Impact of the resolution of brain parcels on connectome-wide association studies in fMRI

doi:10.1016/j.neuroimage.2015.07.071

NeuroImage

Volume 123, December 2015, Pages 212-228

https://doi.org/10.1016/j.neuroimage.2015.07.071 Get rights and content

Highlights

•
The FDR is controlled in GLM-connectome analysis with a single brain parcellation.
•
Replicating a GLM with multiresolution brain parcels can inflate the overall FDR.
•
A new omnibus test for multiresolution GLM-connectome analysis.
•
Simulation show increased GLM sensitivity using less than 50 brain parcels.
•
30 brain parcels summarized well GLM effects in three real datasets.

Abstract

A recent trend in functional magnetic resonance imaging is to test for association of clinical disorders with every possible connection between selected brain parcels. We investigated the impact of the resolution of functional brain parcels, ranging from large-scale networks to local regions, on a mass univariate general linear model (GLM) of connectomes. For each resolution taken independently, the Benjamini–Hochberg procedure controlled the false-discovery rate (FDR) at nominal level on realistic simulations. However, the FDR for tests pooled across all resolutions could be inflated compared to the FDR within resolution. This inflation was severe in the presence of no or weak effects, but became negligible for strong effects. We thus developed an omnibus test to establish the overall presence of true discoveries across all resolutions. Although not a guarantee to control the FDR across resolutions, the omnibus test may be used for descriptive analysis of the impact of resolution on a GLM analysis, in complement to a primary analysis at a predefined single resolution. On three real datasets with significant omnibus test (schizophrenia, congenital blindness, motor practice), markedly higher rate of discovery were obtained at low resolutions, below 50, in line with simulations showing increase in sensitivity at such resolutions. This increase in discovery rate came at the cost of a lower ability to localize effects, as low resolution parcels merged many different brain regions together. However, with 30 or more parcels, the statistical effect maps were biologically plausible and very consistent across resolutions. These results show that resolution is a key parameter for GLM-connectome analysis with FDR control, and that a functional brain parcellation with 30 to 50 parcels may lead to an accurate summary of full connectome effects with good sensitivity in many situations.

Introduction

Brain connectivity in resting-state functional magnetic resonance imaging (fMRI) has been found to be associated with a wide variety of clinical disorders (Fox and Greicius, 2010, Castellanos et al., 2013, Barkhof et al., 2014). Rather than focusing on a limited set of a priori regions of interest, a recent trend is to perform statistical tests of association across the whole connectome, i.e. at every possible brain connection (Shehzad et al., 2014). Such connectome-wide association studies (CWAS) critically depend on the choice of the brain parcels that are used to estimate the connections. Analyses have been performed at different resolutions in the literature (Meskaldji et al., 2013), e.g. voxels (Shehzad et al., 2014), regions (Wang et al., 2007), or distributed networks (Jafri et al., 2008, Marrelec et al., 2008). The main objective of this work was to study the impact of the spatial resolution on the results of a CWAS.

The mass-univariate approach to CWAS (Worsley et al., 1998) consists of independently estimating a GLM at every connection. In the GLM, a series of equations are solved to find a linear mixture of explanatory variables (called covariates) that best fit the connectivity values observed across the many subjects. A p value is generated for each connection to quantify the probability that the estimated strength of association between this connection and a covariate of interest could have arisen randomly in the absence of a true association (Worsley and Friston, 1995). The significance level of each test needs to be corrected for the total number of tests, i.e. the number of brain connections, using for example random field theory (Worsley et al., 1998) or FDR (Benjamini and Hochberg, 1995). Correction for multiple comparisons however generally comes at the cost of a sharp decrease in sensitivity.

A straightforward way to mitigate the impact of multiple comparisons on statistical power is to reduce the number of brain parcels. For example, the AAL template (Tzourio-Mazoyer et al., 2002) includes 116 brain parcels based on anatomical landmarks. Data-driven algorithms can also generate functional brain parcels (Bellec et al., 2006, Thirion et al., 2006, Thirion et al., 2014, Craddock et al., 2012, Blumensath et al., 2013, Gordon et al., 2014). Few investigators have examined how resolution impacts the results of a CWAS. Abou Elseoud et al. (2011) explored the impact of the number of components in a dual-regression independent component analysis on the difference between patients suffering from non-medicated seasonal affective disorder and normal healthy controls. The authors concluded that the number of significant findings was maximized at resolution 45 (in this case, 45 independent components). The impact of the number of brain parcels was also investigated using spatially-constrained spectral clustering (Craddock et al., 2012) at much higher resolutions (from 50 to 3000 +) by Shehzad et al. (2014). The authors concluded that the association between resting-state connectivity and intelligence quotient was consistent across resolutions. It should be noted that, in the above-mentioned studies (Abou Elseoud et al., 2011, Shehzad et al., 2014), the authors did not investigate the implications that the replication of statistical tests at multiple resolutions may have in terms of the control of false positives. We are thus not currently aware of a valid statistical framework to examine the results of a CWAS with data-driven brain parcellations at multiple resolutions.

In this paper, we investigated empirically the impact of the number of brain parcels (resolution) on a mass univariate GLM analysis of connectomes (GLM-connectome). Our first objective was to empirically assess if the Benjamini–Hochberg procedure (Benjamini and Hochberg, 1995) controlled appropriately the FDR with a single brain parcellation, and generated biologically plausible results. Our second objective was to assess if repeating a GLM-connectome analysis independently using multiple parcellations at different resolutions would inflate the overall FDR, pooling tests across all resolutions. Anticipating a lack of control in the absence of any true association, we developed an omnibus test checking for the overall presence of significant associations across all resolutions. Our third objective was to use the omnibus test to evaluate how the resolution impacted the rate of discovery in a GLM analysis, and if the associations derived with the GLM would be consistent across resolutions. We conducted a series of experiments involving both simulated and real datasets to address these three objectives, which have been summarized, along with the main findings, in Table 1.

Section snippets

Functional connectome

The first step to build a connectome is to select a parcellation of the brain, with R parcels. In this work, we relied on a “Bootstrap Analysis of Stable Clusters” (BASC), which can identify consistent functional parcels for a group of subjects (Bellec et al., 2010), using a hierarchical cluster with Ward's criterion both at the individual and the group levels. The functional parcels can be generated at any arbitrary resolution (within the range of the fMRI resolution), and we considered only

Data-generating procedure

We started by simple simulations of independent tests, to assess to which extent the hypothesis of Efron (2008) was robust to different scenarios, and if the omnibus test would systematically ensure that the FDR across resolutions would be well controlled. A number K of test resolutions were generated independently, each one composed of L_k tests, k = 1,…,K. Each resolution included a set proportion of true non-null hypotheses π₁, identical for all resolutions. If π₁L_k was not an integer, the

Data-generating procedure

We designed a simulation framework for multiresolution GLM-connectome analysis in the presence of dependencies between tests, both within resolution and across resolutions. To ensure that these dependencies would be as realistic as possible, semi-synthetic datasets were generated starting from a large real sample (Cambridge) released as part of the 1000 functional connectome project⁶ (Biswal et al., 2010). This sample (Liu et al., 2009

Participants

We evaluated the GLM-connectome on three real datasets: (1) a study (SCHIZO) comparing patients suffering from schizophrenia with healthy control subjects; (2) a study (BLIND) on patients suffering of congenital blindness, compared to sighted controls; and (3) a study (MOTOR) where resting-state data connectivity was compared before and after learning of a motor task. The SCHIZO dataset was contributed by the Center for Biomedical Research Excellence (COBRE) to the 1000 functional connectome

Specificity in multiresolution analysis

This work investigated empirically how the resolution impacts GLM analyses on connectomes, in particular in terms of specificity. We confirmed on realistic simulations the validity of a FDR control using the BH procedure at a single resolution. On three real datasets, there was no sign of substantial departure from the assumptions of a basic parametric GLM. The censoring of time frames with excessive motion will still very likely introduce some departure from the homoscedasticity assumption,

Conclusion

Our overall conclusion is that the GLM analysis of connectomes with control of the FDR using the BH procedure is statistically valid when used at a single resolution, and has the potential to identify biologically plausible associations in a variety of experimental conditions. Caution should be exercised when replicating a GLM-connectome analysis at different resolutions, as the FDR over the tests combined across all resolutions may depart from the FDR within resolution. We proposed a valid

Acknowledgments

Parts of this work were presented at the 2012 and 2013 annual meetings of the organization for human brain mapping, as well as the International Conference on Resting-State Connectivity 2012 (Magdeburg). The authors are grateful to the members of the 1000 functional connectome consortium for publicly releasing the “Cambridge” and “COBRE” data samples, as well as Dr Shehzad for valuable feedback. The computational resources used to perform the data analysis were provided by Compute Canada¹³

References (54)

A. Abou Elseoud et al.
Group-ICA model order highlights patterns of functional brain connectivity
Front. Syst. Neurosci.
(2011)
Y. Ad-Dab'bagh et al.
The CIVET image-processing environment: a fully automated comprehensive pipeline for anatomical neuroimaging research
N.B. Albert et al.
The resting human brain and motor learning
Curr. Biol.
(2009)
G. Albouy et al.
Maintaining vs. enhancing motor sequence memories: respective roles of striatal and hippocampal systems
NeuroImage
(2015)
C.W. Anderson
Quantitative Methods for Current Environmental Issues
(2002)
A. Anticevic et al.
Characterizing thalamo-cortical disturbances in schizophrenia and bipolar illness
Cereb. Cortex
(2014)
F. Barkhof et al.
Resting-state functional MR imaging: a new window to the brain
Radiology
(2014)
P. Bellec
Mining the hierarchy of resting-state brain networks: selection of representative clusters in a multiscale structure
P. Bellec et al.
Identification of large-scale networks in the brain using fMRI
NeuroImage
(2006)
P. Bellec et al.
Multi-level bootstrap analysis of stable clusters in resting-state fMRI
NeuroImage
(2010)

P. Bellec et al.

The pipeline system for Octave and Matlab (PSOM): a lightweight scripting framework and execution engine for scientific workflows

Front. Neuroinforma.

(2012)

P. Bellec et al.

A neuroimaging analysis kit for Matlab and Octave

Y. Benjamini et al.

Controlling the false-discovery rate: a practical and powerful approach to multiple testing

J. R. Stat. Soc. Ser. B

(1995)

Y. Benjamini et al.

The control of the false discovery rate in multiple testing under dependency

Ann. Stat.

(2001)

B.B. Biswal et al.

Toward discovery science of human brain function

Proc. Natl. Acad. Sci. U. S. A.

(2010)

T. Blumensath et al.

Spatially constrained hierarchical parcellation of the brain with resting-state fMRI

NeuroImage

(2013)

V.D. Calhoun et al.

Functional brain networks in schizophrenia: a review

Front. Hum. Neurosci.

(2009)

F.X. Castellanos et al.

Clinical applications of the functional connectome

NeuroImage

(2013)

O. Collignon et al.

Functional specialization for auditory-spatial processing in the occipital cortex of congenitally blind humans

Proc. Natl. Acad. Sci. U. S. A.

(2011)

D.L. Collins et al.

Animal: validation and applications of nonlinear registration-based segmentation

Int. J. Pattern Recognit. Artif. Intell.

(1997)

R.C. Craddock et al.

A whole brain fMRI atlas generated via spatially constrained spectral clustering

Hum. Brain Mapp.

(2012)

B. Efron

Simultaneous inference: when should hypothesis testing problems be combined?

Ann. Appl. Stat.

(2008)

V. Fonov et al.

Unbiased average age-appropriate atlases for pediatric studies

NeuroImage

(2011)

A. Fornito et al.

Schizophrenia, neuroimaging and connectomics

NeuroImage

(2012)

M.D. Fox et al.

Clinical applications of resting state functional connectivity

Front. Syst. Neurosci.

(2010)

F. Giove et al.

Images-based suppression of unwanted global signals in resting-state functional connectivity studies

Magn. Reson. Imaging

(2009)

E.M. Gordon et al.

Generation and evaluation of a cortical area parcellation from resting-state correlations

Cereb. Cortex

(2014)

Cited by (35)

A deep connectome learning network using graph convolution for connectome-disease association study
2023, Neural Networks
Multivariate analysis approaches provide insights into the identification of phenotype associations in brain connectome data. In recent years, deep learning methods including convolutional neural network (CNN) and graph neural network (GNN), have shifted the development of connectome-wide association studies (CWAS) and made breakthroughs for connectome representation learning by leveraging deep embedded features. However, most existing studies remain limited by potentially ignoring the exploration of region-specific features, which play a key role in distinguishing brain disorders with high intra-class variations, such as autism spectrum disorder (ASD), and attention deficit hyperactivity disorder (ADHD). Here, we propose a multivariate distance-based connectome network (MDCN) that addresses the local specificity problem by efficient parcellation-wise learning, as well as associating population and parcellation dependencies to map individual differences. The approach incorporating an explainable method, parcellation-wise gradient and class activation map (p-GradCAM), is feasible for identifying individual patterns of interest and pinpointing connectome associations with diseases. We demonstrate the utility of our method on two largely aggregated multicenter public datasets by distinguishing ASD and ADHD from healthy controls and assessing their associations with underlying diseases. Extensive experiments have demonstrated the superiority of MDCN in classification and interpretation, where MDCN outperformed competitive state-of-the-art methods and achieved a high proportion of overlap with previous findings. As a CWAS-guided deep learning method, our proposed MDCN framework may narrow the bridge between deep learning and CWAS approaches, and provide new insights for connectome-wide association studies.
Genetic Heterogeneity Shapes Brain Connectivity in Psychiatry
2023, Biological Psychiatry
Polygenicity and genetic heterogeneity pose great challenges for studying psychiatric conditions. Genetically informed approaches have been implemented in neuroimaging studies to address this issue. However, the effects on functional connectivity of rare and common genetic risks for psychiatric disorders are largely unknown. Our objectives were to estimate and compare the effect sizes on brain connectivity of psychiatric genomic risk factors with various levels of complexity: oligogenic copy number variants (CNVs), multigenic CNVs, and polygenic risk scores (PRSs) as well as idiopathic psychiatric conditions and traits.
Resting-state functional magnetic resonance imaging data were processed using the same pipeline across 9 datasets. Twenty-nine connectome-wide association studies were performed to characterize the effects of 15 CNVs (1003 carriers), 7 PRSs, 4 idiopathic psychiatric conditions (1022 individuals with autism, schizophrenia, bipolar conditions, or attention-deficit/hyperactivity disorder), and 2 traits (31,424 unaffected control subjects).
Effect sizes on connectivity were largest for psychiatric CNVs (estimates: 0.2–0.65 z score), followed by psychiatric conditions (0.15–0.42), neuroticism and fluid intelligence (0.02–0.03), and PRSs (0.01–0.02). Effect sizes of CNVs on connectivity were correlated to their effects on cognition and risk for disease (r = 0.9, p = 5.93 × 10⁻⁶). However, effect sizes of CNVs adjusted for the number of genes significantly decreased from small oligogenic to large multigenic CNVs (r = −0.88, p = 8.78 × 10⁻⁶). PRSs had disproportionately low effect sizes on connectivity compared with CNVs conferring similar risk for disease.
Heterogeneity and polygenicity affect our ability to detect brain connectivity alterations underlying psychiatric manifestations.
Using connectomics for predictive assessment of brain parcellations
2021, NeuroImage
Citation Excerpt :
Finally, we also observed that, for the SC and intra-subject FC predictions, accounting for the cortical surface structure using Geodesic distances generally performed better than the corresponding use of Euclidean distances. The recent popularity of data-driven in-vivo parcellation approaches, as exemplified by Behrens et al. (2003a); Bellec et al. (2015, 2006); Blumensath et al. (2013); Clos et al. (2013); Craddock et al. (2013); Eickhoff et al. (2015); Fan et al. (2016); Glasser et al. (2016); Shen et al. (2013); Thirion et al. (2014); Yeo et al. (2011), highlights the growing demand for alternatives to conventional atlases, e.g. Desikan et al. (2006); Destrieux et al. (2010); Lancaster et al. (2000); Nieuwenhuys (2013); Nieuwenhuys et al. (2015); Tzourio-Mazoyer et al. (2002). However, it is necessary to recall that such data-driven methods are primarily uni-modal (though with some exceptions, e.g. Fan et al. (2016); Glasser et al. (2016); Parisot et al. (2017)), and as such their resultant parcellations do not conform to the traditional definition of cortical areas, where within-area homogeneity and inter-area uniqueness are required to be congruent across the three commonly-employed modalities of structure, function and connectivity Amunts and Zilles (2015); Felleman and Van Essen (1991); Van Essen et al. (1992).
The organization of the human brain remains elusive, yet is of great importance to the mechanisms of integrative brain function. At the macroscale, its structural and functional interpretation is conventionally assessed at the level of cortical units. However, the definition and validation of such cortical parcellations are problematic due to the absence of a true gold standard. We propose a framework for quantitative evaluation of brain parcellations via statistical prediction of connectomics data. Specifically, we evaluate the extent in which the network representation at the level of cortical units (defined as parcels) accounts for high-resolution brain connectivity. Herein, we assess the pertinence and comparative ranking of ten existing parcellation atlases to account for functional (FC) and structural connectivity (SC) data based on data from the Human Connectome Project (HCP), and compare them to data-driven as well as spatially-homogeneous geometric parcellations including geodesic parcellations with similar size distributions as the atlases. We find substantial discrepancy in parcellation structures that well characterize FC and SC and differences in what well represents an individual’s functional connectome when compared against the FC structure that is preserved across individuals. Surprisingly, simple spatial homogenous parcellations generally provide good representations of both FC and SC, but are inferior when their within-parcellation distribution of individual parcel sizes is matched to that of a valid atlas. This suggests that the choice of fine grained and coarse representations used by existing atlases are important. However, we find that resolution is more critical than the exact border location of parcels.
Tractography density affects whole-brain structural architecture and resting-state dynamical modeling
2021, NeuroImage
Dynamical modeling of the resting-state brain dynamics essentially relies on the empirical neuroimaging data utilized for the model derivation and validation. There is however still no standardized data processing for magnetic resonance imaging pipelines and the structural and functional connectomes involved in the models. In this study, we thus address how the parameters of diffusion-weighted data processing for structural connectivity (SC) can influence the validation results of the whole-brain mathematical models informed by SC. For this, we introduce a set of simulation conditions including the varying number of total streamlines of the whole-brain tractography (WBT) used for extraction of SC, cortical parcellations based on functional and anatomical brain properties and distinct model fitting modalities. The main objective of this study is to explore how the quality of the model validation can vary across the considered simulation conditions. We observed that the graph-theoretical network properties of structural connectome can be affected by varying tractography density and strongly relate to the model performance. We also found that the optimal number of the total streamlines of WBT can vary for different brain atlases. Consequently, we suggest a way how to improve the model performance based on the network properties and the optimal parameter configurations from multiple WBT conditions. Furthermore, the population of subjects can be stratified into subgroups with divergent behaviors induced by the varying WBT density such that different recommendations can be made with respect to the data processing for individual subjects and brain parcellations.
Brain graph super-resolution for boosting neurological disorder diagnosis using unsupervised multi-topology connectional brain template learning
2020, Medical Image Analysis
Citation Excerpt :
This process has two severe limitations: (1) it is very time consuming (it can last up to a few hours when using refined deformable registration), and (2) it is sensitive to the deployed pre-processing methods including registration, atlas-to-image label propagation, which might preclude the comparison of healthy and disordered brain connectomes produced using different MRI pre-processing pipelines. In fact, parcellation methods use predefined coarsely parcellated brain atlases such as Automated Anatomical Labelling (AAL) with 90 ROIs, which define the size or resolution of the resulting brain connectomes (Bellec et al., 2015). For instance, the commonly used statistical parametric mapping (SPM) software (Ashburner, 2012) for generating functional connectomes performs automated anatomical parcellation of brain MR images.
Existing graph analysis techniques generally focus on decreasing the dimensionality of graph data (i.e., removing nodes, edges, or both) in diverse predictive learning tasks in pattern recognition, computer vision, and medical data analysis such as dimensionality reduction, filtering and embedding techniques. However, graph super-resolution is strikingly lacking, i.e., the concept of super-resolving low-resolution (LR) graphs with n_r nodes into high-resolution graphs (HR) with $n_{r^{'}} > n_{r}$ nodes. Particularly, learning how to automatically generate HR brain connectomes, without resorting to the computationally expensive MRI processing steps such as image registration and parcellation, remains unexplored. To fill this gap, we propose the first technique to super-resolve undirected fully connected graphs with application to brain connectomes. First, we root our brain graph super-resolution (BGSR) framework in learning how to estimate a centered LR population-based brain graph representation, coined as connectional brain template (CBT), acting as a proxy in the target BGSR task. Specifically, we hypothesize that the estimation of a well-representative and centered CBT would help better capture the individuality of each LR brain graph via its residual distance from the population-based CBT. This will eventually allow an accurate identification of the most similar individual graphs to a new testing graph in the LR domain for the target prediction task. Second, we leverage the estimated LR CBT (i.e., population mean) to derive residual LR brain graphs, capturing the deviation of all subjects from the estimated CBT. Third, we learn multi-topology LR graph manifolds using different graph topological measurements (e.g., degree, closeness, betweenness) by estimating residual LR similarity matrices modeling the relationship between pairs of residual LR graphs. These are then fused so we can effectively identify for each testing LR subject its most K similar training LR graphs. Last, the missing testing HR graph is predicted by averaging the HR graphs of the K selected training subjects. Predicted HR from LR functional brain graphs boosted classification results for autistic subjects by 16.48% compared with LR functional graphs.
Older adults benefit from more widespread brain network integration during working memory
2020, NeuroImage
Neuroimaging evidence suggests that the aging brain relies on a more distributed set of cortical regions than younger adults in order to maintain successful levels of performance during demanding cognitive tasks. However, it remains unclear how task demands give rise to this age-related expansion in cortical networks. To investigate this issue, functional magnetic resonance imaging was used to measure univariate activity, network connectivity, and cognitive performance in younger and older adults during a working memory (WM) task. Here, individuals performed a WM task in which they held letters online while reordering them alphabetically. WM load was titrated to obtain four individualized difficulty levels with different set sizes. Network integration—defined as the ratio of within-versus between-network connectivity—was linked to individual differences in WM capacity. The study yielded three main findings. First, as task difficulty increased, network integration decreased in younger adults, whereas it increased in older adults. Second, age-related increases in network integration were driven by increases in right hemisphere connectivity to both left and right cortical regions, a finding that helps to reconcile existing theories of compensatory recruitment in aging. Lastly, older adults with higher WM capacity demonstrated higher levels of network integration in the most difficult task condition. These results shed light on the mechanisms of age-related network reorganization by demonstrating that changes in network connectivity may act as an adaptive form of compensation, with older adults recruiting a more distributed cortical network as task demands increase.

View all citing articles on Scopus

View full text

Impact of the resolution of brain parcels on connectome-wide association studies in fMRI

Highlights

Abstract

Introduction

Section snippets

Functional connectome

Data-generating procedure

Data-generating procedure

Participants

Specificity in multiresolution analysis

Conclusion

Acknowledgments

Group-ICA model order highlights patterns of functional brain connectivity

Front. Syst. Neurosci.

The CIVET image-processing environment: a fully automated comprehensive pipeline for anatomical neuroimaging research

The resting human brain and motor learning

Curr. Biol.

Maintaining vs. enhancing motor sequence memories: respective roles of striatal and hippocampal systems

NeuroImage

Quantitative Methods for Current Environmental Issues

Characterizing thalamo-cortical disturbances in schizophrenia and bipolar illness

Cereb. Cortex

Resting-state functional MR imaging: a new window to the brain

Radiology

Mining the hierarchy of resting-state brain networks: selection of representative clusters in a multiscale structure

Identification of large-scale networks in the brain using fMRI

NeuroImage

Multi-level bootstrap analysis of stable clusters in resting-state fMRI

NeuroImage

The pipeline system for Octave and Matlab (PSOM): a lightweight scripting framework and execution engine for scientific workflows

Front. Neuroinforma.

A neuroimaging analysis kit for Matlab and Octave

Controlling the false-discovery rate: a practical and powerful approach to multiple testing

J. R. Stat. Soc. Ser. B

The control of the false discovery rate in multiple testing under dependency

Ann. Stat.

Toward discovery science of human brain function

Proc. Natl. Acad. Sci. U. S. A.

Spatially constrained hierarchical parcellation of the brain with resting-state fMRI

NeuroImage

Functional brain networks in schizophrenia: a review

Front. Hum. Neurosci.

Clinical applications of the functional connectome

NeuroImage

Functional specialization for auditory-spatial processing in the occipital cortex of congenitally blind humans

Proc. Natl. Acad. Sci. U. S. A.

Animal: validation and applications of nonlinear registration-based segmentation

Int. J. Pattern Recognit. Artif. Intell.

A whole brain fMRI atlas generated via spatially constrained spectral clustering

Hum. Brain Mapp.

Simultaneous inference: when should hypothesis testing problems be combined?

Ann. Appl. Stat.

Unbiased average age-appropriate atlases for pediatric studies

NeuroImage

Schizophrenia, neuroimaging and connectomics

NeuroImage

Clinical applications of resting state functional connectivity

Front. Syst. Neurosci.

Images-based suppression of unwanted global signals in resting-state functional connectivity studies

Magn. Reson. Imaging

Generation and evaluation of a cortical area parcellation from resting-state correlations

Cereb. Cortex