Cortical laminar resting‐state signal fluctuations scale with the hypercapnic blood oxygenation level‐dependent response

Abstract

Calibrated functional magnetic resonance imaging can remove unwanted sources of signal variability in the blood oxygenation level-dependent (BOLD) response. This is achieved by scaling, using information from a perfusion-sensitive scan during a purely vascular challenge, typically induced by a gas manipulation or a breath-hold task. In this work, we seek for a validation of the use of the resting-state fluctuation amplitude (RSFA) as a scaling factor to remove vascular contributions from the BOLD response. Given the peculiarity of depth-dependent vascularization in gray matter, BOLD and vascular space occupancy (VASO) data were acquired at submillimeter resolution and averaged across cortical laminae. RSFA from the primary motor cortex was, thus, compared to the amplitude of hypercapnia-induced signal changes (tSD hc ) and with the M factor of the Davis model on a laminar level. High linear correlations were observed for RSFA and tSD hc (R 2 = 0.92 ± 0.06) and somewhat reduced for RSFA and M (R 2 = 0.62 ± 0.19). Laminar profiles of RSFAnormalized BOLD signal changes yielded good agreement with corresponding VASO profiles. Overall, this suggests that RSFA contains strong vascular components and is also modulated by baseline quantities contained in the M factor. We Occupancy; WM, white matter; Mathematical symbols: CBF, cerebral blood flow; CBV, cerebral blood volume; CMR O2 , cerebral metabolic rate of oxygen consumption; CVR, cerebrovascular reactivity; f, normalized CBF; M, calibration constant corresponding to the maximal BOLD signal change; P ET CO 2 , end-tidal partial pressure of carbon dioxide; P ET O 2 , end-tidal partial pressure of oxygen; R 2 , coefficient of determination; R 0 2 , reversible transverse relaxation rate; RSFA, fluctuation amplitude of the resting-state timeseries; r, normalized CMRO2; S, signal amplitude; ΔS, relative BOLD signal change; ΔS hc , relative BOLD signal change induced by mild hypercapnia; T 1 , longitudinal relaxation time; TE, echo time; TI, inversion time; TR, repetition time; tSD hc , temporal standard deviation of the hypercapnia timeseries; tSD rs , temporal standard deviation of the resting-state timeseries; tSNR, temporal signal-to-noise ratio; p, error probability; v, normalized CBV; Z, Z-score; α, Grubb exponent; β, constant describing the coupling between R 0 2 and [dHb]; κ, proportionality constant; 0 , index indicating a baseline ("resting") level; full , index indicating the entire frequency band; high, index indicating the high-frequency band; low, index indicating the low-frequency band; t, index indicating the total blood compartment; v, index indicating the venous compartment; [X], concentration of compound X.Maria Guidi and Laurentius Huber contributed equally to this study.

Validating layer-specific VASO across species

“…The somatosensory CBV changes in response to whisker stimulation peak in the range of 5%-10% (compared to the baseline CBV). Such amplitudes are consistent with previously reported ΔCBV estimates in the pre-clinical domain (Hillman et al 2007;Kennerley et al 2005;2012a;Lee et al 2001;Tian et al 2010), whereas they are about 5-10 times smaller than previously reported VASO signal changes in humans (Beckett et al 2020;Chai et al 2019;Finn et al 2019;Guidi et al 2017;Huber et al 2018;Kurban et al 2020;Persichetti et al 2020;Yang and Yu 2019;Yu et al 2019).…”

“…The VASO CBV profile, however, also exhibits a similarly elevated CBV response in the superficial layers. This is consistent with previous layer-fMRI VASO studies (Beckett et al 2020;Chai et al 2019;Donahue et al 2017;Finn et al 2019;Guidi et al 2017;Huber et al 2014a;2017;Kurban et al 2020;Persichetti et al 2020;Yang and Yu 2019;Yu et al 2019) and might appear in conflict with the large body of preclinical studies. Note that the superficial activity in VASO profiles is located within the cortex and is approx 0.8 mm deeper than the GE-BOLD peak response above the cortical surface.…”

“…Slice selective slab-inversion (SS-SI) VASO is widely used across the literature for non-invasive imaging of microvascular layer-specific signal responses (Beckett et al 2020;Chai et al 2019;Finn et al 2019;Guidi et al 2017;Huber et al 2018;Kurban et al 2020;Persichetti et al 2020;Yang and Yu 2019;Yu et al 2019).…”

Laminar fMRI at ultra-high fields

“…To examine if the laminar neuronal specificity of the signal can be recovered from the data, we used a normalisation approach. Normalisation of laminar profiles have been proposed previously in the context of resting-state fMRI [157,158] using BOLD fluctuation power to probe connectivity across areas and with induced hypercapnia [158]. These normalisation approaches are motivated by the fact that the baseline parameters, such as venous CBV (V0), are assumed to be multiplicative for BOLD sensitivity and conforms to the single-vascular-compartment BOLD signal models (as in Davis et al [159], Buxton et al [160]).…”

“…Thus, calibrating the BOLD signal to another condition or time point (or baseline fluctuations [157,158]) effectively removes the bias stemming from baseline physiological and physical parameters. For the case that neuronal profiles of two stimuli or time-points are just scaled versions of each other (i.e.…”