Noise contributions to the fMRI signal: An overview

Review of weighted exponential random graph models frameworks applied to neuroimaging

Neuro-imaging data can often be represented as statistical networks, especially for functional magnetic resonance imaging (fMRI) data, where brain regions are defined as nodes and the functional interactions between those regions are taken as edges. Such networks are commonly divided into classes depending on the type of edges, namely binary or weighted. A binary network means edges can either be present or absent. Whereas the edges of a weighted network are associated with weight values, and fMRI networks belong to weighted networks. Statistical methods are often adopted to analyse such networks, among which, the exponential random graph model (ERGM) is an important network analysis approach. Typically ERGMs are applied to binary networks, and weighted networks often need to be binarised by arbitrarily selecting a threshold value to define the presence of the edges, which can lead to non-robustness and loss of valuable edge weight information representing the strength of fMRI interaction in fMRI networks. While it is therefore important to gain deeper insight in adopting ERGM on weighted networks, there only exists a few different ERGM frameworks for weighted networks; some of these are not directly implementable on fMRI networks based on their original proposal. We systematically review, implement, analyse and compare five such frameworks via a simulation study and provide guidelines on each modelling framework as well as conclude the suitability of them on fMRI networks based on a range of criteria. We concluded that Multi-Layered ERGM is currently the most suitable framework.

The relative brain signal variability increases in the behavioral variant of frontotemporal dementia and Alzheimer's disease but not in schizophrenia

Overlapping symptoms between Alzheimer's disease (AD), behavioral variant of frontotemporal dementia (bvFTD), and schizophrenia (SZ) can lead to misdiagnosis and delays in appropriate treatment, especially in cases of early-onset dementia. To determine the potential of brain signal variability as a diagnostic tool, we assessed the coefficient of variation of the BOLD signal (CVBOLD) in 234 participants spanning bvFTD (n = 53), AD (n = 17), SZ (n = 23), and controls (n = 141). All underwent functional and structural MRI scans. Data unveiled a notable increase in CVBOLD in bvFTD patients across both datasets (local and international, p < 0.05), revealing an association with clinical scores (CDR and MMSE, r = 0.46 and r = -0.48, p < 0.0001). While SZ and control group demonstrated no significant differences, a comparative analysis between AD and bvFTD patients spotlighted elevated CVBOLD in the frontopolar cortices for the latter (p < 0.05). Furthermore, CVBOLD not only presented excellent diagnostic accuracy for bvFTD (AUC 0.78-0.95) but also showcased longitudinal repeatability. During a one-year follow-up, the CVBOLD levels increased by an average of 35% in the bvFTD group, compared to a 2% increase in the control group (p < 0.05). Our findings suggest that CVBOLD holds promise as a biomarker for bvFTD, offering potential for monitoring disease progression and differentiating bvFTD from AD and SZ.

A Hybrid Routing Pattern in Human Brain Structural Network Revealed By Evolutionary Computation

The human brain functional connectivity network (FCN) is constrained and shaped by the communication processes in the structural connectivity network (SCN). The underlying communication mechanism thus becomes a critical issue for understanding the formation and organization of the FCN. A number of communication models supported by different routing strategies have been proposed, with shortest path (SP), random diffusion (DIF), and spatial navigation (NAV) as the most typical, respectively requiring network global knowledge, local knowledge, and both for path seeking. Yet these models all assumed every brain region to use one routing strategy uniformly, ignoring convergent evidence that supports the regional heterogeneity in both terms of biological substrates and functional roles. In this regard, the current study developed a hybrid communication model that allowed each brain region to choose a routing strategy from SP, DIF, and NAV independently. A genetic algorithm was designed to uncover the underlying region-wise hybrid routing strategy (namely HYB). The HYB was found to outperform the three typical routing strategies in predicting FCN and facilitating robust communication. Analyses on HYB further revealed that brain regions in lower-order functional modules inclined to route signals using global knowledge, while those in higher-order functional modules preferred DIF that requires only local knowledge. Compared to regions that used global knowledge for routing, regions using DIF had denser structural connections, participated in more functional modules, but played a less dominant role within modules. Together, our findings further evidenced that hybrid routing underpins efficient SCN communication and locally heterogeneous structure-function coupling.

Stochastic brain dynamics exhibits differential regional distribution and maturation-related changes

Functional magnetic resonance imaging (fMRI) is a powerful non-invasive method for studying brain function by analyzing blood oxygenation level-dependent (BOLD) signals. These signals arise from intricate interplays of deterministic and stochastic biological elements. Quantifying the stochastic part is challenging due to its reliance on assumptions about the deterministic segment. We present a methodological framework to estimate intrinsic stochastic brain dynamics in fMRI data without assuming deterministic dynamics. Our approach utilizes Approximate Entropy and its behavior in noisy series to identify and characterize dynamical noise in unobservable fMRI dynamics. Applied to extensive fMRI datasets (645 Cam-CAN, 1086 Human Connectome Project subjects), we explore lifelong maturation of intrinsic brain noise. Findings indicate 10% to 60% of fMRI signal power is due to intrinsic stochastic brain elements, varying by age. These components demonstrate a physiological role of neural noise which shows a distinct distributions across brain regions and increase linearly during maturation.

Spatiotemporal cerebral blood flow dynamics underlies emergence of the limbic-sensorimotor-association cortical gradient in human infancy

Infant cerebral blood flow (CBF) delivers nutrients and oxygen to fulfill brain energy consumption requirements for the fastest period of postnatal brain development across lifespan. However, organizing principle of whole-brain CBF dynamics during infancy remains obscure. Leveraging a unique cohort of 100+ infants with high-resolution arterial spin labeled MRI, we found the emergence of the cortical hierarchy revealed by highest-resolution infant CBF maps available to date. Infant CBF across cortical regions increased in a biphasic pattern with initial rapid and sequentially slower rate, with break-point ages increasing along the limbic-sensorimotor-association cortical gradient. Increases in CBF in sensorimotor cortices were associated with enhanced language and motor skills, and frontoparietal association cortices for cognitive skills. The study discovered emergence of the hierarchical limbic-sensorimotor-association cortical gradient in infancy, and offers standardized reference of infant brain CBF and insight into the physiological basis of cortical specialization and real-world infant developmental functioning.

Comparing data-driven physiological denoising approaches for resting-state fMRI: implications for the study of aging

Introduction

Physiological nuisance contributions by cardiac and respiratory signals have a significant impact on resting-state fMRI data quality. As these physiological signals are often not recorded, data-driven denoising methods are commonly used to estimate and remove physiological noise from fMRI data. To investigate the efficacy of these denoising methods, one of the first steps is to accurately capture the cardiac and respiratory signals, which requires acquiring fMRI data with high temporal resolution.

Methods

In this study, we used such high-temporal resolution fMRI data to evaluate the effectiveness of several data-driven denoising methods, including global-signal regression (GSR), white matter and cerebrospinal fluid regression (WM-CSF), anatomical (aCompCor) and temporal CompCor (tCompCor), ICA-AROMA. Our analysis focused on the consequence of changes in low-frequency, cardiac and respiratory signal power, as well as age-related differences in terms of functional connectivity (fcMRI).

Results

Our results confirm that the ICA-AROMA and GSR removed the most physiological noise but also more low-frequency signals. These methods are also associated with substantially lower age-related fcMRI differences. On the other hand, aCompCor and tCompCor appear to be better at removing high-frequency physiological signals but not low-frequency signal power. These methods are also associated with relatively higher age-related fcMRI differences, whether driven by neuronal signal or residual artifact. These results were reproduced in data downsampled to represent conventional fMRI sampling frequency. Lastly, methods differ in performance depending on the age group.

Discussion

While this study cautions direct comparisons of fcMRI results based on different denoising methods in the study of aging, it also enhances the understanding of different denoising methods in broader fcMRI applications.

Performance reserves in brain-imaging-based phenotype prediction

This study examines the impact of sample size on predicting cognitive and mental health phenotypes from brain imaging via machine learning. Our analysis shows a 3- to 9-fold improvement in prediction performance when sample size increases from 1,000 to 1 M participants. However, despite this increase, the data suggest that prediction accuracy remains worryingly low and far from fully exploiting the predictive potential of brain imaging data. Additionally, we find that integrating multiple imaging modalities boosts prediction accuracy, often equivalent to doubling the sample size. Interestingly, the most informative imaging modality often varied with increasing sample size, emphasizing the need to consider multiple modalities. Despite significant performance reserves for phenotype prediction, achieving substantial improvements may necessitate prohibitively large sample sizes, thus casting doubt on the practical or clinical utility of machine learning in some areas of neuroimaging.

Characterizing Spatiotemporal Population Receptive Fields in Human Visual Cortex with fMRI

The use of fMRI and computational modeling has advanced understanding of spatial characteristics of population receptive fields (pRFs) in human visual cortex. However, we know relatively little about the spatiotemporal characteristics of pRFs because neurons' temporal properties are one to two orders of magnitude faster than fMRI BOLD responses. Here, we developed an image-computable framework to estimate spatiotemporal pRFs from fMRI data. First, we developed a simulation software that predicts fMRI responses to a time-varying visual input given a spatiotemporal pRF model and solves the model parameters. The simulator revealed that ground-truth spatiotemporal parameters can be accurately recovered at the millisecond resolution from synthesized fMRI responses. Then, using fMRI and a novel stimulus paradigm, we mapped spatiotemporal pRFs in individual voxels across human visual cortex in 10 participants (both females and males). We find that a compressive spatiotemporal (CST) pRF model better explains fMRI responses than a conventional spatial pRF model across visual areas spanning the dorsal, lateral, and ventral streams. Further, we find three organizational principles of spatiotemporal pRFs: (1) from early to later areas within a visual stream, spatial and temporal windows of pRFs progressively increase in size and show greater compressive nonlinearities, (2) later visual areas show diverging spatial and temporal windows across streams, and (3) within early visual areas (V1-V3), both spatial and temporal windows systematically increase with eccentricity. Together, this computational framework and empirical results open exciting new possibilities for modeling and measuring fine-grained spatiotemporal dynamics of neural responses using fMRI.

A precision neuroscience approach to estimating reliability of neural responses during emotion processing: Implications for task-fMRI

Recent work demonstrating low test-retest reliability of neural activation during fMRI tasks raises questions about the utility of task-based fMRI for the study of individual variation in brain function. Two possible sources of the instability in task-based BOLD signal over time are noise or measurement error in the instrument, and meaningful variation across time within-individuals in the construct itself-brain activation elicited during fMRI tasks. Examining the contribution of these two sources of test-retest unreliability in task-evoked brain activity has far-reaching implications for cognitive neuroscience. If test-retest reliability largely reflects measurement error, it suggests that task-based fMRI has little utility in the study of either inter- or intra-individual differences. On the other hand, if task-evoked BOLD signal varies meaningfully over time, it would suggest that this tool may yet be well suited to studying intraindividual variation. We parse these sources of variance in BOLD signal in response to emotional cues over time and within-individuals in a longitudinal sample with 10 monthly fMRI scans. Test-retest reliability was low, reflecting a lack of stability in between-person differences across scans. In contrast, within-person, within-session internal consistency of the BOLD signal was higher, and within-person fluctuations across sessions explained almost half the variance in voxel-level neural responses. Additionally, monthly fluctuations in neural response to emotional cues were associated with intraindividual variation in mood, sleep, and exposure to stressors. Rather than reflecting trait-like differences across people, neural responses to emotional cues may be more reflective of intraindividual variation over time. These patterns suggest that task-based fMRI may be able to contribute to the study of individual variation in brain function if more attention is given to within-individual variation approaches, psychometrics-beginning with improving reliability beyond the modest estimates observed here, and the validity of task fMRI beyond the suggestive associations reported here.

A Temporal Instability Measure for fMRI Quality Assurance

BACKGROUND

There exist several fMRI quality assurance measures to assess scanner stability. Because they have practical and/or theoretical limitations, a different and more practical measure for instability would be desirable.

PURPOSE

To develop and test a sensitive, reliable and widely applicable temporal instability measure (TIM) for fMRI quality assurance.

STUDY TYPE

Technical development.

PHANTOM

Spherical gel phantom.

POPULATION

A total of 120 datasets from a local Philips scanner with two different receive-only head coils (32ch and 8ch, 60 datasets per coil) were collected as well as 29 additional datasets with three different receive-only head coils (20ch, 32ch, and 64ch) from two additional sites with GE (seven runs with 32ch) and Siemens scanners (seven runs with 32ch and Multiband imaging, five runs with 20ch, 32ch, and 64ch) were borrowed.

FIELD STRENGTH/SEQUENCE

2D Echo-planar-imaging (EPI).

ASSESSMENT

A new TIM was proposed that is based on the eigenratio of the correlation coefficient matrix, where each entry of the matrix is a correlation coefficient between two time-points of the time-series.

STATISTICAL TESTS

Nonparametric bootstrap resampling was used twice to estimate confidence intervals (CI) of the TIM values and to assess the improved sensitivity of this measure. Differences in coil performance were assessed via a nonparametric bootstrap two-sample t-test. P-values <0.05 were considered significant.

RESULTS

The TIM values ranged between 60 parts-per-million and 10,780 parts-per-million across all 149 experiments. The mean CI was 2.96% and 2.16% for the 120 and 29 fMRI datasets, respectively (the repeated bootstrap analysis gave 2.9% and 2.19%, respectively). The 32ch coils of the local Philips data provided more stable measurements than the 8ch coil (observed two-sample t-values = 26.36, -0.2 and -6.2 for TIM, tSNR, and RDC, respectively. PtSNR  = 0.58).

DATA CONCLUSION

The proposed TIM is particularly useful for multichannel coils with spatially nonuniform receive sensitivity and overcomes several limitations of other measures. As such, it provides a reliable test for ascertaining scanner stability for fMRI experiments.

EVIDENCE LEVEL

5.

TECHNICAL EFFICACY

Stage 1.

Impact of sample size and regression of tissue-specific signals on effective connectivity within the core default mode network

Interactions within brain networks are inherently directional, which are inaccessible to classical functional connectivity estimates from resting-state functional magnetic resonance imaging (fMRI) but can be detected using spectral dynamic causal modeling (DCM). The sample size and unavoidable presence of nuisance signals during fMRI measurement are the two important factors influencing the stability of group estimates of connectivity parameters. However, most recent studies exploring effective connectivity (EC) have been conducted with small sample sizes and minimally pre-processed datasets. We explore the impact of these two factors by analyzing clean resting-state fMRI data from 330 unrelated subjects from the Human Connectome Project database. We demonstrate that both the stability of the model selection procedures and the inference of connectivity parameters are highly dependent on the sample size. The minimum sample size required for stable DCM is approximately 50, which may explain the variability of the DCM results reported so far. We reveal a stable pattern of EC within the core default mode network computed for large sample sizes and demonstrate that the use of subject-specific thresholded whole-brain masks for tissue-specific signals regression enhances the detection of weak connections.

Inter-regional correlation estimators for functional magnetic resonance imaging

Functional magnetic resonance imaging (fMRI) functional connectivity between brain regions is often computed using parcellations defined by functional or structural atlases. Typically, some kind of voxel averaging is performed to obtain a single temporal correlation estimate per region pair. However, several estimators can be defined for this task, with various assumptions and degrees of robustness to local noise, global noise, and region size. In this paper, we systematically present and study the properties of 9 different functional connectivity estimators taking into account the spatial structure of fMRI data, based on a simple fMRI data spatial model. These include 3 existing estimators and 6 novel estimators. We demonstrate the empirical properties of the estimators using synthetic, animal, and human data, in terms of graph structure, repeatability and reproducibility, discriminability, dependence on region size, as well as local and global noise robustness. We prove analytically the link between regional intra-correlation and inter-region correlation, and show that the choice of estimator has a strong influence on inter-correlation values. Some estimators, including the commonly used correlation of averages (ca), are positively biased, and have more dependence to region size and intra-correlation than robust alternatives, resulting in spatially-dependent bias. We define the new local correlation of averages estimator with better theoretical guarantees, lower bias, significantly lower dependence on region size (Spearman correlation 0.40 vs 0.55, paired t-test T=27.2, p=1.1e-47), at negligible cost to discriminative power, compared to the ca estimator. The difference in connectivity pattern between the estimators is not distributed uniformly throughout the brain, but rather shows a clear ventral-dorsal gradient, suggesting that region size and intra-correlation plays an important role in shaping functional networks defined using the ca estimator, and leading to non-trivial differences in their connectivity structure. We provide an open source R package and equivalent Python implementation to facilitate the use of the new estimators, together with preprocessed rat time-series.

Comparing the efficacy of data-driven denoising methods for a multi-echo fMRI acquisition at 7T

In functional magnetic resonance imaging (fMRI) of the brain the measured signal is corrupted by several (e.g. physiological, motion, and thermal) noise sources and depends on the image acquisition. Imaging at ultrahigh field strength is becoming increasingly popular as it offers increased spatial accuracy. The latter is of particular benefit in brainstem neuroimaging given the small cross-sectional area of most nuclei. However, physiological noise scales with field strength in fMRI acquisitions. Although this problem is in part solved by decreasing voxel size, it is clear that adequate physiological denoising is of utmost importance in brainstem-focused fMRI experiments. Multi-echo sequences have been reported to facilitate highly effective denoising through TE-dependence of Blood Oxygen Level Dependent (BOLD) signals, in a denoising method referred to as multi-echo independent component analysis (ME-ICA). It has not been explored previously how ME-ICA compares to other data-driven denoising approaches at ultrahigh field strength. In the current study, we compared the efficacy of several denoising methods, including anatomical component based correction (aCompCor), Automatic Removal of Motion Artifacts (ICA-AROMA) aggressive and non-aggressive options, ME-ICA, and a combination of ME-ICA and aCompCor. We assessed several data quality metrics, including temporal signal-to-noise ratio (tSNR), delta variation signal (DVARS), spectral density of the global signal, functional connectivity and Shannon spectral entropy. Moreover, we looked at the ability of each method to uncouple the global signal and respiration. In line with previous reports at lower field strengths, we demonstrate that after applying ME-ICA, the data is best post-processed in order to remove spatially diffuse noise with a method such as aCompCor. Our findings indicate that ME-ICA combined with aCompCor and the aggressive option of ICA-AROMA are highly effective denoising approaches for multi-echo data acquired at 7T. ME-ICA combined with aCompCor potentially preserves more signal-of-interest as compared to the aggressive option of ICA-AROMA.

Exploiting Information in Event-Related Brain Potentials from Average Temporal Waveform, Time-Frequency Representation, and Phase Dynamics

Characterizing the brain's dynamic pattern of response to an input in electroencephalography (EEG) is not a trivial task due to the entanglement of the complex spontaneous brain activity. In this context, the brain's response can be defined as (1) the additional neural activity components generated after the input or (2) the changes in the ongoing spontaneous activities induced by the input. Moreover, the response can be manifested in multiple features. Three commonly studied examples of features are (1) transient temporal waveform, (2) time-frequency representation, and (3) phase dynamics. The most extensively used method of average event-related potentials (ERPs) captures the first one, while the latter two and other more complex features are attracting increasing attention. However, there has not been much work providing a systematic illustration and guidance for how to effectively exploit multifaceted features in neural cognitive research. Based on a visual oddball ERPs dataset with 200 participants, this work demonstrates how the information from the above-mentioned features are complementary to each other and how they can be integrated based on stereotypical neural-network-based machine learning approaches to better exploit neural dynamic information in basic and applied cognitive research.

Reorganization of sensorimotor representations of the intact limb after upper but not lower limb traumatic amputation

It is becoming increasingly clear that limb loss induces wider spread reorganization of representations of the body that are nonadjacent to the affected cortical territory. Data from upper extremity amputees reveal intrusion of the representation of the ipsilateral intact limb into the former hand territory. Here we test for the first time whether this reorganization of the intact limb into the deprived cortex is specific to the neurological organization of the upper limbs or reflects large scale adaptation that is triggered by any unilateral amputation. BOLD activity was measured as human subjects with upper limb and lower limb traumatic amputation and their controls moved the toes on each foot, open and closed each hand and pursed their lips. Subjects with amputation were asked to imagine moving the missing limb while remaining still. Bayesian pattern component modeling of fMRI data showed that intact ipsilateral movements and contralateral movements of the hand and foot were distinctly represented in the deprived sensorimotor cortex years after upper limb amputation. In contrast, there was evidence reminiscent of contralateral specificity for hand and foot movements following lower limb amputation, like that seen in controls. We propose the cortical reorganization of the intact limb to be a function of use-dependent plasticity that is more specific to the consequence of upper limb loss of forcing an asymmetric reliance on the intact hand and arm. The contribution of this reorganization to phantom pain or a heightened risk of overuse and resultant maladaptive plasticity needs investigating before targeting such reorganization in intervention.

Model connectivity: leveraging the power of encoding models to overcome the limitations of functional connectivity

Functional connectivity (FC) is the most popular method for recovering functional networks of brain areas with fMRI. However, because FC is defined as temporal correlations in brain activity, FC networks are confounded by noise and lack a precise functional role. To overcome these limitations, we developed model connectivity (MC). MC is defined as similarities in encoding model weights, which quantify reliable functional activity in terms of interpretable stimulus- or task-related features. To compare FC and MC, both methods were applied to a naturalistic story listening dataset. FC recovered spatially broad networks that are confounded by noise, and that lack a clear role during natural language comprehension. By contrast, MC recovered spatially localized networks that are robust to noise, and that represent distinct categories of semantic concepts. Thus, MC is a powerful data-driven approach for recovering and interpreting the functional networks that support complex cognitive processes.

Glioblastoma induces whole-brain spectral change in resting state fMRI: Associations with clinical comorbidities and overall survival

Glioblastoma, a highly aggressive form of brain tumor, is a brain-wide disease. We evaluated the impact of tumor burden on whole brain resting-state functional magnetic resonance imaging (rs-fMRI) activity. Specifically, we analyzed rs-fMRI signals in the temporal frequency domain in terms of the power-law exponent and fractional amplitude of low-frequency fluctuations (fALFF). We contrasted 189 patients with newly-diagnosed glioblastoma versus 189 age-matched healthy reference participants from an external dataset. The patient and reference datasets were matched for age and head motion. The principal finding was markedly flatter spectra and reduced grey matter fALFF in the patients as compared to the reference dataset. We posit that the whole-brain spectral change is attributable to global dysregulation of excitatory and inhibitory balance and metabolic demand in the tumor-bearing brain. Additionally, we observed that clinical comorbidities, in particular, seizures, and MGMT promoter methylation, were associated with flatter spectra. Notably, the degree of change in spectra was predictive of overall survival. Our findings suggest that frequency domain analysis of rs-fMRI activity provides prognostic information in glioblastoma patients and offers a means of noninvasively studying the effects of glioblastoma on the whole brain.

The arrow-of-time in neuroimaging time series identifies causal triggers of brain function

Moving from association to causal analysis of neuroimaging data is crucial to advance our understanding of brain function. The arrow-of-time (AoT), that is, the known asymmetric nature of the passage of time, is the bedrock of causal structures shaping physical phenomena. However, almost all current time series metrics do not exploit this asymmetry, probably due to the difficulty to account for it in modeling frameworks. Here, we introduce an AoT-sensitive metric that captures the intensity of causal effects in multivariate time series, and apply it to high-resolution functional neuroimaging data. We find that causal effects underlying brain function are more distinctively localized in space and time than functional activity or connectivity, thereby allowing us to trace neural pathways recruited in different conditions. Overall, we provide a mapping of the causal brain that challenges the association paradigm of brain function.

Spatial Smoothing Effect on Group-Level Functional Connectivity during Resting and Task-Based fMRI

Spatial smoothing is a preprocessing step applied to neuroimaging data to enhance data quality by reducing noise and artifacts. However, selecting an appropriate smoothing kernel size can be challenging as it can lead to undesired alterations in final images and functional connectivity networks. However, there is no sufficient information about the effects of the Gaussian kernel size on group-level results for different cases yet. This study investigates the influence of kernel size on functional connectivity networks and network parameters in whole-brain rs-fMRI and tb-fMRI analyses of healthy adults. The analysis includes {0, 2, 4, 6, 8, 10} mm kernels, commonly used in practical analyses, covering all major brain networks. Graph theoretical measures such as betweenness centrality, global/local efficiency, clustering coefficient, and average path length are examined for each kernel. Additionally, principal component analysis (PCA) and independent component analysis (ICA) parameters, namely kurtosis and skewness, are evaluated for the functional images. The findings demonstrate that kernel size directly affects node connections, resulting in modifications to functional network structures and PCA/ICA parameters. However, network metrics exhibit greater resilience to these changes.

Characterizing spatiotemporal population receptive fields in human visual cortex with fMRI

The use of fMRI and computational modeling has advanced understanding of spatial characteristics of population receptive fields (pRFs) in human visual cortex. However, we know relatively little about the spatiotemporal characteristics of pRFs because neurons' temporal properties are one to two orders of magnitude faster than fMRI BOLD responses. Here, we developed an image-computable framework to estimate spatiotemporal pRFs from fMRI data. First, we developed a simulation software that predicts fMRI responses to a time varying visual input given a spatiotemporal pRF model and solves the model parameters. The simulator revealed that ground-truth spatiotemporal parameters can be accurately recovered at the millisecond resolution from synthesized fMRI responses. Then, using fMRI and a novel stimulus paradigm, we mapped spatiotemporal pRFs in individual voxels across human visual cortex in 10 participants. We find that a compressive spatiotemporal (CST) pRF model better explains fMRI responses than a conventional spatial pRF model across visual areas spanning the dorsal, lateral, and ventral streams. Further, we find three organizational principles of spatiotemporal pRFs: (i) from early to later areas within a visual stream, spatial and temporal integration windows of pRFs progressively increase in size and show greater compressive nonlinearities, (ii) later visual areas show diverging spatial and temporal integration windows across streams, and (iii) within early visual areas (V1-V3), both spatial and temporal integration windows systematically increase with eccentricity. Together, this computational framework and empirical results open exciting new possibilities for modeling and measuring fine-grained spatiotemporal dynamics of neural responses in the human brain using fMRI.

Hemodynamic timing in resting-state and breathing-task BOLD fMRI

The blood flow response to a vasoactive stimulus demonstrates regional heterogeneity across both the healthy brain and in cerebrovascular pathology. The timing of a regional hemodynamic response is emerging as an important biomarker of cerebrovascular dysfunction, as well as a confound within fMRI analyses. Previous research demonstrated that hemodynamic timing is more robustly characterized when a larger systemic vascular response is evoked by a breathing challenge, compared to when only spontaneous fluctuations in vascular physiology are present (i.e., in resting-state data). However, it is not clear whether hemodynamic delays in these two conditions are physiologically interchangeable, and how methodological signal-to-noise factors may limit their agreement. To address this, we generated whole-brain maps of hemodynamic delays in nine healthy adults. We assessed the agreement of voxel-wise gray matter (GM) hemodynamic delays between two conditions: resting-state and breath-holding. We found that delay values demonstrated poor agreement when considering all GM voxels, but increasingly greater agreement when limiting analyses to voxels showing strong correlation with the GM mean time-series. Voxels showing the strongest agreement with the GM mean time-series were primarily located near large venous vessels, however these voxels explain some, but not all, of the observed agreement in timing. Increasing the degree of spatial smoothing of the fMRI data enhanced the correlation between individual voxel time-series and the GM mean time-series. These results suggest that signal-to-noise factors may be limiting the accuracy of voxel-wise timing estimates and hence their agreement between the two data segments. In conclusion, caution must be taken when using voxel-wise delay estimates from resting-state and breathing-task data interchangeably, and additional work is needed to evaluate their relative sensitivity and specificity to aspects of vascular physiology and pathology.

Less is more: balancing noise reduction and data retention in fMRI with data-driven scrubbing

Functional MRI (fMRI) data may be contaminated by artifacts arising from a myriad of sources, including subject head motion, respiration, heartbeat, scanner drift, and thermal noise. These artifacts cause deviations from common distributional assumptions, introduce spatial and temporal outliers, and reduce the signal-to-noise ratio of the data-all of which can have negative consequences for the accuracy and power of downstream statistical analysis. Scrubbing is a technique for excluding fMRI volumes thought to be contaminated by artifacts and generally comes in two flavors. Motion scrubbing based on subject head motion-derived measures is popular but suffers from a number of drawbacks, among them the need to choose a threshold, a lack of generalizability to multiband acquisitions, and high rates of censoring of individual volumes and entire subjects. Alternatively, data-driven scrubbing methods like DVARS are based on observed noise in the processed fMRI timeseries and may avoid some of these issues. Here we propose "projection scrubbing", a novel data-driven scrubbing method based on a statistical outlier detection framework and strategic dimension reduction, including independent component analysis (ICA), to isolate artifactual variation. We undertake a comprehensive comparison of motion scrubbing with data-driven projection scrubbing and DVARS. We argue that an appropriate metric for the success of scrubbing is maximal data retention subject to reasonable performance on typical benchmarks such as the validity, reliability, and identifiability of functional connectivity. We find that stringent motion scrubbing yields worsened validity, worsened reliability, and produced small improvements to fingerprinting. Meanwhile, data-driven scrubbing methods tend to yield greater improvements to fingerprinting while not generally worsening validity or reliability. Importantly, however, data-driven scrubbing excludes a fraction of the number of volumes or entire sessions compared to motion scrubbing. The ability of data-driven fMRI scrubbing to improve data retention without negatively impacting the quality of downstream analysis has major implications for sample sizes in population neuroscience research.

MP-PCA denoising of fMRI time-series data can lead to artificial activation "spreading"

MP-PCA denoising has become the method of choice for denoising MRI data since it provides an objective threshold to separate the signal components from unwanted thermal noise components. In rodents, thermal noise in the coils is an important source of noise that can reduce the accuracy of activation mapping in fMRI. Further confounding this problem, vendor data often contains zero-filling and other post-processing steps that may violate MP-PCA assumptions. Here, we develop an approach to denoise vendor data and assess activation "spreading" caused by MP-PCA denoising in rodent task-based fMRI data. Data was obtained from N = 3 mice using conventional multislice and ultrafast acquisitions (1 s and 50 ms temporal resolution, respectively), during visual stimulation. MP-PCA denoising produced SNR gains of 64% and 39% and Fourier Spectral Amplitude (FSA) increases in BOLD maps of 9% and 7% for multislice and ultrafast data, respectively, when using a small [2 2] denoising window. Larger windows provided higher SNR and FSA gains with increased spatial extent of activation that may or may not represent real activation. Simulations showed that MP-PCA denoising can incur activation "spreading" with increased false positive rate and smoother functional maps due to local "bleeding" of principal components, and that the optimal denoising window for improved specificity of functional mapping, based on Dice score calculations, depends on the data's tSNR and functional CNR. This "spreading" effect applies also to another recently proposed low-rank denoising method (NORDIC), although to a lesser degree. Our results bode well for enhancing spatial and/or temporal resolution in future fMRI work, while taking into account the sensitivity/specificity trade-offs of low-rank denoising methods.

The art and science of using quality control to understand and improve fMRI data

Designing and executing a good quality control (QC) process is vital to robust and reproducible science and is often taught through hands on training. As FMRI research trends toward studies with larger sample sizes and highly automated processing pipelines, the people who analyze data are often distinct from those who collect and preprocess the data. While there are good reasons for this trend, it also means that important information about how data were acquired, and their quality, may be missed by those working at later stages of these workflows. Similarly, an abundance of publicly available datasets, where people (not always correctly) assume others already validated data quality, makes it easier for trainees to advance in the field without learning how to identify problematic data. This manuscript is designed as an introduction for researchers who are already familiar with fMRI, but who did not get hands on QC training or who want to think more deeply about QC. This could be someone who has analyzed fMRI data but is planning to personally acquire data for the first time, or someone who regularly uses openly shared data and wants to learn how to better assess data quality. We describe why good QC processes are important, explain key priorities and steps for fMRI QC, and as part of the FMRI Open QC Project, we demonstrate some of these steps by using AFNI software and AFNI’s QC reports on an openly shared dataset. A good QC process is context dependent and should address whether data have the potential to answer a scientific question, whether any variation in the data has the potential to skew or hide key results, and whether any problems can potentially be addressed through changes in acquisition or data processing. Automated metrics are essential and can often highlight a possible problem, but human interpretation at every stage of a study is vital for understanding causes and potential solutions.

Functional connectivity MRI quality control procedures in CONN

Quality control (QC) for functional connectivity magnetic resonance imaging (FC-MRI) is critical to ensure the validity of neuroimaging studies. Noise confounds are common in MRI data and, if not accounted for, may introduce biases in functional measures affecting the validity, replicability, and interpretation of FC-MRI study results. Although FC-MRI analysis rests on the assumption of adequate data processing, QC is underutilized and not systematically reported. Here, we describe a quality control pipeline for the visual and automated evaluation of MRI data implemented as part of the CONN toolbox. We analyzed publicly available resting state MRI data (N = 139 from 7 MRI sites) from the FMRI Open QC Project. Preprocessing steps included realignment, unwarp, normalization, segmentation, outlier identification, and smoothing. Data denoising was performed based on the combination of scrubbing, motion regression, and aCompCor - a principal component characterization of noise from minimally eroded masks of white matter and of cerebrospinal fluid tissues. Participant-level QC procedures included visual inspection of raw-level data and of representative images after each preprocessing step for each run, as well as the computation of automated descriptive QC measures such as average framewise displacement, average global signal change, prevalence of outlier scans, MNI to anatomical and functional overlap, anatomical to functional overlap, residual BOLD timeseries variability, effective degrees of freedom, and global correlation strength. Dataset-level QC procedures included the evaluation of inter-subject variability in the distributions of edge connectivity in a 1,000-node graph (FC distribution displays), and the estimation of residual associations across participants between functional connectivity strength and potential noise indicators such as participant's head motion and prevalence of outlier scans (QC-FC analyses). QC procedures are demonstrated on the reference dataset with an emphasis on visualization, and general recommendations for best practices are discussed in the context of functional connectivity and other fMRI analysis. We hope this work contributes toward the dissemination and standardization of QC testing performance reporting among peers and in scientific journals.

Comparing end-tidal CO2, respiration volume per time (RVT), and average gray matter signal for mapping cerebrovascular reactivity amplitude and delay with breath-hold task BOLD fMRI

Cerebrovascular reactivity (CVR), defined as the cerebral blood flow response to a vasoactive stimulus, is an imaging biomarker with demonstrated utility in a range of diseases and in typical development and aging processes. A robust and widely implemented method to map CVR involves using a breath-hold task during a BOLD fMRI scan. Recording end-tidal CO₂ (P_ETCO₂) changes during the breath-hold task is recommended to be used as a reference signal for modeling CVR amplitude in standard units (%BOLD/mmHg) and CVR delay in seconds. However, obtaining reliable P_ETCO₂ recordings requires equipment and task compliance that may not be achievable in all settings. To address this challenge, we investigated two alternative reference signals to map CVR amplitude and delay in a lagged general linear model (lagged-GLM) framework: respiration volume per time (RVT) and average gray matter BOLD response (GM-BOLD). In 8 healthy adults with multiple scan sessions, we compare spatial agreement of CVR maps from RVT and GM-BOLD to those generated with P_ETCO₂. We define a threshold to determine whether a P_ETCO₂ recording has "sufficient" quality for CVR mapping and perform these comparisons in 16 datasets with sufficient P_ETCO₂ and 6 datasets with insufficient P_ETCO₂. When P_ETCO₂ quality is sufficient, both RVT and GM-BOLD produce CVR amplitude maps that are nearly identical to those from P_ETCO₂ (after accounting for differences in scale), with the caveat they are not in standard units to facilitate between-group comparisons. CVR delays are comparable to P_ETCO₂ with an RVT regressor but may be underestimated with the average GM-BOLD regressor. Importantly, when P_ETCO₂ quality is insufficient, RVT and GM-BOLD CVR recover reasonable CVR amplitude and delay maps, provided the participant attempted the breath-hold task. Therefore, our framework offers a solution for achieving high quality CVR maps in both retrospective and prospective studies where sufficient P_ETCO₂ recordings are not available and especially in populations where obtaining reliable measurements is a known challenge (e.g., children). Our results have the potential to improve the accessibility of CVR mapping and to increase the prevalence of this promising metric of vascular health.

Decoding semantic representations in mind and brain

A key goal for cognitive neuroscience is to understand the neurocognitive systems that support semantic memory. Recent multivariate analyses of neuroimaging data have contributed greatly to this effort, but the rapid development of these novel approaches has made it difficult to track the diversity of findings and to understand how and why they sometimes lead to contradictory conclusions. We address this challenge by reviewing cognitive theories of semantic representation and their neural instantiation. We then consider contemporary approaches to neural decoding and assess which types of representation each can possibly detect. The analysis suggests why the results are heterogeneous and identifies crucial links between cognitive theory, data collection, and analysis that can help to better connect neuroimaging to mechanistic theories of semantic cognition.

Reproducibility of cerebral perfusion measurements using BOLD delay

BOLD delay is an emerging, noninvasive method for assessing cerebral perfusion that does not require the use of intravenous contrast agents and is thus particularly suited for longitudinal monitoring. In this study, we assess the reproducibility of BOLD delay using data from 136 subjects with normal cerebral perfusion scanned on two separate occasions with scanners, sequence parameters, and intervals between scans varying between subjects. The effects of various factors on the reproducibility of BOLD delay, defined here as the differences in BOLD delay values between the scanning sessions, were investigated using a linear mixed model. Reproducibility was additionally assessed using the intraclass correlation coefficient of BOLD delay between sessions. Reproducibility was highest in the posterior cerebral artery territory. The mean BOLD delay test-retest difference after accounting for the aforementioned factors was 1.2 s (95% CI = 1.0 to 1.4 s). Overall, BOLD delay shows good reproducibility, but care should be taken when interpreting longitudinal BOLD delay changes that are either very small or are located in certain brain regions.

Network organization of resting-state cerebral hemodynamics and their aliasing contributions measured by functional near-infrared spectroscopy

Objective. Spontaneous fluctuations of cerebral hemodynamics measured by functional magnetic resonance imaging (fMRI) are widely used to study the network organization of the brain. The temporal correlations among the ultra-slow, <0.1 Hz fluctuations across the brain regions are interpreted as functional connectivity maps and used for diagnostics of neurological disorders. However, despite the interest narrowed in the ultra-slow fluctuations, hemodynamic activity that exists beyond the ultra-slow frequency range could contribute to the functional connectivity, which remains unclear.Approach. In the present study, we have measured the brain-wide hemodynamics in the human participants with functional near-infrared spectroscopy (fNIRS) in a whole-head, cap-based and high-density montage at a sampling rate of 6.25 Hz. In addition, we have acquired resting state fMRI scans in the same group of participants for cross-modal evaluation of the connectivity maps. Then fNIRS data were deliberately down-sampled to a typical fMRI sampling rate of ∼0.5 Hz and the resulted differential connectivity maps were subject to a k-means clustering.Main results. Our diffuse optical topographical analysis of fNIRS data have revealed a default mode network (DMN) in the spontaneous deoxygenated and oxygenated hemoglobin changes, which remarkably resemble the same fMRI network derived from participants. Moreover, we have shown that the aliased activities in the down-sampled optical signals have altered the connectivity patterns, resulting in a network organization of aliased functional connectivity in the cerebral hemodynamics.Significance.The results have for the first time demonstrated that fNIRS as a broadly accessible modality can image the resting-state functional connectivity in the posterior midline, prefrontal and parietal structures of the DMN in the human brain, in a consistent pattern with fMRI. Further empowered by the fast sampling rate of fNIRS, our findings suggest the presence of aliased connectivity in the current understanding of the human brain organization.

Robust high spatio-temporal line-scanning fMRI in humans at 7T using multi-echo readouts, denoising and prospective motion correction

BACKGROUND

Functional magnetic resonance imaging (fMRI), typically using blood oxygenation level-dependent (BOLD) contrast weighted imaging, allows the study of brain function with millimeter spatial resolution and temporal resolution of one to a few seconds. At a mesoscopic scale, neurons in the human brain are spatially organized in structures with dimensions of hundreds of micrometers, while they communicate at the millisecond timescale. For this reason, it is important to develop an fMRI method with simultaneous high spatial and temporal resolution. Line-scanning promises to reach this goal at the cost of volume coverage.

NEW METHOD

Here, we release a comprehensive update to human line-scanning fMRI. First, we investigated multi-echo line-scanning with five different protocols varying the number of echoes and readout bandwidth while keeping the TR constant. In these, we compared different echo combination approaches in terms of BOLD activation (sensitivity) and temporal signal-to-noise ratio. Second, we implemented an adaptation of NOise reduction with DIstribution Corrected principal component analysis (NORDIC) thermal noise removal for line-scanning fMRI data. Finally, we tested three image-based navigators for motion correction and investigated different ways of performing fMRI analysis on the timecourses which were influenced by the insertion of the navigators themselves.

RESULTS

The presented improvements are relatively straightforward to implement; multi-echo readout and NORDIC denoising together, significantly improve data quality in terms of tSNR and t-statistical values, while motion correction makes line-scanning fMRI more robust.

COMPARISON WITH EXISTING METHODS

Multi-echo acquisitions and denoising have previously been applied in 3D magnetic resonance imaging. Their combination and application to 1D line-scanning is novel. The current proposed method greatly outperforms the previous line-scanning acquisitions with single-echo acquisition, in terms of tSNR (4.0 for single-echo line-scanning and 36.2 for NORDIC-denoised multi-echo) and t-statistical values (3.8 for single-echo line-scanning and 25.1 for NORDIC-denoised multi-echo line-scanning).

CONCLUSIONS

Line-scanning fMRI was advanced compared to its previous implementation in order to improve sensitivity and reliability. The improved line-scanning acquisition could be used, in the future, for neuroscientific and clinical applications.

From sMRI to task-fMRI: A unified geometric deep learning framework for cross-modal brain anatomo-functional mapping

Achieving predictions of brain functional activation patterns/task-fMRI maps from its underlying anatomy is an important yet challenging problem. Once successful, it will not only open up new ways to understand how brain anatomy influences functional organization of the brain, but also provide new technical support for the clinical use of anatomical information to guide the localization of cortical functional areas. However, due to the non-Euclidean complex architecture of brain anatomy and the inherent low signal-to-noise ratio (SNR) properties of fMRI signals, the key challenge in building such a cross-modal brain anatomo-functional mapping is how to effectively learn the context-aware information of brain anatomy and overcome the interference of noise-containing task-fMRI labels on the learning process. In this work, we propose a Unified Geometric Deep Learning framework (BrainUGDL) to perform the cross-modal brain anatomo-functional mapping task. Considering that both global and local structures of brain anatomy have an impact on brain functions from their respective perspectives, we innovatively propose the novel Global Graph Encoding (GGE) unit and Local Graph Attention (LGA) unit embedded into two parallel branches, focusing on learning the high-level global and local context information, respectively. Specifically, GGE learns the global context information of each mesh vertex by building and encoding global interactions, and LGA learns the local context information of each mesh vertex by selectively aggregating patch structure enhanced features from its spatial neighbors. The information learnt from the two branches is then fused to form a comprehensive representation of brain anatomical features for final brain function predictions. To address the inevitable measurement noise in task-fMRI labels, we further elaborate a novel uncertainty-filtered learning mechanism, which enables BrainUGDL to realize revised learning from the noise-containing labels through the estimated uncertainty. Experiments across seven open task-fMRI datasets from human connectome project (HCP) demonstrate the superiority of BrainUGDL. To our best knowledge, our proposed BrainUGDL is the first to achieve the prediction of individual task-fMRI maps solely based on brain sMRI data.

Task-based functional MRI challenges in clinical neuroscience: Choice of the best head motion correction approach in multiple sclerosis

Introduction

Functional MRI (fMRI) is commonly used for understanding brain organization and connectivity abnormalities in neurological conditions, and in particular in multiple sclerosis (MS). However, head motion degrades fMRI data quality and influences all image-derived metrics. Persistent controversies regarding the best correction strategy motivates a systematic comparison, including methods such as scrubbing and volume interpolation, to find optimal correction models, particularly in studies with clinical populations prone to characterize by high motion. Moreover, strategies for correction of motion effects gain more relevance in task-based designs, which are less explored compared to resting-state, have usually lower sample sizes, and may have a crucial role in describing the functioning of the brain and highlighting specific connectivity changes.

Methods

We acquired fMRI data from 17 early MS patients and 14 matched healthy controls (HC) during performance of a visual task, characterized motion in both groups, and quantitatively compared the most used and easy to implement methods for correction of motion effects. We compared task-activation metrics obtained from: (i) models containing 6 or 24 motion parameters (MPs) as nuisance regressors; (ii) models containing nuisance regressors for 6 or 24 MPs and motion outliers (scrubbing) detected with Framewise Displacement or Derivative or root mean square VARiance over voxelS; and (iii) models with 6 or 24 MPs and motion outliers corrected through volume interpolation. To our knowledge, volume interpolation has not been systematically compared with scrubbing, nor investigated in task fMRI clinical studies in MS.

Results

No differences in motion were found between groups, suggesting that recently diagnosed MS patients may not present problematic motion. In general, models with 6 MPs perform better than models with 24 MPs, suggesting the 6 MPs as the best trade-off between correction of motion effects and preservation of valuable information. Parsimonious models with 6 MPs and volume interpolation were the best combination for correcting motion in both groups, surpassing the scrubbing methods. A joint analysis regardless of the group further highlighted the value of volume interpolation.

Discussion

Volume interpolation of motion outliers is an easy to implement technique, which may be an alternative to other methods and may improve the accuracy of fMRI analyses, crucially in clinical studies in MS and other neurological populations.

Evaluating the efficacy of multi-echo ICA denoising on model-based fMRI

fMRI is an indispensable tool for neuroscience investigation, but this technique is limited by multiple sources of physiological and measurement noise. These noise sources are particularly problematic for analysis techniques that require high signal-to-noise ratio for stable model fitting, such as voxel-wise modeling. Multi-echo data acquisition in combination with echo-time dependent ICA denoising (ME-ICA) represents one promising strategy to mitigate physiological and hardware-related noise sources as well as motion-related artifacts. However, most studies employing ME-ICA to date are resting-state fMRI studies, and therefore we have a limited understanding of the impact of ME-ICA on complex task or model-based fMRI paradigms. Here, we addressed this knowledge gap by comparing data quality and model fitting performance of data acquired during a visual population receptive field (pRF) mapping (N = 13 participants) experiment after applying one of three preprocessing procedures: ME-ICA, optimally combined multi-echo data without ICA-denoising, and typical single echo processing. As expected, multi-echo fMRI improved temporal signal-to-noise compared to single echo fMRI, with ME-ICA amplifying the improvement compared to optimal combination alone. However, unexpectedly, this boost in temporal signal-to-noise did not directly translate to improved model fitting performance: compared to single echo acquisition, model fitting was only improved after ICA-denoising. Specifically, compared to single echo acquisition, ME-ICA resulted in improved variance explained by our pRF model throughout the visual system, including anterior regions of the temporal and parietal lobes where SNR is typically low, while optimal combination without ICA did not. ME-ICA also improved reliability of parameter estimates compared to single echo and optimally combined multi-echo data without ICA-denoising. Collectively, these results suggest that ME-ICA is effective for denoising task-based fMRI data for modeling analyzes and maintains the integrity of the original data. Therefore, ME-ICA may be beneficial for complex fMRI experiments, including voxel-wise modeling and naturalistic paradigms.

Improving the accuracy of single-trial fMRI response estimates using GLMsingle

Advances in artificial intelligence have inspired a paradigm shift in human neuroscience, yielding large-scale functional magnetic resonance imaging (fMRI) datasets that provide high-resolution brain responses to thousands of naturalistic visual stimuli. Because such experiments necessarily involve brief stimulus durations and few repetitions of each stimulus, achieving sufficient signal-to-noise ratio can be a major challenge. We address this challenge by introducing GLMsingle, a scalable, user-friendly toolbox available in MATLAB and Python that enables accurate estimation of single-trial fMRI responses (glmsingle.org). Requiring only fMRI time-series data and a design matrix as inputs, GLMsingle integrates three techniques for improving the accuracy of trial-wise general linear model (GLM) beta estimates. First, for each voxel, a custom hemodynamic response function (HRF) is identified from a library of candidate functions. Second, cross-validation is used to derive a set of noise regressors from voxels unrelated to the experiment. Third, to improve the stability of beta estimates for closely spaced trials, betas are regularized on a voxel-wise basis using ridge regression. Applying GLMsingle to the Natural Scenes Dataset and BOLD5000, we find that GLMsingle substantially improves the reliability of beta estimates across visually-responsive cortex in all subjects. Comparable improvements in reliability are also observed in a smaller-scale auditory dataset from the StudyForrest experiment. These improvements translate into tangible benefits for higher-level analyses relevant to systems and cognitive neuroscience. We demonstrate that GLMsingle: (i) helps decorrelate response estimates between trials nearby in time; (ii) enhances representational similarity between subjects within and across datasets; and (iii) boosts one-versus-many decoding of visual stimuli. GLMsingle is a publicly available tool that can significantly improve the quality of past, present, and future neuroimaging datasets sampling brain activity across many experimental conditions.

Unsupervised physiological noise correction of functional magnetic resonance imaging data using phase and magnitude information (PREPAIR)

Of the sources of noise affecting blood oxygen level-dependent functional magnetic resonance imaging (fMRI), respiration and cardiac fluctuations are responsible for the largest part of the variance, particularly at high and ultrahigh field. Existing approaches to removing physiological noise either use external recordings, which can be unwieldy and unreliable, or attempt to identify physiological noise from the magnitude fMRI data. Data-driven approaches are limited by sensitivity, temporal aliasing, and the need for user interaction. In the light of the sensitivity of the phase of the MR signal to local changes in the field stemming from physiological processes, we have developed an unsupervised physiological noise correction method using the information carried in the phase and the magnitude of echo-planar imaging data. Our technique, Physiological Regressor Estimation from Phase and mAgnItude, sub-tR (PREPAIR) derives time series signals sampled at the slice TR from both phase and magnitude images. It allows physiological noise to be captured without aliasing, and efficiently removes other sources of signal fluctuations not related to physiology, prior to regressor estimation. We demonstrate that the physiological signal time courses identified with PREPAIR agree well with those from external devices and retrieve challenging cardiac dynamics. The removal of physiological noise was as effective as that achieved with the most used approach based on external recordings, RETROICOR. In comparison with widely used recording-free physiological noise correction tools-PESTICA and FIX, both performed in unsupervised mode-PREPAIR removed significantly more respiratory and cardiac noise than PESTICA, and achieved a larger increase in temporal signal-to-noise-ratio at both 3 and 7 T.

Interpreting mental state decoding with deep learning models

In mental state decoding, researchers aim to identify the set of mental states (e.g., experiencing happiness or fear) that can be reliably identified from the activity patterns of a brain region (or network). Deep learning (DL) models are highly promising for mental state decoding because of their unmatched ability to learn versatile representations of complex data. However, their widespread application in mental state decoding is hindered by their lack of interpretability, difficulties in applying them to small datasets, and in ensuring their reproducibility and robustness. We recommend approaching these challenges by leveraging recent advances in explainable artificial intelligence (XAI) and transfer learning, and also provide recommendations on how to improve the reproducibility and robustness of DL models in mental state decoding.

Increasing diversity in developmental cognitive neuroscience: A roadmap for increasing representation in pediatric neuroimaging research

Understanding of human brain development has advanced rapidly as the field of developmental cognitive neuroscience (DCN) has matured into an established scientific discipline. Despite substantial progress, DCN lags behind other related disciplines in terms of diverse representation, standardized reporting requirements for socio-demographic characteristics of participants in pediatric neuroimaging studies, and use of intentional sampling strategies to more accurately represent the socio-demographic, ethnic, and racial composition of the populations from which participants are sampled. Additional efforts are needed to shift DCN towards a more inclusive field that facilitates the study of individual differences across a variety of cultural and contextual experiences. In this commentary, we outline and discuss barriers within our current scientific practice (e.g., research methods) and beliefs (i.e., what constitutes good science, good scientists, and good research questions) that contribute to under-representation and limited diversity within pediatric neuroimaging studies and propose strategies to overcome those barriers. We discuss strategies to address barriers at intrapersonal, interpersonal, community, systemic, and structural levels. Highlighting strength-based models of inclusion and recognition of the value of diversity in DCN research, along with acknowledgement of the support needed to diversify the field is critical for advancing understanding of neurodevelopment and reducing health inequities.

Neural underpinning of a respiration-associated resting-state fMRI network

Respiration can induce motion and CO2 fluctuation during resting-state fMRI (rsfMRI) scans, which will lead to non-neural artifacts in the rsfMRI signal. In the meantime, as a crucial physiologic process, respiration can directly drive neural activity change in the brain, and may thereby modulate the rsfMRI signal. Nonetheless, this potential neural component in the respiration-fMRI relationship is largely unexplored. To elucidate this issue, here we simultaneously recorded the electrophysiology, rsfMRI, and respiration signals in rats. Our data show that respiration is indeed associated with neural activity changes, evidenced by a phase-locking relationship between slow respiration variations and the gamma-band power of the electrophysiological signal recorded in the anterior cingulate cortex. Intriguingly, slow respiration variations are also linked to a characteristic rsfMRI network, which is mediated by gamma-band neural activity. In addition, this respiration-related brain network disappears when brain-wide neural activity is silenced at an isoelectrical state, while the respiration is maintained, further confirming the necessary role of neural activity in this network. Taken together, this study identifies a respiration-related brain network underpinned by neural activity, which represents a novel component in the respiration-rsfMRI relationship that is distinct from respiration-related rsfMRI artifacts. It opens a new avenue for investigating the interactions between respiration, neural activity, and resting-state brain networks in both healthy and diseased conditions.

Evaluation of noise regression techniques in resting-state fMRI studies using data of 434 older adults

Subject motion is a well-known confound in resting-state functional MRI (rs-fMRI) and the analysis of functional connectivity. Consequently, several clean-up strategies have been established to minimize the impact of subject motion. Physiological signals in response to cardiac activity and respiration are also known to alter the apparent rs-fMRI connectivity. Comprehensive comparisons of common noise regression techniques showed that the "Independent Component Analysis based strategy for Automatic Removal of Motion Artifacts" (ICA-AROMA) was a preferred pre-processing technique for teenagers and adults. However, motion and physiological noise characteristics may differ substantially for older adults. Here, we present a comprehensive comparison of noise-regression techniques for older adults from a large multi-site clinical trial of exercise and intensive pharmacological vascular risk factor reduction. The Risk Reduction for Alzheimer's Disease (rrAD) trial included hypertensive older adults (60-84 years old) at elevated risk of developing Alzheimer's Disease (AD). We compared the performance of censoring, censoring combined with global signal regression, non-aggressive and aggressive ICA-AROMA, as well as the Spatially Organized Component Klassifikator (SOCK) on the rs-fMRI baseline scans from 434 rrAD subjects. All techniques were rated based on network reproducibility, network identifiability, edge activity, spatial smoothness, and loss of temporal degrees of freedom (tDOF). We found that non-aggressive ICA-AROMA did not perform as well as the other four techniques, which performed table with marginal differences, demonstrating the validity of these techniques. Considering reproducibility as the most important factor for longitudinal studies, given low false-positive rates and a better preserved, more cohesive temporal structure, currently aggressive ICA-AROMA is likely the most suitable noise regression technique for rs-fMRI studies of older adults.

Pitfalls and Recommended Strategies and Metrics for Suppressing Motion Artifacts in Functional MRI

In resting-state functional magnetic resonance imaging (rs-fMRI), artefactual signals arising from subject motion can dwarf and obfuscate the neuronal activity signal. Typical motion correction approaches involve the generation of nuisance regressors, which are timeseries of non-brain signals regressed out of the fMRI timeseries to yield putatively artifact-free data. Recent work suggests that concatenating all regressors into a single regression model is more effective than the sequential application of individual regressors, which may reintroduce previously removed artifacts. This work compares 18 motion correction pipelines consisting of head motion, independent components analysis, and non-neuronal physiological signal regressors in sequential or concatenated combinations. The pipelines are evaluated on a dataset of cognitively normal individuals with repeat imaging and on datasets of studies of Autism Spectrum Disorder, Major Depressive Disorder, and Parkinson's Disease. Extensive metrics of motion artifact removal are measured, including resting state network recovery, Quality Control-Functional Connectivity (QC-FC) correlation, distance-dependent artifact, network modularity, and test-retest reliability of multiple rs-fMRI analyses. The results reveal limitations in previously proposed metrics, including the QC-FC correlation and modularity quality, and identify more robust artifact removal metrics. The results also reveal limitations in the concatenated regression approach, which is outperformed by the sequential regression approach in the test-retest reliability metrics. Finally, pipelines are recommended that perform well based on quantitative and qualitative comparisons across multiple datasets and robust metrics. These new insights and recommendations help address the need for effective motion artifact correction to reduce noise and confounds in rs-fMRI.

The optimized combination of aCompCor and ICA-AROMA to reduce motion and physiologic noise in task fMRI data

One of the main challenges in fMRI processing is filtering the task BOLD signals from the noise. Independent component analysis with automatic removal of motion artifacts (ICA-AROMA) reduces motion artifacts by identifying ICA noise components based on their location at the brain edges and cerebrospinal fluid (CSF), high frequency content and correlation with motion regressors. In anatomical component correction (aCompCor), physiological noise regressors extracted from CSF were regressed out from the fMRI time series. In this study, we compared three methods to combine aCompCor and ICA-AROMA denoising in one denoising step. In the first analysis, we regressed the temporal signals of the ICA components identified as noise by ICA-AROMA together with the noise signals determined by aCompCor from the fMRI signals. For the second and third analyses, the correlation between the temporal signals of the ICA components and the aCompCor noise signals was used as an additional criterion to identify the noise components. In the second analysis, the temporal signals of the ICA components classified as noise were regressed from the fMRI signals. In the third analysis, the noise components were removed. To compare the denoising strategies, we examined the fractional amplitude of low-frequency fluctuations (fALFF) and the overlap between the contrast maps. Our results revealed that including the aCompCor noise signals as regressors in ICA-AROMA resulted in more correctly identified noise components, higher fALFF values, and larger activation maps. Moreover, combining the temporal signals of the noise components identified by ICA-AROMA with the aCompCor signals in a noise regression matrix resulted in deactivations. These results suggest that using the correlation between the ICA component temporal signals and the aCompCor signals as noise identification criteria in ICA-AROMA is the best approach for combining both denoising methods.

Connecting the dots between cell populations, whole-brain activity, and behavior

Simultaneously manipulating and monitoring both microscopic and macroscopic brain activity in vivo and identifying the linkage to behavior are powerful tools in neuroscience research. These capabilities have been realized with the recent technical advances of optogenetics and its combination with fMRI, here termed "opto-fMRI." Opto-fMRI allows for targeted brain region-, cell-type-, or projection-specific manipulation and targeted Ca 2 + activity measurement to be linked with global brain signaling and behavior. We cover the history, technical advances, applications, and important considerations of opto-fMRI in anesthetized and awake rodents and the future directions of the combined techniques in neuroscience and neuroimaging.

Controlling for Spurious Nonlinear Dependence in Connectivity Analyses

Recent analysis methods can capture nonlinear interactions between brain regions. However, noise sources might induce spurious nonlinear relationships between the responses in different regions. Previous research has demonstrated that traditional denoising techniques effectively remove noise-induced linear relationships between brain areas, but it is unknown whether these techniques can remove spurious nonlinear relationships. To address this question, we analyzed fMRI responses while participants watched the film Forrest Gump. We tested whether nonlinear Multivariate Pattern Dependence Networks (MVPN) outperform linear MVPN in non-denoised data, and whether this difference is reduced after CompCor denoising. Whereas nonlinear MVPN outperformed linear MVPN in the non-denoised data, denoising removed these nonlinear interactions. We replicated our results using different neural network architectures as the bases of MVPN, different activation functions (ReLU and sigmoid), different dimensionality reduction techniques for CompCor (PCA and ICA), and multiple datasets, demonstrating that CompCor's ability to remove nonlinear interactions is robust across these analysis choices and across different groups of participants. Finally, we asked whether information contributing to the removal of nonlinear interactions is localized to specific anatomical regions of no interest or to specific principal components. We denoised the data 8 separate times by regressing out 5 principal components extracted from combined white matter (WM) and cerebrospinal fluid (CSF), each of the 5 components separately, 5 components extracted from WM only, and 5 components extracted solely from CSF. In all cases, denoising was sufficient to remove the observed nonlinear interactions.

A Geometric View of Signal Sensitivity Metrics in multi-echo fMRI

In multi-echo fMRI (ME-fMRI), two metrics have been widely used to measure the performance of various acquisition and analysis approaches. These are temporal SNR (tSNR) and differential contrast-to-noise ratio (dCNR). A key step in ME-fMRI is the weighted combination of the data from multiple echoes, and prior work has examined the dependence of tSNR and dCNR on the choice of weights. However, most studies have focused on only one of these two metrics, and the relationship between the two metrics has not been examined. In this work, we present a geometric view that offers greater insight into the relation between the two metrics and their weight dependence. We identify three major regimes: (1) a tSNR robust regime in which tSNR is robust to the weight selection with most weight variants achieving close to optimal performance, whereas dCNR shows a pronounced dependence on the weights with most variants achieving suboptimal performance; (2) a dCNR robust regime in which dCNR is robust to the weight selection with most weight variants achieving close to optimal performance, while tSNR exhibits a strong dependence on the weights with most variants achieving significantly lower than optimal performance; and (3) a within-type robust regime in which both tSNR and dCNR achieve nearly optimal performance when the form of the weights are variants of their respective optimal weights and exhibit a moderate decrease in performance for other weight variants. Insight into the behavior observed in the different regimes is gained by considering spherical representations of the weight dependence of the components used to form each metric. For multi-echo acquisitions, dCNR is shown to be more directly related than tSNR to measures of CNR and signal-to-noise ratio (SNR) for both task-based and resting-state fMRI scans, respectively.

Self-navigated prospective motion correction for 3D-EPI acquisition

PURPOSE

Although 3D EPI is more susceptible to motion artifacts than 2D EPI, it presents some benefits for functional MRI, including the absence of spin-history artifacts, greater potential for parallel imaging acceleration, and better functional sensitivity in high-resolution imaging. Here we present a self-navigated 3D-EPI sequence suitable for prospective motion-corrected functional MRI without additional hardware or pulses.

METHODS

For each volume acquisition, the first 24 of the 52 partitions being acquired are accumulated to a new feedback block that was added to the image reconstruction pipeline. After zero-filling the remaining partitions, the feedback block constructs a volumetric self-navigator (vSNav), co-registers it to the reference vSNav acquired during the first volume acquisition, and sends motion estimates to the sequence. The sequence then updates its FOV and acquires subsequent partitions with the adjusted FOV, until the next update is received. The sequence was validated without and with intentional motion in phantom and in vivo on a 3T Skyra.

RESULTS

For phantom scans, the FOV was updated 0.704 s after acquisition of the vSNav partitions, and for in vivo scans after 0.768 s. Both phantom and in vivo data demonstrated stable motion estimates in the absence of motion. For in vivo acquisitions, prospective head-pose estimates using the vSNav's and retrospective estimates with FLIRT (FMRIB's Linear Image Registration Tool) agreed to within 0.23 mm (< 10% of the slice thickness) and 0.14° in all directions.

CONCLUSION

Depending when motion occurs during a volume acquisition, the proposed method fully corrects the FOV and recovers image quality within one volume acquisition.

The effect of acquisition duration on cerebral blood flow-based resting-state functional connectivity

Resting-state functional connectivity (rs-FC) is widely used to examine the functional architecture of the brain, and the blood-oxygenation-level-dependent (BOLD) signal is often utilized for determining rs-FC. However, the BOLD signal is susceptible to various factors that have less influence on the cerebral blood flow (CBF). Therefore, CBF could comprise an alternative for determining rs-FC. Since acquisition duration is one of the essential parameters for obtaining reliable rs-FC, we investigated the effect of acquisition duration on CBF-based rs-FC to examine the reliability of CBF-based rs-FC. Nineteen participants underwent CBF scanning for a total duration of 50 min. Variance of CBF-based rs-FC within the whole brain and 13 large-scale brain networks at various acquisition durations was compared to that with a 50-min duration using the Levene's test. Variance of CBF-based rs-FC at any durations did not differ from that at a 50-min duration (p > .05). Regarding variance of rs-FC within each large-scale brain network, the acquisition duration required to obtain reliable estimates of CBF-based rs-FC was shorter than 10 min and varied across large-scale brain networks. Altogether, an acquisition duration of at least 10 min is required to obtain reliable CBF-based rs-FC. These results indicate that CBF-based resting-state functional magnetic resonance imaging (rs-fMRI) with more than 10 min of total acquisition duration could be an alternative method to BOLD-based rs-fMRI to obtain reliable rs-FC.

A comprehensive investigation of physiologic noise modeling in resting state fMRI; time shifted cardiac noise in EPI and its removal without external physiologic signal measures

Hemodynamic cardiac and respiratory-cycle fluctuations are a source of unwanted non-neuronal signal components, often called physiologic noise, in resting state (rs-) fMRI studies. Here, we use image-based retrospective correction of physiological motion (RETROICOR) with externally measured physiologic signals to investigate cardiac and respiratory hemodynamic phase functions reflected in rs-fMRI data. We find that the cardiac phase function is time shifted locally, while the respiratory phase function is described as single, fixed phase form across the brain. In light of these findings, we propose an update to Physiologic EStimation by Temporal ICA (PESTICA), our publically available software package that estimates physiologic signals when external physiologic measures are not available. This update incorporates: 1) auto-selection of slicewise physiologic regressors and generation of physiologic fixed phase regressors with total slices/TR sampling rate, 2) Fourier series expansion of the cardiac fixed phase regressor to account for time delayed cardiac noise 3) removal of cardiac and respiratory noise in imaging data. We compare the efficacy of the updated method to RETROICOR.

A Multimodal Multilevel Neuroimaging Model for Investigating Brain Connectome Development

Recent advancements of multimodal neuroimaging such as functional MRI (fMRI) and diffusion MRI (dMRI) offers unprecedented opportunities to understand brain development. Most existing neurodevelopmental studies focus on using a single imaging modality to study microstructure or neural activations in localized brain regions. The developmental changes of brain network architecture in childhood and adolescence are not well understood. Our study made use of dMRI and resting-state fMRI imaging data sets from Philadelphia Neurodevelopmental Cohort (PNC) study to characterize developmental changes in both structural as well as functional brain connectomes. A multimodal multilevel model (MMM) is developed and implemented in PNC study to investigate brain maturation in both white matter structural connection and intrinsic functional connection. MMM addresses several major challenges in multimodal connectivity analysis. First, by using a first-level data generative model for observed measures and a second-level latent network modeling, MMM effectively infers underlying connection states from noisy imaging-based connectivity measurements. Secondly, MMM models the interplay between the structural and functional connections to capture the relationship between different brain connectomes. Thirdly, MMM incorporates covariate effects in the network modeling to investigate network heterogeneity across subpopoulations. Finally, by using a module-wise parameterization based on brain network topology, MMM is scalable to whole-brain connectomics. MMM analysis of the PNC study generates new insights in neurodevelopment during adolescence including revealing the majority of the white fiber connectivity growth are related to the cognitive networks where the most significant increase is found between the default mode and the executive control network with a 15% increase in the probability of structural connections. We also uncover functional connectome development mainly derived from global functional integration rather than direct anatomical connections. To the best of our knowledge, these findings have not been reported in the literature using multimodal connectomics. Supplementary materials for this article, including a standardized description of the materials available for reproducing the work, are available as an online supplement.