Introduction

The development of novel functional MRI (fMRI) sequences and high-field MRI scanners has made it possible to investigate the function of the human cerebral cortex along its depth in task-related and resting-state studies (e.g.^1-5). In a previous abstract, we demonstrated that global cortical connectivity along different depth territories (not specific to any ROI) is highly variable and suggested that the prevalence of the identified connectivity states (supremacy of superficial vs. deep layer connectivity) could be indicative of the brain’s mode of working⁶; however, the biological meaning remained ambiguous. Here, we evaluate the influence of a psychological alteration (depression) on these global laminar connectivity states.

Methods

Thirteen healthy subjects and nineteen patients diagnosed with major depressive disorder (Uniklink Aachen) were imaged at a 7T scanner. Patients underwent MRI measurements in two sessions (before and after exposure to treatment, 4-6 weeks apart) and were further assessed using a psychometric test (Beck’s Depression Inventory). The MRI protocol included an MP2RAGE scan, where TR/TE = 4500/2 ms, matrix = 208×300×320 and voxel size = 0.75×0.75×0.75 mm³, and an fMRI scan using GE-EPIK with TR-external phase-correction^1,7-13 (TR/TE = 3500/22 ms, FA = 85°, partial Fourier = 5/8, 3-fold in-plane/3-fold inter-plane (multi-band) acceleration, bandwidth = 875 Hz/Px, αPC/αMain = 9°/90°, Matrix = 336×336×123 and voxel size = 0.63×0.63×0.63 mm³). The MP2RAGE was used to extract the cortical grey matter and generate six surfaces at equal distances from each other along the cortical depth using Freesurfer. Magnitude and phase functional images were reconstructed and pre-processed using SPM, FSL and AFN, including slice timing correction, realignment, temporal filtering and regression of motion parameters + CSF and WM signal + physiological signals (respiratory and cardiac). To correct for the vein-related BOLD signal bias, the pre-processed phase signal was additionally regressed out from the pre-processed magnitude signal¹⁴. Areas where the mean fMRI and the MP2RAGE image were not well co-registered were masked out of the analysis. The clean functional image was mapped to the six surfaces generated from the MP2RAGE, and the signal of vertices in 23 cortical ROIs was averaged for each cortical depth. This resulted in 46×6 time courses. Functional connectivity was calculated as the temporal correlation between pairs of time courses, which returned a 276×276 (ROI-depth-to-ROI-depth) connectivity matrix. A dynamic connectivity analysis was conducted to assess connectivity changes over time by segmenting the fMRI time courses into 120s sliding windows at a step size of 3.5s (1 TR) and computing the temporal correlation for each of these blocks (104). A global depth-to-depth connectivity matrix was computed by averaging the ROI-depth-to-ROI-depth matrix across ROIs. These 6×6 matrices were further simplified by averaging across one dimension to obtain a connectivity profile. Each evaluated time block during the scan time (corresponding to one connectivity matrix in the dynamic analysis) was assigned to one of two states depending on whether the connectivity profile had a positive (stronger superficial) or a negative (stronger deep) slope, denominated as state 1 and 2, respectively (Figure 1). A prevalence analysis was conducted for healthy volunteers and for patients before and after treatment. To assess the potential relevance of these fMRI states, the correlation between the evolution of the prevalence of fMRI-state 1 and the evolution of BDI scores was calculated. For group comparison, patients were further segregated into a good and a bad prognosis group, depending on whether their psychometric scores improved (the usual case) or worsened upon treatment. Statistically significant changes were assessed with paired t-tests (between patients) and unpaired t-tests (between healthy controls and patients).

Results

Preliminary results from patient data showed a moderate correlation between the evolution of laminar state 1 (superficial connectivity) and of the psychometric scores, where decreased prevalence of the superficial connectivity state correlated with improved BDI scores. The prevalence of state 1 was increased in patients at their first session, compared to controls, but it decreased in the second session for patients with improved psychometric scores; in contrast, it remained increased (at higher levels) in three patients who worsened.

Conclusions

Our results indicate that the global connectivity of superficial vs. deep territories of the cortex could be sensitive to psychiatric disease. We are aware that averaging across the whole cortex (per depth) cancels out many distinct responses at particular locations; however, it also helps in studying the depth-relevant component of brain connectivity, which appears to be a promising research target. Further work is needed to verify the potential relevance of this global depth connectivity.

Acknowledgements

We thank Ms Petra Engels, Ms Elke Bechholz and Ms Anita Köth for technical support during MRI acquisition; Ms Rick Claire for abstract corrections, and all the subjects for their excellent cooperation.

References

1. Pais-Roldan, P., et al., Cortical depth-dependent human fMRI of resting-state networks using EPIK. Front Neurosci, 2023. 17: p. 1151544.

2. Huber, L., et al., Layer-dependent functional connectivity methods. Prog Neurobiol, 2021. 207: p. 101835.

3. Sharoh, D., et al., Laminar specific fMRI reveals directed interactions in distributed networks during language processing. Proc Natl Acad Sci U S A, 2019. 116(42): p. 21185-21190.

4. Polimeni, J.R., et al., Laminar analysis of 7T BOLD using an imposed spatial activation pattern in human V1. Neuroimage, 2010. 52(4): p. 1334-46.

5. Huber, L., et al., High-Resolution CBV-fMRI Allows Mapping of Laminar Activity and Connectivity of Cortical Input and Output in Human M1. Neuron, 2017. 96(6): p. 1253-1263 e7.

6. Pais-Roldán, P., et al., High dynamicity of cortical depth-dependent connectivity states. ISMRM, London, UK, 2022. 4764.

7. Zaitsev, M., K. Zilles, and N.J. Shah, Shared k-space echo planar imaging with keyhole. Magn Reson Med, 2001. 45(1): p. 109-17.

8. Shah, N.J., Zilles, K., Verfahren zur Untersuchung eines Objektes mittels Erfassung des Ortsfrequenzraumes. 2003.

9. Zaitsev, M., et al., Dual-contrast echo planar imaging with keyhole: application to dynamic contrast-enhanced perfusion studies. Phys Med Biol, 2005. 50(19): p. 4491-505.

10. Yun, S.D., et al., Parallel imaging acceleration of EPIK for reduced image distortions in fMRI. Neuroimage, 2013. 73: p. 135-43.

11. Yun, S.D. and N.J. Shah, Whole-brain high in-plane resolution fMRI using accelerated EPIK for enhanced characterisation of functional areas at 3T. PLoS One, 2017. 12(9): p. e0184759.

12. Yun, S.D., et al., Evaluating the Utility of EPIK in a Finger Tapping fMRI Experiment using BOLD Detection and Effective Connectivity. Sci Rep, 2019. 9(1): p. 10978.

13. Yun, S.D., et al., Mapping of whole-cerebrum resting-state networks using ultra-high resolution acquisition protocols. Hum Brain Mapp, 2022. 43(11): p. 3386-3403.

14. Menon, R.S., Postacquisition suppression of large-vessel BOLD signals in high-resolution fMRI. Magn Reson Med, 2002. 47(1): p. 1-9.