Rainer Goebel

Analysis of functional image analysis contest (FIAC) data with brainvoyager QX: From single‐subject to cortically aligned group general linear model analysis and self‐organizing group independent component analysis

The Functional Image Analysis Contest (FIAC) 2005 dataset was analyzed using BrainVoyager QX. First, we performed a standard analysis of the functional and anatomical data that includes preprocessing, spatial normalization into Talairach space, hypothesis‐driven statistics (one‐ and two‐factorial, single‐subject and group‐level random effects, General Linear Model [GLM]) of the block‐ and event‐related paradigms. Strong sentence and weak speaker group‐level effects were detected in temporal and frontal regions. Following this standard analysis, we performed single‐subject and group‐level (Talairach‐based) Independent Component Analysis (ICA) that highlights the presence of functionally connected clusters in temporal and frontal regions for sentence processing, besides revealing other networks related to auditory stimulation or to the default state of the brain. Finally, we applied a high‐resolution cortical alignment method to improve the spatial correspondence across brains and re‐run the random effects group GLM as well as the group‐level ICA in this space. Using spatially and temporally unsmoothed data, this cortex‐based analysis revealed comparable results but with a set of spatially more confined group clusters and more differential group region of interest time courses. Hum. Brain Mapp, 2006.

Mapping directed influence over the brain using Granger causality and fMRI

We propose Granger causality mapping (GCM) as an approach to explore directed influences between neuronal populations (effective connectivity) in fMRI data. The method does not rely on a priori specification of a model that contains pre-selected regions and connections between them. This distinguishes it from other fMRI effective connectivity approaches that aim at testing or contrasting specific hypotheses about neuronal interactions. Instead, GCM relies on the concept of Granger causality to define the existence and direction of influence from information in the data. Temporal precedence information is exploited to compute Granger causality maps that identify voxels that are sources or targets of directed influence for any selected region-of-interest. We investigated the method by simulations and by application to fMRI data of a complex visuomotor task. The presented exploratory approach of mapping influences between a region of interest and the rest of the brain can form a useful complement to existing models of effective connectivity.

7T vs. 4T: RF power, homogeneity, and signal-to-noise comparison in head images

Signal‐to‐noise ratio (SNR), RF field (B1), and RF power requirement for human head imaging were examined at 7T and 4T magnetic field strengths. The variation in B1 magnitude was nearly twofold higher at 7T than at 4T (∼42% compared to ∼23%). The power required for a 90° pulse in the center of the head at 7T was approximately twice that at 4T. The SNR averaged over the brain was at least 1.6 times higher at 7T compared to 4T. These experimental results were consistent with calculations performed using a human head model and Maxwells equations. Magn Reson Med 46:24–30, 2001.

The constructive nature of vision: direct evidence from functional magnetic resonance imaging studies of apparent motion and motion imagery

Echoplanar functional magnetic resonance imaging was used to monitor activation changes of brain areas while subjects viewed apparent motion stimuli and while they were engaged in motion imagery. Human cortical areas MT (V5) and MST were the first areas of the ‘dorsal’ processing stream which responded with a clear increase in signal intensity to apparent motion stimuli as compared with flickering control conditions. Apparent motion of figures defined by illusory contours evoked greater activation in V2 and MT/MST than appropriate control conditions. Several areas of the dorsal pathway (V3A, MT/MST, areas in the inferior and superior parietal lobule) as well as prefrontal areas including FEF and BA 9/46 responded strongly when subjects merely imagined moving stimuli which they had seen several seconds before. The activation during motion imagery increased with the synaptic distance of an area from V1 along the dorsal processing stream. Area MT/MST was selectively activated during motion imagery but not during a static imagery control condition. The comparison between the results obtained with objective motion, apparent motion and imagined motion provides further insights into a complex cortical network of motion‐sensitive areas driven by bottom‐up and top‐down neural processes.

Integration of letters and speech sounds in the human brain

Most people acquire literacy skills with remarkable ease, even though the human brain is not evolutionarily adapted to this relatively new cultural phenomenon. Associations between letters and speech sounds form the basis of reading in alphabetic scripts. We investigated the functional neuroanatomy of the integration of letters and speech sounds using functional magnetic resonance imaging (fMRI). Letters and speech sounds were presented unimodally and bimodally in congruent or incongruent combinations. Analysis of single-subject data and group data aligned on the basis of individual cortical anatomy revealed that letters and speech sounds are integrated in heteromodal superior temporal cortex. Interestingly, responses to speech sounds in a modality-specific region of the early auditory cortex were modified by simultaneously presented letters. These results suggest that efficient processing of culturally defined associations between letters and speech sounds relies on neural mechanisms similar to those naturally evolved for integrating audiovisual speech.

Localizing P300 Generators in Visual Target and Distractor Processing: A Combined Event-Related Potential and Functional Magnetic Resonance Imaging Study

Constraints from functional magnetic resonance imaging (fMRI) were used to identify the sources of the visual P300 event-related potential (ERP). Healthy subjects performed a visual three-stimulus oddball paradigm with a difficult discrimination task while fMRI and high-density ERP data were acquired in separate sessions. This paradigm allowed us to differentiate the P3b component of the P300, which has been implicated in the detection of rare events in general (target and distractor), from the P3a component, which is mainly evoked by distractor events. The fMRI-constrained source model explained >99% of the variance of the scalp ERP for both components. The P3b was mainly produced by parietal and inferior temporal areas, whereas frontal areas and the insula contributed mainly to the P3a. This source model reveals that both higher visual and supramodal association areas contribute to the visual P3b and that the P3a has a strong frontal contribution, which is compatible with its more anterior distribution on the scalp. The results point to the involvement of distinct attentional subsystems in target and distractor processing.

Combining multivariate voxel selection and support vector machines for mapping and classification of fMRI spatial patterns

In functional brain mapping, pattern recognition methods allow detecting multivoxel patterns of brain activation which are informative with respect to a subjects perceptual or cognitive state. The sensitivity of these methods, however, is greatly reduced when the proportion of voxels that convey the discriminative information is small compared to the total number of measured voxels. To reduce this dimensionality problem, previous studies employed univariate voxel selection or region-of-interest-based strategies as a preceding step to the application of machine learning algorithms. Here we employ a strategy for classifying functional imaging data based on a multivariate feature selection algorithm, Recursive Feature Elimination (RFE) that uses the training algorithm (support vector machine) recursively to eliminate irrelevant voxels and estimate informative spatial patterns. Generalization performances on test data increases while features/voxels are pruned based on their discrimination ability. In this article we evaluate RFE in terms of sensitivity of discriminative maps (Receiver Operative Characteristic analysis) and generalization performances and compare it to previously used univariate voxel selection strategies based on activation and discrimination measures. Using simulated fMRI data, we show that the recursive approach is suitable for mapping discriminative patterns and that the combination of an initial univariate activation-based (F-test) reduction of voxels and multivariate recursive feature elimination produces the best results, especially when differences between conditions have a low contrast-to-noise ratio. Furthermore, we apply our method to high resolution (2 x 2 x 2 mm(3)) data from an auditory fMRI experiment in which subjects were stimulated with sounds from four different categories. With these real data, our recursive algorithm proves able to detect and accurately classify multivoxel spatial patterns, highlighting the role of the superior temporal gyrus in encoding the information of sound categories. In line with the simulation results, our method outperforms univariate statistical analysis and statistical learning without feature selection.

Who Is Saying "What"? Brain-Based Decoding of Human Voice and Speech

Can we decipher speech content (“what” is being said) and speaker identity (“who” is saying it) from observations of brain activity of a listener? Here, we combine functional magnetic resonance imaging with a data-mining algorithm and retrieve what and whom a person is listening to from the neural fingerprints that speech and voice signals elicit in the listeners auditory cortex. These cortical fingerprints are spatially distributed and insensitive to acoustic variations of the input so as to permit the brain-based recognition of learned speech from unknown speakers and of learned voices from previously unheard utterances. Our findings unravel the detailed cortical layout and computational properties of the neural populations at the basis of human speech recognition and speaker identification.

Mirror-Symmetric Tonotopic Maps in Human Primary Auditory Cortex

Understanding the functional organization of the human primary auditory cortex (PAC) is an essential step in elucidating the neural mechanisms underlying the perception of sound, including speech and music. Based on invasive research in animals, it is believed that neurons in human PAC that respond selectively with respect to the spectral content of a sound form one or more maps in which neighboring patches on the cortical surface respond to similar frequencies (tonotopic maps). The number and the cortical layout of such tonotopic maps in the human brain, however, remain unknown. Here we use silent, event-related functional magnetic resonance imaging at 7 Tesla and a cortex-based analysis of functional data to delineate with high spatial resolution the detailed topography of two tonotopic maps in two adjacent subdivisions of PAC. These maps share a low-frequency border, are mirror symmetric, and clearly resemble those of presumably homologous fields in the macaque monkey.

Individual faces elicit distinct response patterns in human anterior temporal cortex

Visual face identification requires distinguishing between thousands of faces we know. This computational feat involves a network of brain regions including the fusiform face area (FFA) and anterior inferotemporal cortex (aIT), whose roles in the process are not well understood. Here, we provide the first demonstration that it is possible to discriminate cortical response patterns elicited by individual face images with high-resolution functional magnetic resonance imaging (fMRI). Response patterns elicited by the face images were distinct in aIT but not in the FFA. Individual-level face information is likely to be present in both regions, but our data suggest that it is more pronounced in aIT. One interpretation is that the FFA detects faces and engages aIT for identification.