Purpose

Identifying and understanding brain connectivity from non-invasive and in-vivo functional magnetic resonance imaging (fMRI) is desirable both for physiological and pathological processes. With recent advances in simultaneous multi-slice imaging⁽¹⁾, which have improved the temporal resolution of fMRI, it is possible to capitalize on the information contained within the temporal features of BOLD functional connectivity (FC) using a sliding window approach⁽²⁾. In the current study, we first investigated the impact of repetition time (TR) on BOLD fMRI derived dynamic FC (dFC) in a sliding window approach in an event-related, visual-stimulation paradigm. In addition, we tested the feasibility of the optimized dFC method in three epilepsy patients with different epilepsy syndromes and tested the plausibility of the results in relation to the clinical information.

Methods

To characterize the impact of technical parameters (window size, TR); we considered the data from a previous study^(3). Fifteen healthy volunteers participated in the study (“control task”). First, a short block paradigm (duration: 3 min) with blocks of flickering checkerboards and blocks with a fixation cross (duration of each block: 20 s) was used to derive a functionally-defined region of interest (ROI) within the visual cortex. Then, an event-related design (duration of paradigm: 10 min) with brief, pseudo-random checkerboard stimulation periods (duration: 500 ms) was applied. In addition, three patients were recruited in the Neurology/Epileptology department of the University Hospital Tübingen (Germany). Written informed consent was obtained prior to the measurement, in accordance with the ethics committee guidelines. A resting state (eyes closed) paradigm was employed for 30 min (EEG-fMRI). As subject motion can severely influence the FCD estimation^(4), we included subjects and patients with minimal head motion (< 1.5 mm) in our study. The sliding window analysis was performed using the DynamicBC toolbox^(5) with varying window sizes (1 data point shifts). The toolbox is based on Pearson linear correlation to compute bivariate FC between every pair of voxels. Subject specific whole brain gray matter masks (SPM12, unified segmentation) at a probability threshold of 0.2 were used as the ROI for the connectivity analysis. We estimated the functional connectivity degree (FCD) that counts the total number of connections of a given voxel above a predefined threshold (Pearson linear correlation, p<0.001). The FCD values were extracted from the functional ROI (10 s before until 20 s after stimulus onset) and baseline corrected by subtracting the mean signal 5 s prior to the stimulus onset. Similarly, the average FCD time course where extracted from regions defined by the 7-network parcellated atlas⁽⁶⁾. In patients, the FCD time course was extracted for a time window of 20 s before and 30 s after the interictal epileptic discharge (IED) onset, from regions defined by the Freesurfer Desikan-Killiany anatomical atlas.

Results

The visual block-design yielded widespread activation in the visual cortex (Figure 1).We first compared the event-related hemodynamic response (Figure 2, first column) with the FCD transients (Figure. 2, columns 2-5 for window sizes of 7.8 s, 13.2 s, and 18.4 s respectively) for the TRs of 2.64 s, 1.32 s, 0.66 s, and 0.33 s within the defined ROI obtained from the control task. Figure 3 shows the spatial and temporal distribution of the FCD during the event-related paradigm across the regions defined by the^(6) functional network atlas. In addition, it also shows the percent signal change (hemodynamic response) across the functional regions. The feasibility of this approach in patients is shown in Figure 4, where we can observe a clear anatomical relation of the FCD distribution and the clinical hypothesis in the three patients (one focal lesional, two with generalized epilepsy).

Discussion

One of the most striking finding was that it was possible to capture dynamic network changes in the dynamic FCD (dFCD) approach for brief events of 500 ms even with a relatively short window size of 7.8 s. In our paradigm of brief events, a window size of 7.8-13.2 s (sampled beyond the conventional TR of 2.64 s) would be an optimum choice. In addition, we could show that dFCD can also be used to capture dynamic network changes during epileptic spike discharges. A larger cohort of patients is required to assess the stability and clinical utility of such approaches.

Conclusion

Dynamic FCD transients are better detectable with sub-second TR than conventional TR. This approach was capable of capturing neuronal connectivity across various regions of the brain, indicating a potential to study the temporal characteristics of interictal epileptiform discharges and seizures in epilepsy patients or other brain diseases with brief events.

Acknowledgements

This study was supported by a grant of the Werner Reichardt Centre for Integrative Neuroscience (CIN grant: Pool-Project 2012-10) and the DFG (CIN EXC 307). We thank the University of Minnesota Center for Magnetic Resonance Research for providing the multiband-EPI sequence (http://www.cmrr.umn.edu/multiband).

References

1. Larkman DJ, Hajnal JV, Herlihy AH, Coutts GA, Young IR, Ehnholm G. Use of multicoil arrays for separation of signal from multiple slices simultaneously excited. J Magn Reson Imaging 2001;13(2):313-317.

2. Di X, Fu Z, Chan SC, Hung YS, Biswal BB, Zhang Z. Task-related functional connectivity dynamics in a block-designed visual experiment. Frontiers in human neuroscience 2015;9:543.

3. Sahib AK, Mathiak K, Erb M, et al. Effect of temporal resolution and serial autocorrelations in event-related functional MRI. Magnetic resonance in medicine : official journal of the Society of Magnetic Resonance in Medicine / Society of Magnetic Resonance in Medicine 2016.

4. Yan CG, Cheung B, Kelly C, et al. A comprehensive assessment of regional variation in the impact of head micromovements on functional connectomics. NeuroImage 2013;76:183-201.

5. Liao W, Wu GR, Xu Q, et al. DynamicBC: a MATLAB toolbox for dynamic brain connectome analysis. Brain connectivity 2014;4(10):780-790.

6. Yeo BT, Krienen FM, Sepulcre J, et al. The organization of the human cerebral cortex estimated by intrinsic functional connectivity. Journal of neurophysiology 2011;106(3):1125-1165.