Psychological remarks on statistical analysis of fMRI data

A standard approach in fMRI research of psychological phenomena search for activated parts of the brain, under the assumption that these parts of the brain participate in the process, while others, which are not significantly activated, do not. We argue that the reasoning behind this assumption is psychologically questionable, as well as rather problematic from the point of view of functional neuroanatomy and neurophysiology. As a consequence it is a main interest for us to find alternative theoretical models, research approaches, and especially alternative ways to analyse fMRI data, that do not conform to this kind of theorising.

Analysis of continuous fMRI data from resting state and mediation

Preliminary presentation of an ungoing analysis of continuous fMRI data from resting state and meditation using SPM and FSL in combination. We aim at further analysis with other ICA programs and the STEPP program as well.

Component analyses of fMRI

Component analyses (PCA,ICA,CCA,PLS etc) are current popular exploratory tools for fMRI. I will give a review and discuss means for quantifying performance and reproducibility of component representations.

Encoding of electrophysiology in MR images - a way of avoiding imaging artifacts

Recording of EEG during MR imaging is complicated by the presence of imaging artifacts that are typically orders of magnitude larger than the signals of interest. We currently develop a conceptually simple method where the scanner is used for both imaging and recording of non-MR signals. The images are left unaffected. The inherent precise timing allows for recording of electrophysiology in sub-millisecond periods free of gradient activity and it facilitates subtraction of residual artifacts.

Nonlinear hemodynamics

Different models of neural activity can be compared using cross-validation to obtain estimates of predictive likelihoods for each model. Predictions for each model are generated by Markov chain Monte Carlo sampling.

Analysing fMRI experiments using sparse PCA

Principal component analysis (PCA) is a widely used tool in medical image analysis for data reduction, model building, and data understanding and exploration. While PCA is a holistic approach where each new variable is a linear combination of all original variables, sparse PCA (SPCA) aims at producing easily interpreted models through sparse loadings, i.e. each new variable is a linear combination of a subset of the original variables. One of the aims of using SPCA is the possible separation of the results into isolated and easily identifiable effects.
In this talk I will demonstrate that sparse PCA applied to fMRI experiments can be used to detect isolated activation patterns. Results are compared to ML ICA methods.

Measuring spontaneous activity with simultaneous EEG-fMRI

Measuring spontaneous neuronal activity is complicated by several noise sources of structured noise in fMRI. I will review how to model these effects and on the possibility of using simultaneous EEG/fMRI to overcome some of the fundamental problems.

fMRI noise correction by Unsupervised Nuisance Variable Regression

It is often assumed that noise in fMRI data is both white and normal; this assumption however is often violated. One way to correct for non-white and non-noise noise in fMRI is by removing known nuisance variables by simple linear regression (Nuisance Variable Regression, NVR). These nuisance variables are often based on knowledge of the physiological noise sources (eg. Fourier expansion of cardiac and respiratory cycles). I will suggest a way to obtain nuisance variables without having recorded these physiological signals and show that the obtained variables are in fact similar to the ones obtained by standard NVR.

Parallel Factor as an exploratory tool for wavelet transformed EEG-data

Although the multi-way analysis model Parallel Factor (PARAFAC) also named Canonical Decomposition (Candecomp) was introduced in 1970 the model has been used very limited within the neuroscience community. Whereas much attention has been given to the factor analysis model such as PCA and ICA practically no attention has been given to the generalization of factor analysis to higher orders, i.e. PARAFAC. Only very recently it has been demonstrated that PARAFAC decompositions can be useful when analyzing fMRI data. The talk will give an introduction to PARAFAC, show how we have used the model in the analysis of electroencephalographic (EEG) data and sketch how the model can be used to analyze fMRI data.

EEG source connectivity and the prospects for its application to fMRI

Visualizing human subjects as active, self-motivated,and affectively sensitive participants in imaging experiments, rather than modelling them as passively stimulated 'responders' to experimental events motivates, the view that dynamics of brain processes are of essential interest to cognitive neuroscience. These include shifts of attention and performance strategy, search for associations or insight, and active recall, processes in which past associations and affective responses play important roles. EEG source coherence analysis seems to reveal the coupling of synchronized activity between brain regions. This functional connectivity is achieved by studying the correlations between the source signals following the use of brain source montages derived from multiple source models. As more and more evidence reveals that major portions of fMRI BOLD signal dynamics reflect brain processes whose time courses are not simply or linearly related to the timing of stimulus ap! pearance or to gross changes in task demands. The combined used of the EEG/fMRI in functional connenctivity is an important missing link that might further elucidate our understanding 'top-down' as well as 'bottom-up brain processes.

Embodiment and religious behavior

Hvis simulationsteorien og princippet om homøostatisk intentionalitet er afgørende for individets religiøse adfærd, vil copingpsykologiske undersøgelser af personer med forskellige stress-niveauer (homøostatisk aktivitet) demonstrere forskellige grader af repræsentationel og adfærdsmæssig kompleksitet i religiøse tanker og handlinger. Dette skulle kunne påvises ved at opstille fMRI-eksperimenter, hvor fokus rettes på de homøostatiske systemers aktivitet (blodgennemstrøming) i kognitive repræsentationer af gudsforholdet under forsøgspersonernes bønsaktivitet. Religiøse personer med højt stress-niveau fx patienter med en alvorlig diagnose eller forældre, som har mistet et barn etc. skulle i teorien adskille sig fra en kontrolgruppe ved en tydeligere korrelation mellem de religiøse repræsentationer og hjernens belønningssystemer samt kroppens homøostatiske reaktioner: det gælder primært områderne somatosensorisk kortex (særligt ifølge Damasios Somatic Marker Hypothesis! ), ventromediale orbitofrontale kortex, anterior cingulate kortex og Amygdala samt kropslige reaktioner på ANS-aktivitet.

Neurofunctional methods and future implication in clinics

Principles of neurofunctional imaging methods will be presented. The biology behind fMRI will be described. The knowledge of functional localization is of importance for therapeutic intervention: surgery in the brain, optimizing training procedures, restoring brain function after vascular insult. Furthermore, interfacing brain cortex by electrodes for restoring functionality of paralyzed extremities with electrical stimulation systems; intra-cerebral electrode systems for focused stimulation; intra-vascular administered bio-probes for local chemical action will be discussed. Functional interfacing of the brain not only relays on information on brain topography but also on network connectivity and semantic interaction.

Bayesian analysis of a spatio-temporal model for fMRI data

Data obtained during fMRI are a realization of a complex spatio-temporal process with many sources of variation, both biological and technical. Most current model-based methods of analysis are based on a two-step procedure. The initial step is a voxel-wise analysis of the temporal changes in the data while the spatial part of the modelling is done separately as a second step in the analysis. We discuss a spatio-temporal point process model approach for fMRI data where the temporal and spatial activation are modelled simultaneously. The focus of this talk will be on Bayesian inference of model parameters.

What happens to the default system in the active brain?

Recently Fransson (Human Brain Mapping 2005) and Fox et al. (PNAS 2005) have shown that two anti-correlated functional networks are at play in the resting brain during spontaneous fluctuations in the BOLD signal. This presentation shows the results of a similar analysis of fluctuations during complex task solving, including both visual, auditory (verbal) and motor processes. The implications of the results are discussed.