NeuroImage

Recent discussions on the reproducibility of task-related functional magnetic resonance imaging (fMRI) studies have emphasized the importance of power and sample size calculations in fMRI study planning. In general, statistical power and sample size calculations are dependent on the statistical inference framework that is used to test hypotheses. Bibliometric analyses suggest that random field theory (RFT)-based voxel- and cluster-level fMRI inference are the most commonly used approaches for the statistical evaluation of task-related fMRI data. However, general power and sample size calculations for these inference approaches remain elusive. Based on the mathematical theory of RFT-based inference, we here develop power and positive predictive value (PPV) functions for voxel- and cluster-level inference in both uncorrected single test and corrected multiple testing scenarios. Moreover, we apply the theoretical results to evaluate the sample size necessary to achieve desired power and PPV levels based on an fMRI pilot study.

Simultaneous electroencephalography (EEG) and functional MRI (fMRI) offers complementary sensitivity to fast electrophysiological dynamics of EEG and spatially resolved hemodynamics of fMRI, yet previous joint-analysis approaches are confined to fixed task paradigms and struggle with continuous or naturalistic brain states. We FSINC (Fusing Source Imaging based on a Neurovascular Coupling) model, a unified EEG-fMRI source imaging framework that reconstructs cortical activity to simultaneously explain both modalities. FSINC integrates frequency-resolved EEG source activity with fMRI via a data-driven neurovascular coupling model that estimates band-specific coupling coefficients ({beta}) and accommodates a tunable spatial-temporal trade-off through hyperparameters ({lambda}2,{lambda} 3). In realistic simulations, FSINC outperformed conventional methods (wMNE, LORETA) in both spatial and temporal accuracy across EEG SNRs (-10 to 10dB) and numbers of concurrent sources (up to five), with optimal performance at{lambda} 2 = 102 and{lambda} 3=1 (e.g., LE: 0.51{+/-}0.24mm; SDI: 0.03{+/-}0.37mm; temporal accuracy: 0.95 {+/-} 0.05). Applied to simultaneous EEG-fMRI during contrast-reversing visual stimulation (=5.95Hz), FSINC revealed stimulus-locked responses localized to early visual cortex and stimulus-induced modulation of intrinsic alpha oscillations extending into visual and attention networks, patterns that conventional methods failed to capture. Estimated {beta}-weights were broadly consistent with prior reports of negative (theta/alpha) and positive (gamma) BOLD-electrophysiology associations. These findings demonstrate that FSINC enables high-spatiotemporal-resolution source imaging from EEG-fMRI recordings via data-driven hemodynamic modelling, and is expected to be well-suited for continuous and naturalistic brain states (e.g., resting state, natural moving-watching, and narrative listening) that are difficult to interrogate with either modality alone.

Cluster-wise statistical inference is the most widely used technique for functional magnetic resonance imaging (fMRI) data analyses. Cluster-wise statistical inference consists of two steps: i) primary thresholding that excludes less significant voxels by a pre-specified cut-off (e.g., p < 0.001); and ii) cluster-wise thresholding that controls the family-wise error rate (FWER) caused by clusters consisting of false positive suprathreshold voxels. It has been well known that the selection of the primary threshold is critical because it determines both statistical power and false discovery rate. However, in most existing statistical packages, the primary threshold is selected based on prior knowledge (e.g., p < 0.001) without taking into account the information in the data. In this manuscript, we propose a data-driven approach to objectively select the optimal primary threshold based on an empirical Bayes framework. We evaluate the proposed model using extensive simulation studies and an fMRI data example. The results show that our method can effectively increase statistical power while effectively controlling the false discovery rate.

Magnetic Resonance Imaging (MRI) technology has been increasingly used in large-scale association studies. Reproducibility of statistically significant findings generated by MRI-based association studies, especially structural MRI (sMRI) and functional MRI (fMRI), has been recently heavily debated. However, there is still a lack of overall reproducibility assessment for MRI-based association studies. It is also crucial to elucidate the relationship between overall reproducibility and sample size in an experimental design. In this study, we proposed an overall reproducibility index for large-scale high-throughput MRI-based association studies. We performed the overall reproducibility assessments for several recent large sMRI/fMRI databases and observed satisfactory overall reproducibility. Furthermore, we performed the sample size evaluation for the purpose of achieving a desirable overall reproducibility. Additionally, we evaluated the overall reproducibility of GMV changes for UKB vs. PPMI and UKB vs. HCP. We demonstrated that both sample size and some experimental factors play important roles in the overall reproducibility for different experiments. In summary, a systematic assessment of overall reproducibility is fundamental and crucial in the current large-scale high-throughput MRI-based research.

The oscillatory nature of intrinsic brain networks is largely taken for granted in the systems neuroscience community. However, the hypothesis that brain rhythms--and by extension transient bursting oscillations--underlie functional networks has not been demonstrated per se. Electrophysiological measures of functional connectivity are indeed affected by the power bias, which may lead to artefactual observations of spectrally specific network couplings not genuinely driven by neural oscillations, bursting or not. We investigate this crucial question by introducing a unique combination of a rigorous mathematical analysis of the power bias in frequency-dependent amplitude connectivity with a neurobiologically informed model of cerebral background noise based on hidden Markov modeling of resting-state magnetoencephalography (MEG). We demonstrate that the power bias may be corrected by a suitable renormalization depending nonlinearly on the signal-to-noise ratio, with noise identified as non-bursting oscillations. Applying this correction preserves the spectral content of amplitude connectivity, definitely proving the importance of brain rhythms in intrinsic functional networks. Our demonstration highlights a dichotomy between spontaneous oscillatory bursts underlying network couplings and non-bursting oscillations acting as background noise but whose function remains unsettled. Significance statementBrain rhythms are paramount electrophysiological correlates of human cerebral activity as they coordinate neurons across distinct brain areas and establish neural synchronization. Spontaneous cortical oscillations, and particularly transient "bursts" of oscillations, also appear as the main electrophysiological subtrate of intrinsic functional brain networks. However, we argue that this oscillatory theory of brain networks, despite being widely accepted, should be requestioned due to a critical bias in electrophysiological measures of network connectivity. Here, we combined mathematical and neurobiological modeling techniques with magnetoencephalography recordings to set this theory on firm and rigorous grounds. Key to our analysis are scarcely studied non-bursting cortical oscillations. Although their precise function remains elusive, our results reveal that dissociating bursting and non-bursting oscillations is fundamental to the rigorous interpretation of electrophysiological network connectivity.

Multivariate statistics and machine learning methods have become a common tool to extract information represented in the brain. What is less recognized is that, in the process, it has become more difficult to perform data-driven discovery and functional localization. This is because multivariate pattern analysis (MVPA) studies tend to restrict themselves to a subset of the available data, or because sound inference to map model parameters back to brain anatomy is lacking. Here, we present a high-dimensional (including brain-wide) multivariate classification pipeline for the detection and localization of brain functions during tasks. In particular, we probe it at visual and socio-affective states in a task-oriented functional magnetic resonance imaging (fMRI) experiment. Classification models for a group of human participants and existing rigorous cluster inference methods are used to construct group anatomical-statistical parametric maps, which correspond to the most likely neural correlates of each psychological state. This led to the discovery of a multidimensional pattern of macroscale brain activity which reliably encodes for the perception of happiness in the visual cortex, lingual gyri and the posterior perivermian cerebellum. We failed to find similar evidence for sadness and anger. Anatomical consistency of discriminating features across subjects and contrasts despite the high number of dimensions suggests MVPA is a viable tool for a complete functional mapping pipeline, and not just the prediction of psychological states. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=134 SRC="FIGDIR/small/438425v3_ufig1.gif" ALT="Figure 1"> View larger version (34K): org.highwire.dtl.DTLVardef@1961forg.highwire.dtl.DTLVardef@26d25aorg.highwire.dtl.DTLVardef@bc4b35org.highwire.dtl.DTLVardef@1eda0a6_HPS_FORMAT_FIGEXP M_FIG C_FIG

In diffusion MRI, the Ensemble Average diffusion Propagator (EAP) provides relevant microstructural information and meaningful descriptive maps of the white matter previously obscured by traditional techniques like the Diffusion Tensor. The direct estimation of the EAP, however, requires a dense sampling of the Cartesian q-space. Due to the huge amount of samples needed for an accurate reconstruction, more efficient alternative techniques have been proposed in the last decade. Even so, all of them imply acquiring a large number of diffusion gradients with different b-values. In order to use the EAP in practical studies, scalar measures must be directly derived, being the most common the return-to-origin probability (RTOP) and the return-to-plane and return-to-axis probabilities (RTPP, RTAP).\n\nIn this work, we propose the so-called \"Apparent Measures Using Reduced Acquisitions\" (AMURA) to drastically reduce the number of samples needed for the estimation of diffusion properties. AMURA avoids the calculation of the whole EAP by assuming the diffusion anisotropy is roughly independent from the radial direction. With such an assumption, and as opposed to common multi-shell procedures based on iterative optimization, we achieve closed-form expressions for the measures using information from one single shell. This way, the new methodology remains compatible with standard acquisition protocols commonly used for HARDI (based on just one b-value). We report extensive results showing the potential of AMURA to reveal microstructural properties of the tissues compared to state of the art EAP estimators, and is well above that of Diffusion Tensor techniques. At the same time, the closed forms provided for RTOP, RTPP, and RTAP-like magnitudes make AMURA both computationally efficient and robust.

Many functional magnetic resonance imaging (fMRI) studies conclude that two conditions engage "overlapping, yet partly distinct" patterns of activation. Yet, there is currently no commonly accepted method for determining the extent of this overlap. While correlations between activation patterns can serve as a measure of their correspondence, empirical correlations are strongly biased towards zero due to measurement noise, preventing their use in testing hypotheses about the actual degree of pattern correspondence. In this paper, we derive the maximum-likelihood estimate for the correlation of the true (noise-less) activation patterns and examine its behavior in the low signal-to-noise regime that is typical for fMRI studies. We show that although the maximum-likelihood estimate corrects for much of the influence of measurement noise, it is ultimately biased. We examine different ways of drawing inferences about the size of the underlying true correlations. We find that a subject-wise bootstrap on the maximum-likelihood group estimate performs best over the tested conditions. We extend the proposed method to test more general hypotheses about the representational geometry of activation patterns for more conditions, and highlight best practices, as well as common pitfalls and problems, in testing such hypotheses.

Many recent studies have demonstrated the inflated type 1 error rate of the original Gaussian random field (GRF) methods for inference of neuroimages and identified resampling (permutation and bootstrapping) methods that have better performance. There has been no evaluation of resampling procedures when using robust (sandwich) statistical images with different topological features (TF) used for neuroimaging inference. Here, we consider estimation of distributions TFs of a statistical image and evaluate resampling procedures that can be used when exchangeability is violated. We compare the methods using realistic simulations and study sex differences in life-span age-related changes in gray matter volume in the Nathan Kline Institute Rockland sample. We find that our proposed wild bootstrap and the commonly used permutation procedure perform well in sample sizes above 50 under realistic simulations with heteroskedasticity. The Rademacher wild bootstrap has fewer assumptions than the permutation and performs similarly in samples of 100 or more, so is valid in a broader range of conditions. We also evaluate the GRF-based pTFCE method and show that it has inflated error rates in samples less than 200. Our R package, pbj, is available on Github and allows the user to reproducibly implement various resampling-based group level neuroimage analyses.

Mar[c]enko-Pastur (MP) PCA denoising is emerging as an effective means for noise suppression in MRI acquisitions with redundant dimensions. However, MP-PCA performance is severely compromised by spatially correlated noise - an issue typically affecting most modern MRI acquisitions - almost to the point of returning the original images with little or no noise removal. In this study, we develop and apply two new strategies that enable efficient and robust denoising even in the presence of severe spatial correlations. This is achieved by measuring a-priori information about the noise variance and combing these estimates with PCA denoising thresholding concepts. The two denoising strategies developed here are: 1) General PCA (GPCA) denoising that uses a-priori noise variance estimates without assuming specific noise distributions; and 2) Threshold PCA (TPCA) denoising which removes noise components with a threshold computed from a-priori estimated noise variance to determine the upper bound of the MP distribution. These strategies were tested in simulations with known ground truth and applied for denoising diffusion MRI data acquired using pre-clinical (16.4T) and clinical (3T) MRI scanners. In synthetic phantoms, MP-PCA failed to denoise spatially correlated data, while GPCA and TPCA correctly classified all signal/noise components. In cases where the noise variance was not accurately estimated (as can be the case in many practical scenarios), TPCA still provides excellent denoising performance. Our experiments in pre-clinical diffusion data with highly corrupted by spatial correlated noise revealed that both GPCA and TPCA robustly denoised the data while MP-PCA denoising failed. In in vivo diffusion MRI data acquired on a clinical scanner in healthy subjects, MP-PCA weakly removed noised, while TPCA was found to have the best performance, likely due to misestimations of the noise variance. Thus, our work shows that these novel denoising approaches can strongly benefit future pre-clinical and clinical MRI applications.

Functional near-infrared spectroscopy (fNIRS) measures the hemoglobin concentration changes associated with neuronal activity. Diffuse optical tomography (DOT) consists of reconstructing the optical density changes measured from scalp channels to the oxy-/deoxy-hemoglobin (i.e., HbO/HbR) concentration changes within the cortical regions. In the present study, we adapted a nonlinear source localization method developed and validated in the context of Electro- and Magneto-Encephalography (EEG/MEG): the Maximum Entropy on the Mean (MEM), to solve the inverse problem of DOT reconstruction. We first introduced depth weighting strategy within the MEM framework for DOT reconstruction to avoid biasing the reconstruction results of DOT towards superficial regions. We also proposed a new initialization of the MEM model improving the temporal accuracy of the original MEM framework. To evaluate MEM performance and compare with widely used depth weighted Minimum Norm Estimate (MNE) inverse solution, we applied a realistic simulation scheme which contained 4000 simulations generated by 250 different seeds at different locations and 4 spatial extents ranging from 3 to 40cm2 along the cortical surface. Our results showed that overall MEM provided more accurate DOT reconstructions than MNE. Moreover, we found that MEM was remained particularly robust in low signal-to-noise ratio (SNR) conditions. The proposed method was further illustrated by comparing to functional Magnetic Resonance Imaging (fMRI) activation maps, on real data involving finger tapping tasks with two different montages. The results showed that MEM provided more accurate HbO and HbR reconstructions in spatial agreement with the main fMRI cluster, when compared to MNE. HighlightsO_LIWe introduced a new fNIRS reconstruction method - Maximum Entropy on the Mean. C_LIO_LIWe implemented depth weighting strategy within the MEM framework. C_LIO_LIWe improved the temporal accuracy of the original MEM reconstruction. C_LIO_LIPerformances of MEM and MNE were evaluated with realistic simulations and real data. C_LIO_LIMEM provided more accurate and robust reconstructions than MNE. C_LI

In the statistical analysis of functional Magnetic Resonance Imaging (fMRI) brain data it remains a challenge to account for simultaneously testing activation in over 100.000 volume units or voxels. A popular method that reduces the dimensionality of this test problem is cluster-based inference. We propose a new testing procedure that allows to control the family-wise error (FWE) rate at the cluster level but improves cluster-based test decisions in two ways by (1) taking into account a measure for data analytical stability and (2) allowing a more voxel-based interpretation of the results. For each voxel, we define the re-selection rate conditional on a given FWE-corrected threshold and use this rate, which is a measure of stability, into the selection process. In our procedure, we set a more liberal and a more conservative FWE controlling threshold. Clusters that survive the liberal but not the conservative threshold are retained if sufficient evidence for voxelwise stability is available. Cluster that survive the conservative threshold are retained anyhow, and clusters that do not survive the liberal threshold are not further considered. Using the Human Connectome Project Data (Van Essen et al., 2012), we demonstrate how in a group analysis our method results not only in a higher number of selected voxels but also in a larger overlap between different test images. Additionally, we demonstrate the ability of our procedure to control the FWE, also in relatively small sample sizes.

During preclinical drug testing, the systemic administration of scopolamine (SCO), a cholinergic antagonist, is widely used. However, it suffers important limitations, like non-specific behavioural effects partly due to its peripheral side-effects. Therefore, neuroimaging measures would enhance its translational value. To this end, in Wistar rats, we measured whisker-stimulation induced functional MRI activation after SCO, peripherally acting butylscopolamine (BSCO), or saline administration in a cross-over design. Besides the commonly used gradient-echo echo-planar imaging (GE EPI), we also used an arterial spin labeling method in isoflurane anesthesia. With the GE EPI measurement, SCO decreased the evoked BOLD response in the barrel cortex (BC), while BSCO increased it in the anterior cingulate cortex. In a second experiment, we used GE EPI and spin-echo (SE) EPI sequences in a combined (isoflurane + i.p. dexmedetomidine) anesthesia to account for anesthesia-effects. Here, we also examined the effect of donepezil. In the combined anesthesia, with the GE EPI, SCO decreased the activation in the BC and the inferior colliculus (IC). BSCO reduced the response merely in the IC. Our results revealed that SCO attenuated the evoked BOLD activation in the BC as a probable central effect in both experiments. The likely peripheral vascular actions of SCO with the given fMRI sequences depended on the type of anesthesia or its dose.

Since their introduction in the late eighties, Bayesian approaches for neuroimaging have opened the way to new powerful and quantitative analysis of brain data. Here, we apply this statistical framework to evaluate empirically the gain of fused EEG-MEG source reconstruction, compared to unimodal (EEG or MEG) one. Combining EEG and MEG information for source reconstruction has been consistently evidenced to enhance localization performances using simulated data. However, given considerable efforts to conduct simultaneous recordings, empirical evaluation becomes necessary to quantify the real information gain. And this is obviously not straightforward due to the ill-posedness of the inverse problem. Here, we consider Bayesian model comparison to quantify the ability of EEG, MEG and fused (EEG/MEG) inversions of individual data to resolve spatial source models. These models consisted in group-level cortical distributions inferred from real EEG, MEG and EEG/MEG brain responses. We applied this comparative evaluation to the timely issue of the generators of auditory mismatch responses evoked by unexpected sounds. These included the well-known Mismatch Negativity (MMN) but also earlier deviance responses. As expected, fused localization was evidenced to outperform unimodal inversions with larger model separability. The present methodology confirms with real data the theoretical interest of simultaneous EEG/MEG recordings and fused inversion to highly inform (spatially and temporally) source modeling. Precisely, a bilateral fronto-temporal network could be identified for both the MMN and early deviance response. Interestingly, multimodal inversions succeeded in revealing spatio-temporal details of the functional organization within the supratemporal plane that have not been reported so far, nor were visible here with unimodal inversions. The present refined auditory network could serve as priors for auditory modeling studies.

According to theoretical studies, the MRI signal decay due to transverse relaxation, in brain tissue with magnetic inclusions (e.g. blood vessels, myelin, iron-rich cells), exhibits a transition from a Gaussian behaviour at short echo times to exponential at long echo times. Combined, the Gaussian and exponential decay parameters carry information about the inclusions (e.g., size, volume fraction) and provide a unique insight into brain tissue microstructure. However, gradient echo decays obtained experimentally typically only capture the long-time exponential behaviour. Here, we provide experimental evidence of non-exponential transverse relaxation signal decay at short times in human subcortical grey matter, from MRI data acquired in vivo at 3T. The detection of the non-exponential behaviour of the signal decay allows the subsequent characterization of the magnetic inclusions in the subcortex. The gradient-echo data was collected with short inter-echo spacings, a minimal echo time of 1.25 ms and novel acquisition strategies tailored to mitigate the effect of motion and cardiac pulsation. The data was fitted using both a standard exponential model and non-exponential theoretical models describing the impact of magnetic inclusions on the MRI signal. The non-exponential models provided superior fits to the data, indicative of a better representation of the observed signal. The strongest deviations from exponential behaviour were detected in the substantia nigra and globus pallidus. Numerical simulations of the signal decay, conducted from histological maps of iron concentration in the substantia nigra, closely replicated the experimental data - highlighting that non-heme iron can be at the source of the non-exponential signal decay. To investigate the potential of the non-exponential signal decay as a tool to characterize brain microstructure, we attempted to estimate the properties of the inclusions at the source of this decay behaviour using two available analytical models of transverse relaxation. Under the assumption of the static dephasing regime, the magnetic susceptibility and volume fractions of the inclusions was estimated to range from 1.8 to 4 ppm and from 0.02 to 0.04 respectively. Alternatively, under the assumption of the diffusion narrowing regime, the typical inclusion size was estimated to be [~]2.4 m. Both simulations and experimental data point towards an intermediate regime with a non-negligible effect of water diffusion to signal decay. Non-exponential transverse relaxation decay provides new means to characterize the spatial distribution of magnetic material within subcortical grey matter tissue with increased specificity, with potential applications to Parkinsons disease and other pathologies.

Network neuroscience has emerged as an indispensable tool for studying brain structure and function. Currently, the network-based statistic (NBS) procedure is widely used for dealing with massive multiple testing/comparison problems in brain networks. However, the NBS requires choosing a hard cluster-forming threshold, lacking objective rules. A powerful and flexible statistical framework is urgently needed with growing interest in finer-grained network explorations across modalities and scales. Here, we introduce a permutation-based framework--"Threshold-Free Network-Oriented Statistics" (TFNOS). It integrates two "threshold-free" pathways: traversing all cluster-forming thresholds (TT) and using predefined clusters (PC). The TT procedure, building upon the threshold-free network-based statistics, requires setting additional parameters. The PC procedures comprise six variants given the degree of freedom in pooling data, null distribution construction, and controlled error rate. Using numerical simulations, we evaluated the performance of the TT procedure under 600 parameter combinations, then benchmarked TFNOS procedures and baselines across different topologies of effects, sample sizes, and effect sizes, and finally provided illustrative examples with real data. We offer recommended parameter values that allow the TT procedure to stably maintain leading power, while empirically controlling the false discovery rate (FDR) beyond only weakly controlling the familywise error rate (FWER). Notably, the relevant parameters commonly employed in the field appear overly liberal. Furthermore, for the PC procedures, FDR-controlling variants showed improved power compared to FWER-controlling variants, and some of them are simple but do not compromise power. The nonparametric PC procedures allow the selection of any test statistics considered appropriate. Overall, the TFNOS is a generalized framework for inference on edges/nodes of undirected/directed brain networks. We provide empirical and principled criteria for selecting appropriate procedures and may enhance the reproducibility and sensitivity of future brain research.

BackgroundReproducibility of research findings has been recently questioned in many fields of science, including psychology and neurosciences. One factor influencing reproducibility is the simultaneous testing of multiple hypotheses, which increases the number of false positive findings unless the p-values are carefully corrected. While this multiple testing problem is well known and has been studied for decades, it continues to be both a theoretical and practical problem.\n\nNew MethodHere we assess the reproducibility of research involving multiple-testing corrected for family-wise error rate (FWER) or false discovery rate (FDR) by techniques based on random field theory (RFT), cluster-mass based permutation testing, adaptive FDR, and several classical methods. We also investigate the performance of these methods under two different models.\n\nResultsWe found that permutation testing is the most powerful method among the considered approaches to multiple testing, and that grouping hypotheses based on prior knowledge can improve power. We also found that emphasizing primary and follow-up studies equally produced most reproducible outcomes.\n\nComparison with Existing Method(s)We have extended the use of two-group and separate-classes models for analyzing reproducibility and provide a new open-source software \"MultiPy\" for multiple hypothesis testing.\n\nConclusionsOur results suggest that performing strict corrections for multiple testing is not sufficient to improve reproducibility of neuroimaging experiments. The methods are freely available as a Python toolkit \"MultiPy\" and we aim this study to help in improving statistical data analysis practices and to assist in conducting power and reproducibility analyses for new experiments.

Bansal and Peterson (2018) found that in simple stationary Gaussian simulations Random Field Theory incorrectly estimates the number of clusters of a Gaussian field that lie above a threshold. Their results contradict the existing literature and appear to have arisen due to errors in their code. Using reproducible code we demonstrate that in their simulations Random Field Theory correctly predicts the expected number of clusters and therefore that many of their results are invalid.

Scientific progress is based on the ability to compare opposing theories and thereby develop consensus among existing hypotheses or create new ones. We argue that data aggregation (i.e. combine data across studies or research groups) for neuroscience is an important tool in this process. An important prerequisite is the ability to directly compare fMRI results over studies. In this paper, we discuss how an observed effect size in an fMRI data-analysis can be transformed into a standardized effect size. We demonstrate how these enable direct comparison and data aggregation over studies. Furthermore, we also discuss the influence of key parameters in the design of an fMRI experiment (such as number of scans and the sample size) on (statistical) properties of standardized effect sizes. In the second part of the paper, we give an overview of two approaches to aggregate fMRI results over studies. The first corresponds to extending the two-level general linear model approach as is typically used in individual fMRI studies with a third level. This requires the parameter estimates corresponding to the group models from each study together with estimated variances and meta-data. Unfortunately, there is a risk of running into unit mismatches when the primary studies use different scales to measure the BOLD response. To circumvent, it is possible to aggregate (unitless) standardized effect sizes which can be derived from summary statistics. We discuss a general model to aggregate these and different approaches to deal with between-study heterogeneity. Furthermore, we hope to further promote the usage of standardized effect sizes in fMRI research.

Diffusion Spectrum Imaging (DSI) using dense Cartesian sampling of q-space has been shown to provide important advantages for modeling complex white matter architecture. However, its adoption has been limited by the lengthy acquisition time required. Sparser sampling of q-space combined with compressed sensing (CS) reconstruction techniques has been proposed as a way to reduce the scan time of DSI acquisitions. However prior studies have mainly evaluated CS-DSI in post-mortem or non-human data. At present, the capacity for CS-DSI to provide accurate and reliable measures of white matter anatomy and microstructure in the living human brain remains unclear. We evaluated the accuracy and inter-scan reliability of 6 different CS-DSI schemes that provided up to 80% reductions in scan time compared to a full DSI scheme. We capitalized on a dataset of twenty-six participants who were scanned over eight independent sessions using a full DSI scheme. From this full DSI scheme, we subsampled images to create a range of CS-DSI images. This allowed us to compare the accuracy and inter-scan reliability of derived measures of white matter structure (bundle segmentation, voxel-wise scalar maps) produced by the CS-DSI and the full DSI schemes. We found that CS-DSI estimates of both bundle segmentations and voxel-wise scalars were nearly as accurate and reliable as those generated by the full DSI scheme. Moreover, we found that the accuracy and reliability of CS-DSI was higher in white matter bundles that were more reliably segmented by the full DSI scheme. As a final step, we replicated the accuracy of CS-DSI in a prospectively acquired dataset (n=20, scanned once). Together, these results illustrate the utility of CS-DSI for reliably delineating in vivo white matter architecture in a fraction of the scan time, underscoring its promise for both clinical and research applications.