Bridging subject variability patterns between fMRI and MEG studies
Poster No:
1316
Submission Type:
Abstract Submission
Authors:
Brian Zhaoyi Mo1, Stephen Smith1, Mark Woolrich1
Institutions:
1University of Oxford, Oxford, United Kingdom
First Author:
Co-Author(s):
Introduction:
Recent methods for modelling brain activity are providing new insight into functional brain connectivity and networks. These methods can capture meaningful variation across the population in both fMRI and MEG studies, showing potential for prediction of cognition or clinical trait(Finn, 2015; da Silva Castanheira, 2021; Sareen, 2021). Due to the inherent differences in the measurement of brain activity, linking functional networks between MEG and fMRI at the individual participant level remains a challenge(Hall, 2014). Nevertheless, by leveraging patterns of subject variability, the connection between fMRI and MEG can be better understood. Using resting fMRI and MEG data from the same participants, we compared fMRI and MEG functional connectivity(FC) at both the group and individual level.
Methods:
We used the Cam-CAN dataset(N = 612 subjects; age range: 18-88 yrs)(Taylor, 2017; Shafto, 2014), in which both resting-state fMRI and MEG are available in the same subjects among the healthy cohort. To facilitate cross-modality comparative analysis, we used a modified Glasser parcellation(Glasser52) for both modalities. For fMRI, we computed parcel time series by regressing the fMRI data on parcels' spatial maps. We then estimated fMRI FC using full correlation. For MEG, we computed parcel time series using source reconstruction(LCMV beamforming), before extracting the first PC in each parcel, followed by symmetric orthogonalization to reduce spatial leakage. We then estimated MEG FC using 2 different methods: 1) Time-averaged(static) FC, obtained using Amplitude Envelope Correlation(AEC) across broad-band frequencies(1-45 Hz) and specific frequency bands: delta(1-4 Hz), theta(4-8 Hz), alpha(8-13 Hz), beta(13-24 Hz), and gamma(30-45 Hz). 2) Dynamic FC, obtained using the Time-Delay Embedded Hidden Markov Model(TDE-HMM)(Baker, 2014; Vidaurre, 2018) with dual estimation. For the group-level, cross-modal analysis, we calculated Pearson's r between the modality-specific, group-level FCs, obtained by averaging over the individual FC. For individual-level analysis, we extracted subject "fingerprints" by removing group-level information. We then conducted three downstream tasks to characterise the subject variability patterns:(1)Modality-specific within-subject discriminability, by correlating cross-session fingerprints for each modality (including the use denoising via Principal Component Analysis(PCA) to enhance discriminability).(2)Modality-specific age prediction, using subject fingerprints with a 10-fold cross-validated ElasticNet model.(3)Cross-modality fingerprint alignment, using a regression model to assess subject prediction accuracy.
Results:
The modality-specific analysis results at the individual level are shown in Fig1B for subject discriminability and in Fig1C for predicting age. Note that PCA denoising further improved the discriminability, though it also impacted the between-subject geometric. The cross-modal analysis results at the group level are shown in Fig1A, indicating reasonably consistent connectivity patterns at the group level. Finally, the cross-modality analysis results at the individual level are shown in Fig1D, where we attempted to align the cross-modality subject variability by predicting MEG fingerprints using fMRI fingerprints.
Conclusions:
In modality-specific analysis, both fMRI and dynamic MEG features, as well as most MEG static features, showed strong performance in age prediction and subject differentiation. In group-level cross-modal analysis, and as shown in prior studies(Brookes, 2011; Wirsich, 2021), we found reasonable network similarity between resting-state MEG and fMRI. At the individual level, while shared variability between MEG and fMRI fingerprints was observed, cross-modality subject identification using linear methods has shown limited success. These results suggest that while both modalities show meaningful subject variability, the pattern of variability across subjects is largely complementary and modality-specific.
Modeling and Analysis Methods:
Bayesian Modeling
Connectivity (eg. functional, effective, structural)
EEG/MEG Modeling and Analysis 1
fMRI Connectivity and Network Modeling 2
Keywords:
Aging
Data analysis
FUNCTIONAL MRI
MEG
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?
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.
Brookes, M. J. (2011). Measuring functional connectivity using MEG: methodology and comparison with fcMRI. Neuroimage, 56(3), 1082-1104.
da Silva Castanheira, J. (2021). Brief segments of neurophysiological activity enable individual differentiation. Nature communications, 12(1), 5713.
Finn, E. S. (2015). Functional connectome fingerprinting: identifying individuals using patterns of brain connectivity. Nature neuroscience, 18(11), 1664-1671.
Hall, E. L. (2014). The relationship between MEG and fMRI. Neuroimage, 102, 80-91.
Sareen, E. (2021). Exploring MEG brain fingerprints: Evaluation, pitfalls, and interpretations. NeuroImage, 240, 118331.
Shafto, M. A. (2014). The Cambridge Centre for Ageing and Neuroscience (Cam-CAN) study protocol: a cross-sectional, lifespan, multidisciplinary examination of healthy cognitive ageing. BMC neurology, 14, 1-25.
Taylor, J. R. (2017). The Cambridge Centre for Ageing and Neuroscience (Cam-CAN) data repository: Structural and functional MRI, MEG, and cognitive data from a cross-sectional adult lifespan sample. neuroimage, 144, 262-269.
Vidaurre, D. (2018). Spontaneous cortical activity transiently organises into frequency specific phase-coupling networks. Nature communications, 9(1), 2987.
Wirsich, J. (2021). The relationship between EEG and fMRI connectomes is reproducible across simultaneous EEG-fMRI studies from 1.5 T to 7T. NeuroImage, 231, 117864.