Quantitative fMRI is in the focus of current neuroimaging. However, several reports suggest that the use of a breath-hold or gas-manipulation task (both causing discomfort) may be unnecessary.¹ Such challenges are believed to elicit a purely vascular response. Alternative ways of measuring the cerebrovascular reactivity were suggested from work focusing on resting-state fluctuations.^2,3 These signal fluctuations, which are believed to result from global changes in venous oxygenation have been used as a measure of vascular reactivity and even for inter-subject BOLD signal calibration.⁴ Previous investigations have mainly tackled the problem on a voxel, ROI, and inter-subject basis. Recently, it was shown that physiologically driven signal fluctuations in BOLD time series exhibit a specific depth-dependent signature,⁵ which appears to be similar to hypercapnia responses.⁶

The aim of this work is to compare the resting-state fluctuation amplitude (RSFA) with the hypercapnic BOLD response on a laminar basis. To assess potential employment in normalizing the BOLD response, fluctuation amplitudes were additionally compared to the laminar M parameter calculated from the Davis model.

Subjects and sequence: N=5 participants were examined on a Siemens 7T scanner with a combined BOLD and VASO sequence⁷ at a nominal in-plane resolution of 0.8×0.8 mm² and slice thickness of 1.2-1.5mm. Imaging focused on the hand-knob of one hemisphere with 5-7 slices perpendicular to the cortical surface. The gas-manipulation session consisted of administration of mild hypercapnia (5%CO2) for two 3min periods (total scan time 15 min). The resting-state session consisted of a 12-min scan with acquisition of either BOLD contrast (TR=1.5s, 3 participants) or BOLD interleaved to VASO (TR=3s, 2 participants).

Extraction of profiles: The resting-state time series were bandpass filtered to obtain low- (0.01-0.1 Hz) and high-frequency series (0.1-0.25 Hz). Layering was done using an equidistant approach directly in EPI space. In total, 20 laminae were constructed in each subject, of which the inner-most 10 to 12 were selected for the correlation analysis. Temporal standard deviation maps were obtained and lamina-dependent values extracted. The laminar M parameter was computed with the Davis model⁸ as previously described.⁶

Fig. 1 depicts results from the M1/S1 region of one subject. Significant hypercapnia responses can be detected with BOLD and VASO at the ultra-high resolutions used here. This permits layer-dependent analysis in EPI space (colorful edges overlayed on EPI images). Fig. 2 shows that cortical profiles of the calibration parameter M obtained with hypercapnia and resting-state signal fluctuations are very similar. Both are highest at the cortical surface and decrease with cortical depth.

Scatter plots between laminar BOLD RSFA and laminar temporal standard deviation (tSD) of the hypercapnia-BOLD time courses were constructed and the Pearson correlation coefficient calculated (Fig. 2). Results in Fig. 3 show that the correlations are generally quite high, however, with some degree of variation across participants.

The previously observed positive correlation between hypercapnic response and RSFA is confirmed at a laminar level. The specific signature of decreasing RSFA and M with cortical depth is consistent with the underlying variation of venous blood volume.⁵ The source of the correlation might primarily result from large BOLD fluctuations at the pial surface, which would explain a lower correlation with the parameter M.The results shown here suggest that a layer-dependent fMRI signal calibration factor might be obtained from resting-state fMRI time series without the need to conduct uncomfortable hypercapnia-inducing breathing manipulations. Both, RSFA and calibration parameter M capture the cortical profile of vascular reactivity variations with cortical depth. The presented validation experiments suggest that fMRI signal calibration with RSFA may gain importance in future experiments to reveal layer-dependent brain activity information from high-resolution fMRI.