Torben E. Lund

Non-white noise in fMRI: does modelling have an impact?

The sources of non-white noise in Blood Oxygenation Level Dependent (BOLD) functional magnetic resonance imaging (fMRI) are many. Familiar sources include low-frequency drift due to hardware imperfections, oscillatory noise due to respiration and cardiac pulsation and residual movement artefacts not accounted for by rigid body registration. These contributions give rise to temporal autocorrelation in the residuals of the fMRI signal and invalidate the statistical analysis as the errors are no longer independent. The low-frequency drift is often removed by high-pass filtering, and other effects are typically modelled as an autoregressive (AR) process. In this paper, we propose an alternative approach: Nuisance Variable Regression (NVR). By inclusion of confounding effects in a general linear model (GLM), we first confirm that the spatial distribution of the various fMRI noise sources is similar to what has already been described in the literature. Subsequently, we demonstrate, using diagnostic statistics, that removal of these contributions reduces first and higher order autocorrelation as well as non-normality in the residuals, thereby improving the validity of the drawn inferences. In addition, we also compare the performance of the NVR method to the whitening approach implemented in SPM2.

Endogenous cortical rhythms determine cerebral specialization for speech perception and production.

Across multiple timescales, acoustic regularities of speech match rhythmic properties of both the auditory and motor systems. Syllabic rate corresponds to natural jaw-associated oscillatory rhythms, and phonemic length could reflect endogenous oscillatory auditory cortical properties. Hemispheric lateralization for speech could result from an asymmetry of cortical tuning, with left and right auditory areas differentially sensitive to spectro-temporal features of speech. Using simultaneous electroencephalographic (EEG) and functional magnetic resonance imaging (fMRI) recordings from humans, we show that spontaneous EEG power variations within the gamma range (phonemic rate) correlate best with left auditory cortical synaptic activity, while fluctuations within the theta range correlate best with that in the right. Power fluctuations in both ranges correlate with activity in the mouth premotor region, indicating coupling between temporal properties of speech perception and production. These data show that endogenous cortical rhythms provide temporal and spatial constraints on the neuronal mechanisms underlying speech perception and production.

Motion or activity: their role in intra- and inter-subject variation in fMRI.

Functional MRI (fMRI) carries the potential for non-invasive measurements of brain activity. Typically, what are referred to as activation images are actually thresholded statistical parametric maps. These maps possess large inter-session variability. This is especially problematic when applying fMRI to pre-surgical planning because of a higher requirement for intra-subject precision. The purpose of this study was to investigate the impact of residual movement artefacts on intra-subject and inter-subject variability in the observed fMRI activation. Ten subjects were examined using three different word-generation tasks. Two of the subjects were examined 10 times on 10 different days using the same paradigms. We systematically investigated one approach of correcting for residual movement effects: the inclusion of regressors describing movement-related effects in the design matrix of a General Linear Model (GLM). The data were analysed with and without modeling the residual movement artefacts and the impact on inter-session variance was assessed using F-contrasts. Inclusion of motion parameters in the analysis significantly reduced both the intra-subject as well as the inter-subject-variance.

Hypercapnic normalization of BOLD fMRI: comparison across field strengths and pulse sequences

The blood oxygenation level-dependent (BOLD) functional magnetic resonance imaging (fMRI) signal response to neural stimulation is influenced by many factors that are unrelated to the stimulus. These factors are physiological, such as the resting venous cerebral blood volume (CBV(v)) and vessel size, as well as experimental, such as pulse sequence and static magnetic field strength (B(0)). Thus, it is difficult to compare task-induced fMRI signals across subjects, field strengths, and pulse sequences. This problem can be overcome by normalizing the neural activity-induced BOLD fMRI response by a global hypercapnia-induced BOLD signal. To demonstrate the effectiveness of the BOLD normalization approach, gradient-echo BOLD fMRI at 1.5, 4, and 7 T and spin-echo BOLD fMRI at 4 T were performed in human subjects. For neural stimulation, subjects performed sequential finger movements at 2 Hz, while for global stimulation, subjects breathed a 5% CO(2) gas mixture. Under all conditions, voxels containing primarily large veins and those containing primarily active tissue (i.e., capillaries and small veins) showed distinguishable behavior after hypercapnic normalization. This allowed functional activity to be more accurately localized and quantified based on changes in venous blood oxygenation alone. The normalized BOLD signal induced by the motor task was consistent across different magnetic fields and pulse sequences, and corresponded well with cerebral blood flow measurements. Our data suggest that the hypercapnic normalization approach can improve the spatial specificity and interpretation of BOLD signals, allowing comparison of BOLD signals across subjects, field strengths, and pulse sequences. A theoretical framework for this method is provided.

Motion verb sentences activate left posterior middle temporal cortex despite static context.

The left posterior middle temporal region, anterior to V5/MT, has been shown to be responsive both to images with implied motion, to simulated motion, and to motion verbs. In this study, we investigated whether sentence context alters the response of the left posterior middle temporal region. ‘Fictive motion’ sentences are sentences in which an inanimate subject noun, semantically incapable of self movement, is coupled with a motion verb, yielding an apparent semantic contradiction (e.g. ‘The path comes into the garden.’). However, this context yields no less activation in the left posterior middle temporal region than sentences in which the motion can be applied to the subject noun. We speculate that the left posterior middle temporal region activity in fictive motion sentences reflects the fact that the hearer applies motion to the depicted scenario by scanning it egocentrically.

Concrete spatial language: see what I mean?

Conveying complex mental scenarios is at the heart of human language. Advances in cognitive linguistics suggest this is mediated by an ability to activate cognitive systems involved in non-linguistic processing of spatial information. In this fMRI-study, we compare sentences with a concrete spatial meaning to sentences with an abstract meaning. Using this contrast, we demonstrate that sentence meaning involving motion in a concrete topographical context, whether linked to animate or inanimate subjects nouns, yield more activation in a bilateral posterior network, including fusiform/parahippocampal, and retrosplenial regions, and the temporal-occipital-parietal junction. These areas have previously been shown to be involved in mental navigation and spatial memory tasks. Sentences with an abstract setting activate an extended largely left-lateralised network in the anterior temporal, and inferior and superior prefrontal cortices, previously found activated by comprehension of complex semantics such as narratives. These findings support a model of language, where the understanding of spatial semantic content emerges from the recruitment of brain regions involved in non-linguistic spatial processing.

Experimentally and computationally fast method for estimation of a mean kurtosis.

Results from several recent studies suggest the magnetic resonance diffusion‐derived metric mean kurtosis (MK) to be a sensitive marker for tissue pathology; however, lengthy acquisition and postprocessing time hamper further exploration. The purpose of this study is to introduce and evaluate a new MK metric and a rapid protocol for its estimation.

Amygdala and heart rate variability responses from listening to emotionally intense parts of a story

Emotions are often understood in relation to conditioned responses. Narrative emotions, however, cannot be reduced to a simple associative relationship between emotion words and their experienced counterparts. Intensity in stories may arise without any overt emotion depicting words and vice versa. In this fMRI study we investigated BOLD responses to naturally fluctuating emotions evoked by listening to a story. The emotional intensity profile of the text was found through a rating study. The validity of this profile was supported by heart rate variability (HRV) data showing a significant correspondence across participants between intensity ratings and HRV measurements obtained during fMRI. With this ecologically valid stimulus we found that narrative intensity was accompanied by activation in temporal cortices, medial geniculate nuclei in the thalamus and amygdala, brain regions that are all part of the system for processing conditioned emotional responses to auditory stimuli. These findings suggest that this system also underpins narrative emotions in spite of their complex nature. Traditional language regions and premotor cortices were also activated during intense parts of the story whereas orbitofrontal cortex was found linked to emotion with positive valence, regardless of level of intensity.

Modelling cardiac signal as a confound in EEG-fMRI and its application in focal epilepsy studies.

Cardiac noise has been shown to reduce the sensitivity of functional Magnetic Resonance Imaging (fMRI) to an experimental effect due to its confounding presence in the blood oxygenation level-dependent (BOLD) signal. Its effect is most severe in particular regions of the brain and a method is yet to take it into account in routine fMRI analysis. This paper reports the development of a general and robust technique to improve the reliability of EEG-fMRI studies to BOLD signal correlated with interictal epileptiform discharges (IEDs). In these studies, ECG is routinely recorded, enabling cardiac effects to be modelled, as effects of no interest. Our model is based on an over-complete basis set covering a linear relationship between cardiac-related MR signal and the phase of the cardiac cycle or time after pulse (TAP). This method showed that, on average, 24.6 +/- 10.9% of grey matter voxels contained significant cardiac effects and 22.3 +/- 24.1% of those voxels exhibiting significantly IED-correlated BOLD signal also contained significant cardiac effects. We quantified the improvement of the TAP model over the original model, without cardiac effects, by evaluating changes in efficiency, with respect to estimating the contrast of the effects of interest. Over voxels containing significant, cardiac-related signal, efficiency was improved by 18.5 +/- 4.8%. Over the remaining voxels, no improvement was demonstrated. This suggests that, while improving sensitivity in particular regions of the brain, there is no risk that the TAP model will reduce sensitivity elsewhere.

An fMRI study of the neural correlates of graded visual perception

The neural correlates of clearly perceived visual stimuli have been reported previously in contrast to unperceived stimuli, but it is uncertain whether intermediate or graded perceptual experiences correlate with different patterns of neural activity. In this study, the subjective appearance of briefly presented visual stimuli was rated individually by subjects with respect to perceptual clarity: clear, vague or no experience of a stimulus. Reports of clear experiences correlated with activation in a widespread network of brain areas, including parietal cortex, prefrontal cortex, premotor cortex, supplementary motor areas, insula and thalamus. The reports of graded perceptual clarity were reflected in graded neural activity in a network comprising the precentral gyrus, intraparietal sulcus, basal ganglia and the insula. In addition, the reports of vague experiences demonstrated unique patterns of activation. Different degrees of perceptual clarity were reflected both in the degree to which activation was found within parts of the network serving a clear conscious percept, and additional unique activation patterns for different degrees of perceptual clarity. Our findings support theories proposing the involvement of a widespread network of brain areas during conscious perception.