Investigating Fast and Slow Dynamics in Resting-State EPTI via Frequency-Informed Window Sizes
Poster No:
1460
Submission Type:
Abstract Submission
Authors:
Micah Holness1, Anastasia Bohsali1, Brad Baker1, Sir-Lord Wiafe1, Fuyixue Wang2, Vince Calhoun3, Lisa Krishnamurthy4
Institutions:
1TReNDS Center, Georgia State University, Atlanta, GA, 2Massachusetts General Hospital, Harvard University, Boston, MA, 3TReNDS Center, Georgia State/Georgia Tech/Emory University, Atlanta, GA, 4TReNDS Center, Emory University, Atlanta, GA
First Author:
Co-Author(s):
Introduction:
Recent advancements in fMRI acquisition have led to creation of echo-planar time-resolved imaging (EPTI) data (Wang, 2019), which contains fast to slow frequency signatures across grouped echo times (TEs) (Krishnamurthy, 2024). For dynamic functional connectivity (dFNC), window size has been a continuous debate since dynamics depend on window length to determine the resolution of temporal correlations within each sliding window (i.e., longer windows smooth correlations while shorter windows maintain temporal resolution of correlations over time). Furthermore, there is evidence that smaller pockets of coherent brain activity occur with faster frequencies (Chang, 2010). Many studies have approached this issue through adaptive windows, instantaneous correlations, and time-series decomposition into various frequency bands (Iraji, 2020). Our study takes an alternative approach to window-based dFNC tailored to the multi-frequency content of EPTI data, by approximating a matrix of window sizes that are informed by the frequency content of component time-series within each echo group, in order to capture both fast and slow dynamics within the same time-course. Furthermore, our study compares dFNC metrics (such as, mean dwell time and fraction of time spent in a connectivity state) across various window sizes and echo groups to achieve a more robust comparison of changes in connectivity that might otherwise be biased by a single window size, which effectively serves as a frequency filter for dynamics.
Methods:
Single-shot EPTI (3T Siemens,112 echoes, TE=5-60ms, TR=1.7s) was acquired on 3 healthy participants (2 male, 1 female) during rest. Data were motion-corrected, warped to MNI, and smoothed by an 8mm kernel. Using GIFT, we conducted an initial dFNC analysis of windowed time-series correlations with a set window size of 22 TRs, and identified 7 distinct states that appeared across echo groups (8 echo groups; 14 echoes per echo group), exhibiting largely different manifestations of connectivity (Pearson's R < 0.8). We then calculated the max frequency distribution across echo groups by bootstrapping over 20 randomly sampled components with replacement for 100 iterations and calculating the max frequency from fourier transform of the averaged components' time-series. The 95th percentile of each distribution served as upper limit of the window sizes for each echo group and .01 Hz (33 TRs) as lower limit, expanding the range into a set of 10 distinct window sizes for each echo group. We used the following window size formula: window size (TRs) = 1 / (3 x frequency (Hz)); in order to convert the max frequencies into window sizes. A secondary dFNC analysis was computed for all window sizes, achieving a set of high and low frequency dFNC states (N=360) that resulted from a broad range of window sizes.
Results:
We compared appearance and transitions of the 7 distinct states across both echo groups and window sizes. Our results revealed certain states appeared in earlier echo groups (states 1, 2, 3, & 6), such as echo group 1 (5 – 12ms); while other states (4, 5, 6, & 7) were favored by latter echo groups, such as echo groups 3 – 8 (19 – 60ms). In addition, echo group 2 (12 – 19ms) was identified as the transition echo group, showing highest whole-brain temporal variability (mean std =.0078) across both echo groups and windows, as well as containing both states from high-frequency early states and low-frequency latter states (states 2, 4, 5, 6, & 7). State 6 appeared across all echo groups, inferring that the large-scale state may involve both high and low frequency correlations.
Conclusions:
Our results reveal that we can use both echo groups and window sizes to observe changes in states over the course of the scan (windowed timepoints) or over the course of a single TR (echoes) - with the second method potentially allowing increased temporal resolution for extending future examination of signal-related EPTI changes to patient cases, such as lesions due to stroke.
Modeling and Analysis Methods:
Connectivity (eg. functional, effective, structural)
fMRI Connectivity and Network Modeling 1
Task-Independent and Resting-State Analysis 2
Novel Imaging Acquisition Methods:
BOLD fMRI
Imaging Methods Other
Keywords:
Acquisition
ADULTS
Blood
Data analysis
FUNCTIONAL MRI
NORMAL HUMAN
Systems
Other - Echo Planar Time-Resolved Imaging (EPTI)
1|2Indicates the priority used for review
Abstract Information
By submitting your proposal, you grant permission for the Organization for Human Brain Mapping (OHBM) to distribute your work in any format, including video, audio print and electronic text through OHBM OnDemand, social media channels, the OHBM website, or other electronic publications and media.
The Open Science Special Interest Group (OSSIG) is introducing a reproducibility challenge for OHBM 2025. This new initiative aims to enhance the reproducibility of scientific results and foster collaborations between labs. Teams will consist of a “source” party and a “reproducing” party, and will be evaluated on the success of their replication, the openness of the source work, and additional deliverables. Click here for more information. Propose your OHBM abstract(s) as source work for future OHBM meetings by selecting one of the following options:
Please indicate below if your study was a "resting state" or "task-activation” study.
Healthy subjects only or patients (note that patient studies may also involve healthy subjects):
Was this research conducted in the United States?
Are you Internal Review Board (IRB) certified? Please note: Failure to have IRB, if applicable will lead to automatic rejection of abstract.
Were any human subjects research approved by the relevant Institutional Review Board or ethics panel? NOTE: Any human subjects studies without IRB approval will be automatically rejected.
Were any animal research approved by the relevant IACUC or other animal research panel? NOTE: Any animal studies without IACUC approval will be automatically rejected.
Please indicate which methods were used in your research:
For human MRI, what field strength scanner do you use?
Which processing packages did you use for your study?
Provide references using APA citation style.
2. Iraji, A. (2020). Tools of the trade: estimating time-varying connectivity patterns from fMRI data. Social Cognitive and Affective Neuroscience, 16(8), 849-874.
3. Krishnamurthy, L. (2024). Data-driven analysis of echo planar time-resolved (EPTI) MRI suggests frequency-specific mechanisms of brain fluctuations with unique TE signatures. [Unpublished manuscript]
4. Wang, F. (2019). Echo planar time-resolved imaging (EPTI). Magnetic Resonance in Medicine, 81(6), 3417-3923.