Physicians, neuroscientist and researchers who are interested in resting state fMRI under influence of hypercapniafMRI using blood oxygenation level-dependent (BOLD) contrast has appeared to measure vascular oxygenation change due to neuron activity¹. Resting-state functional MRI (rs-fMRI) investigating low-frequency fluctuations of spontaneous neuronal activity in the brain can be measured based on BOLD signal by gradient echo EPI. Among the papers of resting-state networks, the default-mode network (DMN) is highly reproducible². Furthermore, some clinical studies have indicated the disrupted connectivity of DMN could be found in neuronal disorders, such as Alzheimer’s disease³ and attention deficit hyperactivity disorder⁴. However, alteration of BOLD signals under hypercapnia has been noticed and it might cause misinterpretation of fMRI results with concurrent hypercapnia⁵. Carbon dioxide (CO2) is a potent vasodilator that could increase the cerebral blood flow prominently⁶. Because BOLD signal coming from the cerebral blood flow changes, the neuron-activity related BOLD signals will be interfered by the BOLD signals coming from the inhaled CO2 (hBOLD)⁷. We hypothesize that the influence in hypercapnia might be different for the connectivity of DMN and hypercapnia grades. Therefore, in our study, we explored the change of functional connectivity of DMN in different CO2 concentration.

Experiment design: 10 healthy males (mean age 23, range 21-25) given informed consent were scanned with inhalation of different CO2 concentrations. All images were performed on a 3.0 Tesla MR scanner (GE Signa HDx, GE Healthcare, Milwaukee,Wis). A high-resolution T1W image for coregistration of all participants were scanned by a 3D IR SPGR (TR/TI/TE 7.84/450/2.98, FOV: 256 mm, 256x256, 1 mm thickness). BOLD signals were measured using the EPI sequence , and 34 slices (FOV: 192 mm, 64x64, 3 mm thickness) were acquired to cover the whole brain. Anatomical T1 weighted images were acquired to rule out morphological deficits. Each experiment consisted of 80 scans with a sampling time of 2.5 seconds. The first 5 dummy scans were discarded. The CO2 gas mixture (air, 3%, 5% and 7%) was given at the hypercanpic phase during 6 min for each experiment. BOLD-EPI scan was started after 3 min from gas delivery for steady-state hypercapnia.

Data Analysis: Image data were preprocessed using SPM8 package. The GE-EPI images were firstly realigned, then co-registered and normalized to the MNI space. The template provided by SPM was further used for normalization with re-sampling to 2 mm cubic voxels and a Gaussian kernel of 6 mm for spatial smoothing. The Functional Connectivity (FC) was calculated by Resting-State fMRI Data Analysis Toolkit (REST). The band-pass filtering (0.03 to 0.08 Hz) was used to reduce low-frequency drift and high-frequency physiological respiratory and cardiac noise. For the group analysis, the voxel-by-voxel statistical significance of FC value was analyzed using the one-sample t test (p<0.01). The activated voxels inside the region DMN were counted to calculate the activated volumes for different CO2 concentrations (air 3%, 5% and 7%).

Figure 1 depicts the functional connectivity matrix of cerebrum regions for different CO2 concentration. The matrices provide the correlations of Pearson coefficients at a frequency band of 0.03–0.08 Hz of 90 (sub)cortical regions as defined in the Automated Anatomical Labeling (AAL) atlas. histogram of BOLD percentage change at all 90 AAL ROIs for 3%, 5% and 7% CO2 concentration. Figure 2 shows the DMN connectivity maps obtained through group analysis. The activated voxels were defined by the T value over 3. The activated volumes ratios were calculated by dividing the activated volume in air experiment for 3%, %% and 7% CO2 concentration (Figure 3). Our results show that the functional connectivity of DMN is changed in hypercapnia. FC is slightly changed in 3% CO2 fractions and is gradual reduced as the CO2 fraction increases. The reasons might be related to the alteration of cerebral hemodynamics, which was suggested as a nonlinear function of CO2 concentration⁸. Our results also demonstrat the FC reduction is different for different DMN region. It could result that the BOLD variations by hypercapnia are different with local brain area and hypercapnia grades. In this study, we provide simple but important concepts in approaching the influence of inhaled CO2 on the resting-state fMRI experiments.