Limits...
Multi-scale integration and predictability in resting state brain activity.

Kolchinsky A, van den Heuvel MP, Griffa A, Hagmann P, Rocha LM, Sporns O, Goñi J - Front Neuroinform (2014)

Bottom Line: We look at how these measures scale when larger spatial regions as well as larger connectome sub-networks are considered.We also find a set of connectome regions that are both internally integrated and coupled to the rest of the brain, and which resemble previously reported resting-state networks.Finally, we argue that information-theoretic measures are useful for characterizing the functional organization of the brain at multiple scales.

View Article: PubMed Central - PubMed

Affiliation: Department of Informatics, School of Informatics and Computing, Indiana University Bloomington, IN, USA ; Instituto Gulbenkian de Ciência Oeiras, Portugal.

ABSTRACT
The human brain displays heterogeneous organization in both structure and function. Here we develop a method to characterize brain regions and networks in terms of information-theoretic measures. We look at how these measures scale when larger spatial regions as well as larger connectome sub-networks are considered. This framework is applied to human brain fMRI recordings of resting-state activity and DSI-inferred structural connectivity. We find that strong functional coupling across large spatial distances distinguishes functional hubs from unimodal low-level areas, and that this long-range functional coupling correlates with structural long-range efficiency on the connectome. We also find a set of connectome regions that are both internally integrated and coupled to the rest of the brain, and which resemble previously reported resting-state networks. Finally, we argue that information-theoretic measures are useful for characterizing the functional organization of the brain at multiple scales.

No MeSH data available.


Scaling of the information-theoretic measures of seed predictability. Colored lines indicate mean values across all seed ROIs in both hemispheres, while shaded areas indicate values within 1st and 3rd quartile. Colors indicate values for neighbors/subsystems/environments chosen according to Euclidean (red), Connectome (blue), and Randomized (gray) distance metrics. (A) Average Seed-Neighbor MI between seeds and their corresponding kth rank neighbors chosen according to the three distance metrics. (B) Seed-Subsystem MI between seeds and subsystems built according to the three distance metrics. The illustration in the top left corner diagrams how this measure is computed for a given seed and subsystem of size 3. (C) Seed-Environment MI between seeds and environments built according to the three distance metrics. The illustration in the top right corner diagrams how this measure is computed for a given seed and subsystem size 3 (environment size 4). The horizontal dotted line indicates 0.3 bits of Seed-Environment MI, a threshold used later in our definition of Euclidean Coupling Range.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4109611&req=5

Figure 3: Scaling of the information-theoretic measures of seed predictability. Colored lines indicate mean values across all seed ROIs in both hemispheres, while shaded areas indicate values within 1st and 3rd quartile. Colors indicate values for neighbors/subsystems/environments chosen according to Euclidean (red), Connectome (blue), and Randomized (gray) distance metrics. (A) Average Seed-Neighbor MI between seeds and their corresponding kth rank neighbors chosen according to the three distance metrics. (B) Seed-Subsystem MI between seeds and subsystems built according to the three distance metrics. The illustration in the top left corner diagrams how this measure is computed for a given seed and subsystem of size 3. (C) Seed-Environment MI between seeds and environments built according to the three distance metrics. The illustration in the top right corner diagrams how this measure is computed for a given seed and subsystem size 3 (environment size 4). The horizontal dotted line indicates 0.3 bits of Seed-Environment MI, a threshold used later in our definition of Euclidean Coupling Range.

Mentions: Figure 3A shows Seed-Neighbor MI between seeds and their neighbors chosen according to the three distance metrics, averaged over all ROIs in both hemispheres as seeds. ROIs that are closer in Euclidean and Connectome space have a higher MI, with closely ranked Euclidean neighbors (up to neighbor ~8) showing a higher coupling than Connectome neighbors (this reproduces the effect seen in Figure 2B, where proximate Euclidean and Connectome pairs tend to have higher Pairwise MI). As expected, Pairwise MI with Randomized neighbors displays no systematic regularity with neighbor rank. Mean Seed-Neighbor MI for Euclidean neighbors becomes most similar to the mean Seed-Neighbor MI for Randomized neighbors at approximately the 50th neighbor (for Euclidean neighbors, this corresponds to a distance of approximately 40 mm). This is the Euclidean scale at which functional correlations between pairs of physically proximate ROIs decay to baseline levels.


Multi-scale integration and predictability in resting state brain activity.

Kolchinsky A, van den Heuvel MP, Griffa A, Hagmann P, Rocha LM, Sporns O, Goñi J - Front Neuroinform (2014)

Scaling of the information-theoretic measures of seed predictability. Colored lines indicate mean values across all seed ROIs in both hemispheres, while shaded areas indicate values within 1st and 3rd quartile. Colors indicate values for neighbors/subsystems/environments chosen according to Euclidean (red), Connectome (blue), and Randomized (gray) distance metrics. (A) Average Seed-Neighbor MI between seeds and their corresponding kth rank neighbors chosen according to the three distance metrics. (B) Seed-Subsystem MI between seeds and subsystems built according to the three distance metrics. The illustration in the top left corner diagrams how this measure is computed for a given seed and subsystem of size 3. (C) Seed-Environment MI between seeds and environments built according to the three distance metrics. The illustration in the top right corner diagrams how this measure is computed for a given seed and subsystem size 3 (environment size 4). The horizontal dotted line indicates 0.3 bits of Seed-Environment MI, a threshold used later in our definition of Euclidean Coupling Range.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4109611&req=5

Figure 3: Scaling of the information-theoretic measures of seed predictability. Colored lines indicate mean values across all seed ROIs in both hemispheres, while shaded areas indicate values within 1st and 3rd quartile. Colors indicate values for neighbors/subsystems/environments chosen according to Euclidean (red), Connectome (blue), and Randomized (gray) distance metrics. (A) Average Seed-Neighbor MI between seeds and their corresponding kth rank neighbors chosen according to the three distance metrics. (B) Seed-Subsystem MI between seeds and subsystems built according to the three distance metrics. The illustration in the top left corner diagrams how this measure is computed for a given seed and subsystem of size 3. (C) Seed-Environment MI between seeds and environments built according to the three distance metrics. The illustration in the top right corner diagrams how this measure is computed for a given seed and subsystem size 3 (environment size 4). The horizontal dotted line indicates 0.3 bits of Seed-Environment MI, a threshold used later in our definition of Euclidean Coupling Range.
Mentions: Figure 3A shows Seed-Neighbor MI between seeds and their neighbors chosen according to the three distance metrics, averaged over all ROIs in both hemispheres as seeds. ROIs that are closer in Euclidean and Connectome space have a higher MI, with closely ranked Euclidean neighbors (up to neighbor ~8) showing a higher coupling than Connectome neighbors (this reproduces the effect seen in Figure 2B, where proximate Euclidean and Connectome pairs tend to have higher Pairwise MI). As expected, Pairwise MI with Randomized neighbors displays no systematic regularity with neighbor rank. Mean Seed-Neighbor MI for Euclidean neighbors becomes most similar to the mean Seed-Neighbor MI for Randomized neighbors at approximately the 50th neighbor (for Euclidean neighbors, this corresponds to a distance of approximately 40 mm). This is the Euclidean scale at which functional correlations between pairs of physically proximate ROIs decay to baseline levels.

Bottom Line: We look at how these measures scale when larger spatial regions as well as larger connectome sub-networks are considered.We also find a set of connectome regions that are both internally integrated and coupled to the rest of the brain, and which resemble previously reported resting-state networks.Finally, we argue that information-theoretic measures are useful for characterizing the functional organization of the brain at multiple scales.

View Article: PubMed Central - PubMed

Affiliation: Department of Informatics, School of Informatics and Computing, Indiana University Bloomington, IN, USA ; Instituto Gulbenkian de Ciência Oeiras, Portugal.

ABSTRACT
The human brain displays heterogeneous organization in both structure and function. Here we develop a method to characterize brain regions and networks in terms of information-theoretic measures. We look at how these measures scale when larger spatial regions as well as larger connectome sub-networks are considered. This framework is applied to human brain fMRI recordings of resting-state activity and DSI-inferred structural connectivity. We find that strong functional coupling across large spatial distances distinguishes functional hubs from unimodal low-level areas, and that this long-range functional coupling correlates with structural long-range efficiency on the connectome. We also find a set of connectome regions that are both internally integrated and coupled to the rest of the brain, and which resemble previously reported resting-state networks. Finally, we argue that information-theoretic measures are useful for characterizing the functional organization of the brain at multiple scales.

No MeSH data available.