studies, we investigate several robust statistics to perform random-effect analysis on the mean population effect (sign statistic, Wilcoxon's signed rank statistic and empirical t statistic). The tests' significance levels are calibrated using the same sign permutation framework, allowing for exact specificity control under a mild symmetry assumption about the subjects' distribution. The benefit of the robust approach is illustrated on a speech listening experiment involving three month-old infants.

Human Retinotopic Mapping Using fMRI

Abstract: We present in this report a new method for the retinotopic mapping of the human visual cortex using fMRI. This fast method allow s to delineate any human's occipital retinotopic visual areas after 30 minutes in an MR scanner. Based on the known retinotopic properties o f the visual cortex and on the procedures described in the literature, we first detail the experimental protocol we used. We then present th e functional data analysis we perform to get the retinotopic angular maps. The algorithm to get a model of the cortical surface from the ana tomical MR image is also rapidly presented. We then show the retinotopic maps projected on the latter model and compare them with the litera ture. Lastly, we present the choices we made to delineate these areas and extract regions of interest that can be used for further studying the human visual cortical system.

Improving Sensitivity and Reliability of fMRI Group Studies through High Level Combination of Individual Subjects Results

Abstract: fMRI group studies are usually based on the computation of a mean signal for each voxel across subjects (Random Effects Analyzes), assuming that all subjects are in the same anatomical space (Talairach space). Although this is the standard approach, it lacks efficiency because its underlying hypotheses are often violated. We present here a new framework that detects structures of interest from each subject’s data, then searches for correspondences across subjects and outlines the most reproducible activation in the group studied. This framework enables a strict control on the number of false positives. It is shown here that this analysis demonstrates increased validity and improves both the sensitivity and reliability of group analyzes compared to standard methods. Moreover, it directly provides information on the activated regions spatial position correspondence or variability across subjects, which is difficult to obtain in standard voxel-based analyzes.

Reading the brain visual system as an inverse problem

Abstract: Standard inference in neuroimaging consists in describing brain activations elicited and modulated by different kinds of stimuli. Recently however, paradigms have been studied in which the converse operation is performed, thus inferring behavioral or mental states associated with activation images. Here, we use the well-known retinotopy of the visual cortex to infer the visual content of real scenes from the activation patterns that they elicit. We present an explicit decoding technique, based on the current knowledge of the retinotopic structure of the visual areas. Our algorithm can predict the stimulus identity with significant accuracy.