CoChat
Sign Up
Cited by 253Functional Brain Connectivity StudiesTopological and Geometric Data AnalysisNeural dynamics and brain function

The DOI matches its Crossref record; 69 of the 69 references it cites were found in a registry or an index.

Towards a new approach to reveal dynamical organization of the brain using topological data analysis

Open access
Saggar, Manish, Sporns, Olaf, Gonzalez-Castillo, Javier, Bandettini, Peter A., Carlsson, Gunnar, Glover, Gary, Reiss, Allan L.·2018·Nature Communications
DOI 10.1038/s41467-018-03664-4PMC PMC5895632
Open this paper in CoChat — free accountOpen PDF

Abstract

Little is known about how our brains dynamically adapt for efficient functioning. Most previous work has focused on analyzing changes in co-fluctuations between a set of brain regions over several temporal segments of the data. We argue that by collapsing data in space or time, we stand to lose useful information about the brain’s dynamical organization. Here we use Topological Data Analysis to reveal the overall organization of whole-brain activity maps at a single-participant level—as an interactive representation—without arbitrarily collapsing data in space or time. Using existing multitask fMRI datasets, with the known ground truth about the timing of transitions from one task-block to next, our approach tracks both within- and between-task transitions at a much faster time scale (~4–9 s) than before. The individual differences in the revealed dynamical organization predict task performance. In summary, our approach distills complex brain dynamics into interactive and behaviorally relevant representations.

Overview

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1038/s41467-018-03664-4

Cites

69 of 69 references verified0 citation errors · 0 not checkable

Statistics

Figures

Fig. 1 Application of Mapper to 4D fMRI data. a Pre-processed four-dimensional fMRI data from each participant was fed into the analysis. For each participant, the entire data matrix (i.e., #TRs × #Voxels) was used for analysis. b Non-linear dimensionality reduction was done during the filtering step to project fMRI data into a lower two-dimension set (represented by dimensions (f, g)). c Two-dimensional binning was then performed by dividing the lower-dimension space into smaller bins (determined by the resolution parameter, R) with certain overlap (determined by the gain parameter, G). d Partial clustering was then performed to get a compressed representation by collapsing data into fewer nodes, where each node represents a cluster, and the size of each node depicts the number of data points inside each cluster. e After clustering, nodes that share data points (i.e., time frames in this case) are linked together with an edge to create the final compressed combinatorial representation (or graph)
Fig. 1 Application of Mapper to 4D fMRI data. a Pre-processed four-dimensional fMRI data from each participant was fed into the analysis. For each participant, the entire data matrix (i.e., #TRs × #Voxels) was used for analysis. b Non-linear dimensionality reduction was done during the filtering step to project fMRI data into a lower two-dimension set (represented by dimensions (f, g)). c Two-dimensional binning was then performed by dividing the lower-dimension space into smaller bins (determined by the resolution parameter, R) with certain overlap (determined by the gain parameter, G). d Partial clustering was then performed to get a compressed representation by collapsing data into fewer nodes, where each node represents a cluster, and the size of each node depicts the number of data points inside each cluster. e After clustering, nodes that share data points (i.e., time frames in this case) are linked together with an edge to create the final compressed combinatorial representation (or graph)CC BY · PubMed Central
Fig. 2 Revealing the shape of brain’s dynamical organization. a Depicts the shape graph for one of the representative participants (S01) from the CMP dataset. The shape graph was annotated using colors and pie-chart visualization scheme to depict how the tasks were represented in each shape graph. b Shows a zoomed-in version of the densely connected region of the shape graph to show the use of pie-chart visualization
Fig. 2 Revealing the shape of brain’s dynamical organization. a Depicts the shape graph for one of the representative participants (S01) from the CMP dataset. The shape graph was annotated using colors and pie-chart visualization scheme to depict how the tasks were represented in each shape graph. b Shows a zoomed-in version of the densely connected region of the shape graph to show the use of pie-chart visualizationCC BY · PubMed Central
Fig. 3 Quantifying the mesoscale community structure of shape graphs. a Shows two shape graphs from two representative participants (S14 and S07). As depicted, S14 had a low modularity score, i.e., nodes with different task types are connected to each other without any preference for same task nodes. On the other hand, shape graph from S07 depicts high modularity structure with nodes preferentially connected to other nodes of same task type. b The modularity score (Qmod) was observed to be associated with task performance, such that higher modularity in the shape graph was associated with better task performance
Fig. 3 Quantifying the mesoscale community structure of shape graphs. a Shows two shape graphs from two representative participants (S14 and S07). As depicted, S14 had a low modularity score, i.e., nodes with different task types are connected to each other without any preference for same task nodes. On the other hand, shape graph from S07 depicts high modularity structure with nodes preferentially connected to other nodes of same task type. b The modularity score (Qmod) was observed to be associated with task performance, such that higher modularity in the shape graph was associated with better task performanceCC BY · PubMed Central
Fig. 4 Quantifying the mesoscale core−periphery structure of shape graphs. a Shows shape graphs for two representative participants (S14 and S01). In the left column, we annotated the nodes using task-type and in the right column the nodes are annotated using the coreness scores (CS). As evident, the more central and densely connected nodes have high CS and more sparsely and peripheral nodes have less CS. b Shows boxplots of CS derived from the CMP data as well as from the three null models (neuRosim, phase randomization with random, and constant phase). Across the four CMP tasks, nodes containing resting time frames had the lowest CS, while the nodes containing time frames from cognitively demanding tasks mainly lied relatively deeper inside the shape graph (i.e. higher CS). The CS of nodes with resting time frames was also lower than all the three corresponding null models (ps < 0.001), and the CS of nodes with working memory or math frames was higher than all the three corresponding null models (ps < 0.005). No significant difference was observed for the CS of nodes with time frames from the video task and the corresponding null models. c Shows the brain regions that were positively associated with coreness scores, i.e., higher the CS, higher the activation in these regions. The group-maps (in red-yellow scale) for positive relation with CS are overlaid on the task-associated meta-analysis maps from NeuroSynth44. As shown, higher CS was associated with higher engagement of task-related brain regions
Fig. 4 Quantifying the mesoscale core−periphery structure of shape graphs. a Shows shape graphs for two representative participants (S14 and S01). In the left column, we annotated the nodes using task-type and in the right column the nodes are annotated using the coreness scores (CS). As evident, the more central and densely connected nodes have high CS and more sparsely and peripheral nodes have less CS. b Shows boxplots of CS derived from the CMP data as well as from the three null models (neuRosim, phase randomization with random, and constant phase). Across the four CMP tasks, nodes containing resting time frames had the lowest CS, while the nodes containing time frames from cognitively demanding tasks mainly lied relatively deeper inside the shape graph (i.e. higher CS). The CS of nodes with resting time frames was also lower than all the three corresponding null models (ps < 0.001), and the CS of nodes with working memory or math frames was higher than all the three corresponding null models (ps < 0.005). No significant difference was observed for the CS of nodes with time frames from the video task and the corresponding null models. c Shows the brain regions that were positively associated with coreness scores, i.e., higher the CS, higher the activation in these regions. The group-maps (in red-yellow scale) for positive relation with CS are overlaid on the task-associated meta-analysis maps from NeuroSynth44. As shown, higher CS was associated with higher engagement of task-related brain regionsCC BY · PubMed Central
Fig. 5 Capturing temporal transitions at the level individual time frames. a Shows a temporal connectivity matrix (TCM) for participant S01 from the CMP dataset. A TCM (size: #time frames × #time frames) shows how each time frame is connected (or similar) to all other time frames. The time frames are connected (i.e. non-zero value in the TCM) if they share a node in the shape graph or if the nodes containing these time frames are connected by an edge in the shape graph. We have manually highlighted the task blocks on the TCM to show the task-related modular structure. b Shows that the degree of TCM nodes can capture the transition between tasks. Here we show average degree across all 18 participants as a solid-line, shaded region shows the standard error around the mean. The dotted lines denote detected change points, which mark the transitions for the eight task blocks. c Shows the effectiveness of capturing temporal transitions extracted from the CMP data as compared to transitions extracted from the phase-randomization null model. The blue dashed line denotes onset of task block and the blue dotted line indicates offset of task block. As compared to the null model data (denoted using diamonds), a time frame by time frame analysis revealed that the degree of TCM generated using real data (denoted using + signs) could capture both the onset and offset of each task within a matter of a few time frames. The red inverted triangles on top denote point-by-point significant difference between transitions from real as compared to null data
Fig. 5 Capturing temporal transitions at the level individual time frames. a Shows a temporal connectivity matrix (TCM) for participant S01 from the CMP dataset. A TCM (size: #time frames × #time frames) shows how each time frame is connected (or similar) to all other time frames. The time frames are connected (i.e. non-zero value in the TCM) if they share a node in the shape graph or if the nodes containing these time frames are connected by an edge in the shape graph. We have manually highlighted the task blocks on the TCM to show the task-related modular structure. b Shows that the degree of TCM nodes can capture the transition between tasks. Here we show average degree across all 18 participants as a solid-line, shaded region shows the standard error around the mean. The dotted lines denote detected change points, which mark the transitions for the eight task blocks. c Shows the effectiveness of capturing temporal transitions extracted from the CMP data as compared to transitions extracted from the phase-randomization null model. The blue dashed line denotes onset of task block and the blue dotted line indicates offset of task block. As compared to the null model data (denoted using diamonds), a time frame by time frame analysis revealed that the degree of TCM generated using real data (denoted using + signs) could capture both the onset and offset of each task within a matter of a few time frames. The red inverted triangles on top denote point-by-point significant difference between transitions from real as compared to null dataCC BY · PubMed Central

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Show all 69 references

Cited by

Sorted by citations

See all 253

Related

Metadata

Authors
Saggar, M., Sporns, O., Gonzalez-Castillo, J., Bandettini, P. A., Carlsson, G., Glover, G., & Reiss, A. L.
Affiliations
  • Stanford Medicine (US)
  • Indiana University Bloomington (US)
  • National Institutes of Health (US)
  • Ayasdi (United States) (US)
  • Stanford University (US)
Corresponding author
Saggar, Manish
ORCID
Venue
Nature Communications
Publisher
Springer Science and Business Media LLC
Published
2018-04-11
Volume
9
Issue
1
Pages
1399-1399
ISSN
2041-1723 (ISSN-L)
DOI
10.1038/s41467-018-03664-4
PMC
PMC5895632
OpenAlex
W2797507693
Cited by
253
References
69
Citations by year
2026: 29 · 2025: 32 · 2024: 19 · 2023: 40 · 2022: 24 · 2021: 40 · 2020: 27 · 2019: 32 · 2018: 10
FWCI
16.23
Type
journal article
Language
en
Open access
Licence
Locations
  • Nature Communications (publishedVersion, cc-by)
  • Europe PMC (PubMed Central) (publishedVersion, other-oa)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • RePEc: Research Papers in Economics (submittedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
  • IUScholarWorks Open (Indiana University) (submittedVersion)
Funders
  • National Alliance for Research on Schizophrenia and Depression
  • National Institutes of Health (93-M-0170, NCT00001360, 5u54mh091657-05, 1U54‐MH‐091657)
  • McDonnell Center for Systems Neuroscience (ZIAMH002783, 1U54MH091657)
  • National Institute of Mental Health (K99MH104605, 1U54 MH091657, U54-MH091657, 93- M-0170, NCT00001360, R00MH104605, ZIAMH002783)
Topics
  • Life Sciences › Neuroscience › Cognitive Neuroscience › Functional Brain Connectivity Studies
  • Physical Sciences › Computer Science › Computational Theory and Mathematics › Topological and Geometric Data Analysis
  • Life Sciences › Neuroscience › Cognitive Neuroscience › Neural dynamics and brain function
Keywords
brain dynamics, topological data analysis, task transitions, interactive visualization, individual differences, task performance prediction
MeSH terms
Adult, Brain, Female, Humans, Magnetic Resonance Imaging, Male, Memory, Short-Term, Nerve Net, Rest, Task Performance and Analysis, Connectome, Datasets as Topic

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review