Resting State functional Magnetic Resonance Imaging (rs-fMRI) has revealed brain-wide correlation patterns throughout the human brain, interpreted as Functional Connectivity. Dynamic Functional Connectivity (DFC) has recently expanded on this technique via sliding window correlation analysis, revealing moment-to-moment changes in functional connectivity across an imaging session. However, the meaning of these transitions in terms of neural activity and behavior are not well understood.In this work, I utilized Dynamic Functional Connectivity analytical techniques in conjunction with Wide Field Optical Mapping (WFOM) in the awake, freely behaving mouse. I hypothesized that neural and hemodynamic activity observed with WFOM would exhibit similar transitions between functional connectivity states as reported by fMRI DFC studies. I also explored whether changes in functional connectivity would correspond to changes in behavior.
Simultaneous neural and hemodynamic activity was collected using WFOM from five freely behaving head-fixed Thy1-jRGECO1a mice. Behavioral metrics of movement, whisking and pupillometry were acquired simultaneously. Raw neuroimaging data were dimensionally reduced to representative time courses across the dorsal surface of the cortex for each subject utilizing a semi-supervised clustering technique. Functional Connectivity analysis revealed rich spatiotemporal structures within neural and hemodynamic activity, which were consistent across imaging sessions and subjects.
I observed broad changes in Functional Connectivity metrics during rest, locomotion, and transitional epochs between the two by directly comparing windows captured during these epochs. It was also observed that Functional Connectivity metrics immediately following locomotion offset could be distinguished from periods of sustained rest. Similar to human fMRI studies, a distinct increase in bilateral connectivity of anterior lateral prefrontal cortex was observed, which became significantly less synchronized with posterior brain regions during sustained periods of rest.
I next used an unsupervised clustering technique on the same data to test if these properties could be observed in an indirect manner. This approach has been previously used in numerous human fMRI studies, and contextualized this work to human fMRI studies. A sliding window was used to calculate moment-to-moment Functional Connectivity maps across each imaging session. These dynamic correlation maps were clustered into multiple states, which could then be used to calculate the most representative state for any given epoch. Unsupervised clustering revealed strikingly similar dynamic states to our previous observations. These dynamic states also exhibited independent distributions of behavioral activity both in neural and hemodynamic models, leading us to conclude that there is not only a meaningful link between Functional Connectivity in neural and hemodynamic activity, but that behavioral shifts largely drive these changes.
My findings provide strong evidence that Dynamic Functional Connectivity has neural origins, and hemodynamic responses are able to depict correlation patterns that tracks rapid changes in behavior and internal brain states such as the level of arousal or alertness. Future studies are necessary to further investigate this speculation, but this offers an excellent framework to better understand the rich, dynamic properties of brain activity.
| Identifer | oai:union.ndltd.org:columbia.edu/oai:academiccommons.columbia.edu:10.7916/n51c-c778 |
| Date | January 2023 |
| Creators | Thibodeaux, David Nicholas |
| Source Sets | Columbia University |
| Language | English |
| Detected Language | English |
| Type | Theses |
Page generated in 0.0017 seconds