Functional connectomics from resting-state fMRI

May combine two claims.

The histogram of edge-strength mean group differences is centred at zero, with no gross asymmetry.

“Indeed, the histogram of edge-strength mean group differences is centred at zero, with no gross asymmetry.”

May combine two claims.

Average between-subject network similarity correlations were FreeSurfer=0.613, MSMfolding=0.618 and MSMmultimodal=0.649.

“We found the following average correlation values for the 3 alignment methods: FreeSurfer=0.613, MSMfolding=0.618 and MSMmultimodal=0.649 (Figure 5A).”

May combine two claims.

The node-pair contributing most strongly to the regression against fluid intelligence has only a weak average connection in the population but varies across subjects in a way that correlates with fluid intelligence.

“Indeed, the node-pair which contributes most strongly to the regression against fluid intelligence (shown in more detail in B) has only a weak connection on average in the population - but it varies across subjects in a way that correlates with fluid intelligence.”

May combine two claims.

The three alignment methods gave sex classification accuracies of FreeSurfer=78%, MSMfolding=82% and MSMmultimodal=87%.

“The 3 alignment methods described above gave classification accuracies of: FreeSurfer=78%, MSMfolding=82% and MSMmultimodal=87%.”

Probably not a substantial claim: may not pass S3 (load-bearing).

The 24 edges having the strongest weights in the regression against fluid intelligence are shown.

“The 24 edges (node-pairs) having the strongest weights in the regression are shown;”

May combine two claims.

In both cases, strong spatial correspondence to specific task activation maps from HCP task-fMRI datasets was shown, helping interpretation for functional modes identified from resting-state data.

“In both cases we show strong spatial correspondence to specific task activation maps from the HCP task-fMRI datasets, thus helping to provide interpretation for the functional modes identified from the resting-state data.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Figure 9 shows two example results from feeding 98 nodes' timeseries into temporal-ICA, analysing HCP rfMRI data containing more time points (>600,000, concatenating data from 131 subjects) than spatial points (~90,000 grayordinates).

“Figure 9 shows 2 example results from feeding 98 nodes’ timeseries into temporal-ICA, where we analysed the high quality HCP rfMRI data containing more time points (>600,000, concatenating data from 131 subjects) than spatial points (~90,000 grayordinates).”

May combine two claims.

Different resting-state networks have distinct timecourses, while each network has a relatively consistent timecourse across its set of involved regions.

“These patterns can be distinguished from each other because, while each has a relatively consistent timecourse across its set of involved regions, the different networks have distinct timecourses [2].”

• cites [2] Beckmann CF 2005 Investigations into Resting-State Connectivity using Independent Component Analysis Philosophical Transactions of the Royal Society 360 1001 1013 PMC1854918 16087444 sourceNot checked yet

  Not checked in the free run.

May combine two claims.

Resting-state networks persist even during sleep and under anaesthesia.

“These “resting-state networks” (RSNs) persist even during sleep and under anaesthesia [3]”

• cites [3] Picchioni D 2013 Sleep and the functional connectome Neuroimage 80 387 396 23707592 PMC3733088 sourceNot checked yet

  Not checked in the free run.

May combine two claims.

Seed-based correlation contains fine spatial detail, but only informs about average correlation with the selected seed(s).

“Such an approach contains fine spatial detail, but only informs about average correlation with the selected seed(s).”

May combine two claims.

Effective artefact removal is particularly important for resting-state analyses, which rely fundamentally on correlations between different voxels' timeseries, because these correlations will be corrupted by artefacts that span multiple voxels.

“Effective artefact removal is particularly important for resting-state analyses, which rely fundamentally on correlations between different voxels’ timeseries, as these will be corrupted by artefacts that span multiple voxels.”

Classified as probably not a substantial claim.

A high-dimensional parcellation into many nodes allows a richer analysis of the network connections by shifting the emphasis from large-scale maps with fine spatial detail into a network description of nodes and edges.

“a high-dimensional parcellation into many nodes (potentially hundreds) allows a richer analysis of the network connections – by shifting the emphasis from large-scale maps with fine spatial detail into a network description of “nodes” and “edges”, new information can be obtained.”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

The sluggish response of the brain's haemodynamics to neural activity ultimately places a limit on the usefulness of further improvements in temporal resolution.

“although the sluggish response of the brain’s haemodynamics (to neural activity) will ultimately place a limit on the usefulness of further improvements in temporal resolution.”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

Task-fMRI has the advantage of fitting a pre-specified temporal model, which provides greater robustness against artefactual influences.

“(In contrast, task-fMRI has the advantage of fitting a pre-specified temporal model, which provides greater robustness against artefactual influences.)”

Probably not a substantial claim: may not pass S4 (based).

Accurate alignment of functionally corresponding cortical parcels is an exceedingly challenging problem, largely owing to individual variability of cortical folding patterns and variability of functional parcel locations relative to these folds.

“accurate alignment of functionally corresponding cortical parcels is an exceedingly challenging problem, largely owing to individual variability of cortical folding patterns along with variability of functional parcel locations relative to these folds”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

To combine or compare connectomes across subjects, it is important to strive for the same parcellation in each subject; comparisons between two network models are inherently flawed if they are derived from non-corresponding sets of nodes.

“In order to combine or compare connectomes across subjects, it is important to strive for “the same” parcellation in each subject – comparisons between two network models are inherently flawed if they are derived from non-corresponding sets of nodes.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Multiband accelerated EPI enables major improvements in spatial and/or temporal resolution, for example acquiring data with 2mm spatial resolution in less than a second.

““multiband accelerated EPI” acquires multiple slices simultaneously [39, 40]. Such approaches enable major improvements in spatial and/or temporal resolution, for example acquiring data with 2mm spatial resolution in less than a second.”

• cites [39] Moeller S 2010 Multiband multislice GE-EPI at 7 tesla, with 16-fold acceleration using partial parallel imaging with application to high spatial and temporal whole-brain fMRI Magn Reson Med 63 1144 1153 20432285 PMC2906244 sourceNot checked yet

  Not checked in the free run.

• cites [40] Setsompop K 2012 Blipped-controlled aliasing in parallel imaging for simultaneous multislice echo planer imaging with reduced g-factor penalty Magnetic Resonance in Medicine 67 1210 1224 21858868 PMC3323676 sourceNot checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing).

Higher temporal resolution of fMRI data can increase the information content of the data, for example in terms of reflecting the richness of neural dynamics.

“and also increase the information content of the data (e.g., in terms of reflecting the richness of the neural dynamics) [41, 42]”

• cites [41] Feinberg DA 2010 Multiplexed Echo Planar Imaging for Sub-second Whole Brain FMRI and Fast Diffusion Imaging PLoS ONE 5 e15710 21187930 PMC3004955 sourceNot checked yet

  Not checked in the free run.

• cites [42] Smith SM 2012 Temporally-independent functional modes of spontaneous brain activity Proc Natl Acad Sci USA (PNAS) 109 3131 3136 PMC3286957 22323591 sourceNot checked yet

  Not checked in the free run.

Classified as probably not a substantial claim.

A detailed nodes+edges network model may reveal which nodes, as sub-parts of large-scale networks, are responsible for the correlations seen between the larger-scale networks.

“a detailed nodes+edges network model may reveal which nodes (sub-parts of the large-scale networks) are responsible for the correlations seen between the larger-scale networks.”

May combine two claims.

Correlation cannot reveal anything about causality, or even whether connectivity is direct versus indirect.

“Furthermore, correlation cannot reveal anything about causality, or even whether connectivity is direct vs indirect [49].”

• cites [49] Marrelec G 2006 Partial correlation for functional brain interactivity investigation in functional MRI NeuroImage 32 228 237 16777436 sourceNot checked yet

  Not checked in the free run.

May combine two claims.

Low-dimensional ICA generates a richer description of multiple networks in the brain, but at the expense of no longer dictating in advance the regions with which the connectivity map(s) are related.

“This therefore generates a richer description of multiple networks in the brain, but at the expense of no longer dictating in advance the regions with which the connectivity map(s) are related.”

May combine two claims.

In a 3-node network A→B→C, all three nodes' timeseries will be correlated, so correlation will incorrectly estimate a fully-connected network including an A-C connection.

“For example, in a 3-node network A→B→C, all three nodes’ timeseries will be correlated, so correlation will incorrectly estimate a fully-connected network (including an A-C connection).”

May combine two claims.

DCM, originally developed for task data, has recently been extended to allow modelling of resting-state data and to search across a wide range of possible network matrix models.

“while DCM, originally developed for task data, has recently been extended [54] to allow modelling of resting-state data, and to search across a wide range of possible network matrix models”

• cites [54] Friston KJ 2011 Network discovery with DCM NeuroImage 56 1202 1221 21182971 PMC3094760 sourceNot checked yet

  Not checked in the free run.

May combine two claims.

L1-norm regularisation of the inverse covariance matrix shrinks the estimated values in the partial correlation matrix, so that very small, noisy values are forced to zero, and all other values are better estimated.

“L1-norm regularisation of the inverse covariance matrix [55] shrinks the estimated values in the partial correlation matrix, so that very small, noisy values are forced to zero, and all other values are better estimated.”

• cites [55] Friedman J 2008 Sparse inverse covariance estimation with the Graphical Lasso Biostat 9 432 441 PMC3019769 18079126 sourceNot checked yet

  Not checked in the free run.

May combine two claims.

Graph-theoretic techniques are dependent on accurate network modelling at the lower level, and the abstraction of these metrics from the underlying data makes it challenging to evaluate results.

“Naturally, these techniques are dependent on accurate network modelling at the lower level (carefully constructed network matrices); the abstraction of these metrics from the underlying data makes it challenging to evaluate results”

May combine two claims.

The HCP is generating a detailed in vivo mapping of functional connectivity in a large cohort of over 1000 subjects, and is making these datasets freely available for use by the neuroimaging community, with over 200 subjects' datasets already acquired and publicly released.

“The HCP is generating a detailed in vivo mapping of functional connectivity in a large cohort (over 1000 subjects), and is making these datasets freely available for use by the neuroimaging community (over 200 subjects’ datasets are already acquired and publicly released – available via humanconnectome.org).”

May combine two claims.

Bayes nets model the full set of covariances or conditional dependencies between all nodes' timeseries and show some promise.

“Bayes nets model the full set of covariances (or conditional dependencies) between all nodes’ timeseries [59] and show some promise.”

• cites [59] Ramsey JD 2010 Six problems for causal inference from fMRI NeuroImage 49 1545 1558 19747552 sourceNot checked yet

  Not checked in the free run.

May combine two claims.

It is frequently the thresholded full correlation matrix that is input into graph theoretical analyses, rather than an estimate of direct network connections, and interpretations of graph-theory-based measures are often questionable as a result.

“Another concern is that it is frequently the (thresholded) full correlation matrix that is input into graph theoretical analyses, rather than an estimate of direct network connections; interpretations of graph-theory-based measures (such as communication path length) are often, as a result, somewhat questionable.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Many methods, such as correlation, give no directionality information at all.

“Many methods, such as correlation, give no directionality information at all.”

May combine two claims.

The Thousand Functional Connectomes dataset has greater heterogeneity of scanning parameters and lower data quality than the HCP.

“This growing dataset comes with the caveat of having greater heterogeneity of scanning parameters (and lower data quality) than the HCP”

May combine two claims.

Partial correlation aims to more accurately estimate the direct connections, though not their directionalities.

“another simple estimation method, partial correlation, aims to more accurately estimate the “direct” connections (though not their directionalities).”

Probably not a substantial claim: may not pass S3 (load-bearing).

The MGH-UCLA HCP is using extremely powerful MRI gradients to generate leading-edge 3T diffusion MRI data.

“The MGH-UCLA HCP is using extremely powerful MRI gradients to generate leading-edge 3T diffusion MRI data.”

• cites [16] Setsompop K 2013 Pushing the limits of in vivo diffusion MRI for the Human Connectome Project Neuroimage 80 220 233 23707579 PMC3725309 sourceNot checked yet

  Not checked in the free run.

Classified as probably not a substantial claim.

It cannot be assumed that the brain's connections are static.

“it cannot be assumed that the brain’s connections are static.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

The WU-Minn HCP is generating high quality and high resolution 3T rfMRI and diffusion MRI data from over 1000 healthy adults, as well as task-fMRI, genotyping and behavioural data.

“the WashU-UMinn-Oxford (“WU-Minn”) HCP is generating high quality and high resolution 3T rfMRI and diffusion MRI data from over 1000 healthy adults, as well as task-fMRI, genotyping and behavioural data.”

• cites [67] Van Essen DC 2013 The WU-Minn Human Connectome Project: An Overview NeuroImage 80 62 79 23684880 PMC3724347 sourceNot checked yet

  Not checked in the free run.

May combine two claims.

In the 3-node network example, regression of B out of A and C removes the correlation between A and C, and hence the spurious third edge A-C.

“In this case, regression of B out of A and C removes the correlation between A and C, and hence the spurious third edge (A–C).”

May combine two claims.

Simpler network modelling methods are in general more robust with respect to fitting the model to the data and faster to compute.

“The simpler methods are in general more robust (with respect to fitting the model to the data), and faster to compute.”

May combine two claims.

Simpler network modelling methods do not require the scope of possible network models to be pre-constrained, making them computationally feasible for attempting network discovery.

“they do not require the scope of possible network models to be pre-constrained, making them computationally feasible for attempting network discovery.”

May combine two claims.

HCP subjects are drawn from a population of healthy adult twins and their non-twin siblings in the age range of 22-35 years, thus aiding the study of the influence of heritability and environmental factors on the connectome.

“Subjects are drawn from a population of healthy adult twins and their non-twin siblings (in the age range of 22–35 years), thus aiding the study of the influence of heritability and environmental factors on the connectome.”

May combine two claims.

Temporal lag between pairs of timeseries is often used to estimate directionality, though it is arguably too confounded by slow and variable haemodynamics to be useful with fMRI data, and is better-suited to electrophysiological modalities.

“Temporal lag between pairs of timeseries [58] is often used, though it is arguably too confounded by the slow and variable haemodynamics to be useful with fMRI data [56], and is better-suited to electrophysiological modalities.”

• cites [56] Smith SM 2011 Network Modelling Methods for FMRI NeuroImage 54 875 891 20817103 sourceNot checked yet

  Not checked in the free run.

May combine two claims.

Estimation of partial correlation effectively involves inverting the full correlation matrix, a process which is often quite unstable, depending on the quality of the data and the number of original timepoints.

“The estimation of partial correlation effectively involves inverting the full correlation matrix, a process which is often quite unstable, depending on the quality of the data and the number of original timepoints.”

May combine two claims.

Causality estimation when the brain is at rest may involve many different functional processes mixed together, and would be estimated as the average relative causality over all possible spontaneous fluctuations.

“causality estimation when the brain is “at rest” may involve many different functional processes mixed together, and would be estimated as the average “relative causality” over all possible spontaneous fluctuations.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

The Thousand Functional Connectomes and related projects are seeking to generate even larger numbers of subjects than the HCP, bringing together data from a wide range of imaging studies and scanners, and covering a wider range of subject groups, including a range of pathologies.

“The Thousand Functional Connectomes (and related) projects [74, 75], led by Mike Milham, are seeking to generate even larger numbers of subjects than the HCP, bringing together data from a wide range of imaging studies and scanners, and covering a wider range of subject groups, including a range of pathologies.”

• cites [74] Biswal BB 2010 Towards discovery science of human brain function Proc Natl Acad Sci USA (PNAS) 107 4734 4739 PMC2842060 20176931 sourceNot checked yet

  Not checked in the free run.

• cites [75] Yan CG 2013 Standardizing the intrinsic brain: Towards robust measurement of inter-individual variation in 1000 functional connectomes Neuroimage 80 246 262 23631983 PMC4074397 sourceNot checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing).

FreeSurfer provides one way to improve intersubject alignment of features on the cortical surface using folding features.

“Once each subject’s cortical surface has been estimated from the structural data using FreeSurfer, FreeSurfer also provides one way to improve intersubject alignment of features on the surface.”

May combine two claims.

Partial correlation matrices being as strong and consistent across subjects as full correlation is unusual and indicates high quality of the data and a large number of timepoints in each subject's rfMRI dataset.

“For the partial correlation matrices to be as strong and consistent across subjects as the full correlation is unusual and is indicative of the high quality of the data and large number of timepoints in each subject’s rfMRI dataset.”

May combine two claims.

Multimodal MSM improves the alignment of function across subjects, as it improves the accuracy with which derived network matrices can predict sex, and this supports that sex prediction is primarily driven by network connection strength and not systematic functional misalignments.

“This suggests that multimodal MSM improves the alignment of function across subjects (as it improves the accuracy with which the derived network matrices can predict sex), and also supports the hypothesis that this sex prediction is indeed primarily being driven by network connection strength and not (for example) by systematic functional misalignments across (groups of) subjects.”

Probably not a substantial claim: may not pass S3 (load-bearing).

This kind of analysis has previously been used to predict subjects' psychiatric condition and cognitive state.

“This kind of analysis has previously been used (albeit from full correlation network matrices) to predict subjects’ psychiatric condition [77] and cognitive state [78];”

• cites [77] Craddock RC 2009 Disease state prediction from resting state functional connectivity Magnetic Resonance in Medicine 62 1619 1628 19859933 PMC3749911 sourceNot checked yet

  Not checked in the free run.

• cites [78] Shirer WR 2012 Decoding subject-driven cognitive states with whole-brain connectivity patterns Cereb Cortex 22 158 165 21616982 PMC3236795 sourceNot checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

An important issue is to distinguish between changes in correlation due to nodes being part of multiple overlapping functional networks versus internal connections within any given network being non-constant.

“One important issue is to distinguish between changes in correlation due to some nodes being part of multiple overlapping functional networks, vs. the internal connections within any given network being non-constant.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Multimodal Surface Matching (MSM) is an alternative surface warping approach developed by the authors.

“We have recently developed an alternative surface warping approach, Multimodal Surface Matching (MSM) [70].”

• cites [70] Robinson E 2013 Multimodal Surface Matching: Fast and Generalisable Cortical Registration using Discrete Optimisation Springer Berlin Heidelberg 24683992Not checked yet

  Not checked in the free run.

May combine two claims.

A problem can arise when applying a window length that does not encompass at least several cycles of resting fluctuations, because the rfMRI signal is dominated by low frequencies (~0.015Hz), and a sliding window containing a fraction of a cycle is expected to show correlations appearing to fluctuate wildly over time even if the underlying network structure is stationary.

“In addition, a problem can arise when applying a window length that does not encompass (at least) several cycles of the resting fluctuations, as the rfMRI signal is dominated by low frequencies (~0.015Hz), and a sliding window containing a fraction of a cycle is expected to show correlations appearing to fluctuate wildly over time, even if the underlying network structure is stationary.”

• AssumptionRests on an assumption: The rfMRI signal is dominated by low frequencies (~0.015Hz).

Probably not a substantial claim: may not pass S3 (load-bearing).

FIX achieves classification accuracy on HCP data of over 99%.

“FIX achieves classification accuracy on HCP data of over 99%.”

• cites Salimi-Khorshidi, unpublished manuscriptNot checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

No single parcellation technique has emerged as the best approach, and investigations are ongoing.

“At this point, no single parcellation technique has emerged as the best approach, and investigations are ongoing.”

May combine two claims.

Better alignments should yield more accurate network matrices, because functional parcels defined at the group level should be more accurately aligned to the true subject-specific versions of those same parcels.

“Therefore, better alignments should yield more accurate network matrices, because the functional parcels defined at the group level should be more accurately aligned to the true subject-specific versions of those same parcels.”

• AssumptionRests on an assumption: The functional parcels defined at the group level correspond to true subject-specific versions of those same parcels.

Probably not a substantial claim: may not pass S3 (load-bearing).

Nonstationarities in correlation have been studied utilising either wavelet decompositions or sliding-window correlation.

“For example, nonstationarities in correlation have been studied utilising either wavelet decompositions or sliding-window correlation [88].”

• cites [88] Hutchison RM 2013 Dynamic functional connectivity: Promise, issues, and interpretations Neuroimage 80 360 378 23707587 PMC3807588Not checked yet

  Not checked in the free run.

May combine two claims.

An increase in cross-subject network similarity is not conclusive proof of a better alignment method; it might indicate over-strong application of prior information.

“However, although this is encouraging, an increase in cross-subject network similarity is not conclusive proof of a better alignment method; for example, it might indicate over-strong application of prior information (e.g., in the extreme, setting all subjects to a constant network), which would reduce sensitivity when attempting to relate the network matrices to other measures.”

May combine two claims.

A limitation is that multiple distinct functional networks might overlap each other in space and also vary over time in how they interact with each other.

“A limitation, however, is that multiple distinct functional networks might overlap each other in space and also vary over time in how they interact with each other.”

May combine two claims.

rfMRI does not provide the more direct view of neuronal activity and information flow that can be obtained from electrophysiological techniques, or inform us regarding the biochemistry of brain connectivity.

“It does not provide the more direct view of neuronal activity and information flow that can be obtained from electrophysiological techniques, or inform us regarding the biochemistry of brain connectivity.”

May combine two claims.

[42] showed several temporal functional modes with plausible functional interpretability and large spatial overlap; default-mode network patterns can be decomposed into several modes of distinct but overlapping spatial layout.

“[42] showed several such modes, with plausible functional interpretability and large amounts of spatial overlap; for example, the patterns of average correlation and anti-correlation associated with the default-mode network can be decomposed into several modes of distinct but overlapping spatial layout (and hence presumably distinct function).”

• cites [42] Smith SM 2012 Temporally-independent functional modes of spontaneous brain activity Proc Natl Acad Sci USA (PNAS) 109 3131 3136 PMC3286957 22323591 sourceNot checked yet

  Not checked in the free run.

May combine two claims.

Evidence indicates that rfMRI signal includes complex spatiotemporal patterns of spontaneous activity that propagate across the brain and are therefore not well modelled by spatially fixed network modelling approaches.

“A related challenge is raised by evidence that rfMRI signal includes complex spatiotemporal patterns of spontaneous activity that propagate across the brain [92], and are therefore not at all well modelled by spatially fixed network modelling approaches.”

• cites [92] Majeed W 2010 Spatiotemporal dynamics of low frequency BOLD fluctuations in rats and humans Neuroimage 54 1140 1150 20728554 PMC2997178Not checked yet

  Not checked in the free run.

May combine two claims.

rfMRI provides the ability to create a richly detailed mapping of the functional connectome, at the millimetre scale, covering the whole brain, with in vivo, non-invasive imaging across thousands of subjects.

“However, rfMRI does provide us with something valuable and unique – the ability to create a richly detailed mapping of the functional connectome, at the millimetre scale, covering the whole brain, with in vivo, non-invasive imaging across thousands of subjects.”

May combine two claims.

rfMRI does not tell us directly about the microscopic neuronal-level structural connectome, or even about the macroscopic-level structural connectome such as is inferred from diffusion MRI.

“It does not tell us directly about the microscopic, neuronal-level, structural connectome, or even about the macroscopic-level structural connectome such as is inferred from diffusion MRI.”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

Another limitation with temporal-ICA is the assumption that connections within a given mode are not changing in strength over time; this is an oversimplistic view of brain networks.

“Another limitation with temporal-ICA is the assumption that the connections within a given mode are not changing in strength over time; this is surely an oversimplistic view of brain networks.”

May combine two claims.

Two distinct functional modes might be mutually exclusive if they share one node that strongly participates in only one function at any given time; this would imply a negative correlation between the modes' timeseries, and this situation will not be well identified through temporal-ICA.

“For example, two distinct functional modes might in reality be mutually exclusive, if they share one node that strongly participates in only one function at any given time; this would imply a negative correlation between the modes’ timeseries, and this situation will not be well identified through the application of temporal-ICA.”