Extracting electrophysiological correlates of functional magnetic resonance imaging data using the canonical polyadic decomposition

The hemodynamic response function (HRF) links neuronal activity to functional magnetic resonance imaging (fMRI) signals. While most fMRI studies use a canonical HRF, increasing evidence from studies of healthy subjects suggests that the HRF depends on anatomical location and disease states. Here, we investigate how HRF variability relates to anatomical location and pathology in the epileptic brain, using a large simultaneous electroencephalogram and fMRI dataset. Applying HRF deconvolution and temporal decomposition, we built the first whole-brain HRF library specific to epilepsy, identifying four distinct shape groups. We mapped HRF features across parcellations of two atlases using novel Bayesian hierarchical models. In non-epileptogenic regions, HRF shape and spatial distributions align with findings from healthy subjects. Within pathological regions, they vary significantly according to pathology. Our results indicate that HRF variability is associated with pathology, in addition to its dependence on anatomical location, motivating region- and pathology-based HRF modulation in epilepsy studies.

In an event-related functional MRI data analysis, an accurate and robust extraction of the hemodynamic response function (HRF) and its associated statistics (e.g., magnitude, width, and time to peak) is critical to infer quantitative information about the relative timing of the neuronal events in different brain regions. The aim of this paper is to develop a multiscale adaptive smoothing model (MASM) to accurately estimate HRFs pertaining to each stimulus sequence across all voxels. MASM explicitly accounts for both spatial and temporal smoothness information, while incorporating such information to adaptively estimate HRFs in the frequency domain. One simulation study and a real data set are used to demonstrate the methodology and examine its finite sample performance in HRF estimation, which confirms that MASM significantly outperforms the existing methods including the smooth finite impulse response model, the inverse logit model and the canonical HRF.

The selection of the appropriate hemodynamic response function (HRF) for signal modeling in functional magnetic resonance imaging (fMRI) is important. Although the use of the boxcar-shaped hemodynamic response function (BHRF) and canonical hemodynamic response (CHRF) has gained increasing popularity in rodent fMRI studies, whether the selected HRF affects the results of rodent fMRI has not been fully elucidated. Here we investigated the signal change and t-statistic sensitivities of BHRF, CHRF, and impulse response function (IRF). The effect of HRF selection on different tasks was analyzed by using data collected from two groups of rats receiving either 3 mA whisker pad or 3 mA forepaw electrical stimulations (n = 10 for each group). Under whisker pad stimulation with large blood-oxygen-level dependent (BOLD) signal change (4.31 ± 0.42%), BHRF significantly underestimated signal changes (P < 0.001) and t-statistics (P < 0.001) compared with CHRF or IRF. CHRF and IRF did not provide significantly different t-statistics (P > 0.05). Under forepaw stimulation with small BOLD signal change (1.71 ± 0.34%), different HRFs provided insignificantly different t-statistics (P > 0.05). Therefore, the selected HRF can influence data analysis in rodent fMRI experiments with large BOLD responses but not in those with small BOLD responses.

BOLD fMRI has become a prevalent method to study cerebral sensory processing in rodent disease models, including pain and mechanical hypersensitivity. fMRI data analysis is frequently combined with a general-linear-model (GLM) -based analysis, which uses the convolution of a hemodynamic response function (HRF) with the stimulus paradigm. However, several studies indicated that the HRF differs across species, sexes, brain structures, and experimental factors, including stimulation modalities or anesthesia, and hence might strongly affect the outcome of BOLD analyzes. While considerable work has been done in humans and rats to understand the HRF, much less is known in mice. As a prerequisite to investigate mechano-sensory processing and BOLD fMRI data in male and female mice, we (1) designed a rotating stimulator that allows application of two different mechanical modalities, including innocuous von Frey and noxious pinprick stimuli and (2) determined and statistically compared HRFs across 30 brain structures and experimental conditions, including sex and, stimulus modalities. We found that mechanical stimulation lead to brain-wide BOLD signal changes thereby allowing extraction of HRFs from multiple brain structures. However, we did not find differences in HRFs across all brain structures and experimental conditions. Hence, we computed a whole-brain mouse HRF, which is based on 88 functional scans from 30 mice. A comparison of this mouse-specific HRF with our previously reported rat-derived HRF showed significantly slower kinetics in mice. Finally, we detected pronounced differences in cerebral BOLD activation between male and female mice with mechanical stimulation, thereby exposing divergent processing of noxious and innocuous stimuli in both sexes.

PurposeRecent studies have applied the new magnetic resonance encephalography (MREG) sequence to the study of interictal epileptic discharges (IEDs) in the electroencephalogram (EEG) of epileptic patients. However, there are no criteria to quantitatively evaluate different processing methods, to properly use the new sequence.MethodsWe evaluated different processing steps of this new sequence under the common generalized linear model (GLM) framework by assessing the reliability of results. A bootstrap sampling technique was first used to generate multiple replicated data sets; a GLM with different processing steps was then applied to obtain activation maps, and the reliability of these maps was assessed.ResultsWe applied our analysis in an event‐related GLM related to IEDs. A higher reliability was achieved by using a GLM with head motion confound regressor with 24 components rather than the usual 6, with an autoregressive model of order 5 and with a canonical hemodynamic response function (HRF) rather than variable latency or patient‐specific HRFs. Comparison of activation with IED field also favored the canonical HRF, consistent with the reliability analysis.ConclusionThe reliability analysis helps to optimize the processing methods for this fast fMRI sequence, in a context in which we do not know the ground truth of activation areas. Magn Reson Med 78:1–1, 2017. © 2016 International Society for Magnetic Resonance in Medicine

SummaryHumans show a remarkable ability to discriminate others' gaze direction, even though a given direction can be conveyed by many physically dissimilar configurations of different eye positions and head views. For example, eye contact can be signaled by a rightward glance in a left-turned head or by direct gaze in a front-facing head. Such acute gaze discrimination implies considerable perceptual invariance. Previous human research found that superior temporal sulcus (STS) responds preferentially to gaze shifts [1], but the underlying representation that supports such general responsiveness remains poorly understood. Using multivariate pattern analysis (MVPA) of human functional magnetic resonance imaging (fMRI) data, we tested whether STS contains a higher-order, head view-invariant code for gaze direction. The results revealed a finely graded gaze direction code in right anterior STS that was invariant to head view and physical image features. Further analyses revealed similar gaze effects in left anterior STS and precuneus. Our results suggest that anterior STS codes the direction of another's attention regardless of how this information is conveyed and demonstrate how high-level face areas carry out fine-grained, perceptually relevant discrimination through invariance to other face features.

The impact of temporal regularization on estimates of the BOLD hemodynamic response function: A comparative analysis

In this paper, we used time-domain functional near infrared spectroscopy (TD-fNIRS) to evaluate the haemodynamic response function (HRF) in the occipital cortex following visual stimulation in glaucomatous eyes as compared to healthy eyes. A total of 98 subjects were enrolled in the study and clinically classified as healthy subjects, glaucoma patients (primary open-angle glaucoma) and mixed subjects (i.e. with a different classification for the two eyes). After quality check data were used from HRF of 73 healthy and 62 glaucomatous eyes. The amplitudes of the oxygenated and deoxygenated haemoglobin concentrations, together with their latencies with respect to the stimulus onset, were estimated by fitting their time course with a canonical HRF. Statistical analysis showed that the amplitudes of both haemodynamic parameters show a significant association with the pathology and a significant discriminating ability, while no significant result was found for latencies. Overall, our findings together with the ease of use and noninvasiveness of TD-NIRS, make this technique a promising candidate as a supporting tool for a better evaluation of the glaucoma pathology.

.SignificanceDeaf and hard-of-hearing infants’ hemodynamic response function (HRF) has not yet been characterized. However, without an appropriate estimate of their HRF, neuroimaging modalities relying on hemodynamic responses, e.g., functional magnetic resonance imaging (fMRI) or functional near-infrared spectroscopy (fNIRS), cannot be used reliably in this population, e.g., prior to and following cochlear implantation. We contribute to better theoretical models of and more suitable therapeutic interventions for the neural changes induced by auditory and language deprivation in deaf infants.AimWe aim to characterize the parameters of the HRF of deaf and hard-of-hearing infants in response to Italian.ApproachWe measured 2- to 20-month-old Italian-exposed infants’ HRF to Italian using fNIRS in the bilateral temporal, i.e., auditory, cortices. We characterized the HRF for all infants and for three clinically relevant subsets: (i) monolingual Italian infants, (ii) genetically deaf infants, and (iii) infants aged 5 to 12 months. We computed the following parameters: peak amplitude, time-to-peak, full width at half maximum, and where present, the amplitude and latency of the initial dip and/or the final undershoot, using a model-based parameter fitting approach. We statistically compared these HRF parameters to those of typically hearing infants.ResultsDeaf and typically hearing infants showed largely similar HRFs, with both groups reaching comparable main peak amplitudes. Minor differences have been found in the latencies of some response components.ConclusionsOur results provide the first detailed characterization of the hemodynamic response to native-language speech in deaf and hard-of-hearing infants, improving clinical and therapeutic approaches through more accurate analysis of fMRI and fNIRS data.

The simultaneous acquisition and subsequent analysis of EEG and fMRI data is challenging owing to increased noise levels in the EEG data. A common method to integrate data from these two modalities is to use aspects of the EEG data, such as the amplitudes of event-related potentials (ERP) or oscillatory EEG activity, to predict fluctuations in the fMRI data. However, this relies on the acquisition of high quality datasets to ensure that only the correlates of neuronal activity are being studied. In this study, we investigate the effects of head-motion-related artefacts in the EEG signal on the predicted T2*-weighted signal variation. We apply our analyses to two independent datasets: 1) four participants were asked to move their feet in the scanner to generate small head movements, and 2) four participants performed an episodic memory task. We created T2*-weighted signal predictors from indicators of abrupt head motion using derivatives of the realignment parameters, from visually detected artefacts in the EEG as well as from three EEG frequency bands (theta, alpha and beta). In both datasets, we found little correlation between the T2*-weighted signal and EEG predictors that were not convolved with the canonical haemodynamic response function (cHRF). However, all convolved EEG predictors strongly correlated with the T2*-weighted signal variation in various regions including the bilateral superior temporal cortex, supplementary motor area, medial parietal cortex and cerebellum. The finding that movement onset spikes in the EEG predict T2*-weighted signal intensity only when the time course of movements is convolved with the cHRF, suggests that the correlated signal might reflect a BOLD response to neural activity associated with head movement. Furthermore, the observation that broad-spectral EEG spikes tend to occur at the same time as abrupt head movements, together with the finding that abrupt movements and EEG spikes show similar correlations with the T2*-weighted signal, indicates that the EEG spikes are produced by abrupt movement and that continuous regressors of EEG oscillations contain motion-related noise even after stringent correction of the EEG data. If not properly removed, these artefacts complicate the use of EEG data as a predictor of T2*-weighted signal variation.

The hemodynamic response function (HRF) describes the dynamic process of the blood oxygen level dependent (BOLD) signal evoked by a very short stimulus of unit intensity. Understanding HRF is a key issue for exploring the underlying dynamics of nervous system activation. Some researches suggested that the shape of HRF varied with different brain regions and different subjects, even at different time point in the same brain region of a subject. In this work, epilepsy discharges were adopted as markers of short stimuli, and then we use EEG-fMRI method to investigate HRF in temporal lobe epilepsy with interictal regional slow wave. Five patients with idiopathic temporal lobe epilepsy were enrolled in this study, the peak delay and amplitude of the estimated HRF was compared with Gamma, Glover and the SPM canonical HRF, and the results show that the delay of the estimated HRF is different from either one of the three theoretical model, thus further study on both theoretical model and real data estimation is needed.

Pain is a complex experience that is thought to emerge from the activity of multiple brain areas, some of which are inconsistently detected using traditional fMRI analysis. One hypothesis is that the traditional analysis of pain-related cerebral responses, by relying on the correlation of a predictor and the canonical hemodynamic response function (HRF)- the general linear model (GLM)- may under-detect the activity of those areas involved in stimulus processing that do not present a canonical HRF. In this study, we employed an innovative data-driven processing approach- an inter-run synchronization (IRS) analysis- that has the advantage of not establishing any pre-determined predictor definition. With this method we were able to evidence the involvement of several brain regions that are not usually found when using predictor-based analysis. These areas are synchronized during the administration of mechanical punctate stimuli and are characterized by a BOLD response different from the canonical HRF. This finding opens to new approaches in the study of pain imaging.

In BOLD fMRI data analysis, robust and accurate estimation of the Hemodynamic Response Function (HRF) is still under investigation. Parametric methods assume the shape of the HRF to be known and constant throughout the brain, whereas non-parametric methods mostly rely on artificially increasing the signal-to-noise ratio. We extend and develop a previously proposed method that makes use of basic yet relevant temporal information about the underlying physiological process of the brain BOLD response in order to infer the HRF in a Bayesian framework. A general hypothesis test is also proposed, allowing to take advantage of the knowledge gained regarding the HRF to perform activation detection. The performances of the method are then evaluated by simulation. Great improvement is shown compared to the Maximum-Likelihood estimate in terms of estimation error, variance, and bias. Robustness of the estimators with regard to the actual noise structure or level, as well as the stimulus sequence, is also proven. Lastly, fMRI data with an event-related paradigm are analyzed. As suspected, the regions selected from highly discriminating activation maps resulting from the method exhibit a certain inter-regional homogeneity in term of HRF shape, as well as noticeable inter-regional differences.

Non-parametric Bayesian models of response function in dynamic image sequences

Non-parametric hemodynamic response function (HRF) estimation in noisy functional Magnetic Resonance Imaging (fMRI) plays an important role when investigating the temporal dynamic of a brain region response during activations. Assuming the drift Lipschitz continuous; a new algorithm for non-parametric HRF estimation is derived in this paper. The proposed algorithm estimates the HRF by applying a first order differencing to the fMRI time series samples. It is shown that the proposed HRF estimator is √(N) consistent. Its performance is assessed using both simulated and a real fMRI data sets obtained from an event-related fMRI experiment. The application results reveal that the proposed HRF estimation method is efficient both computationally and in term of accuracy.