Materials and Methods
In total, 26 pilots on active duty and 24 healthy controls participated in this study. Fourteen pilots were flight instructors from the Civil Aviation Flight University of China. Their aircraft types were C172R, PA44, SR-20, PA44, MA600, and CE525. Twelve pilots were first officers from different airlines. Their aircraft types were Boeing 737 or Airbus 320. The two groups were matched for sex, handedness, and education characteristics. Individuals with a history of neurological illness, traumatic brain injury, substance-related disorders, or standard contraindications to MRI were excluded from this study.
The experimental procedures (No. 2018-042002) were approved by the Ethics Committee of University of Electronic Science and Technology of China (Chengdu, China), and written informed consent was obtained from all participants.
Data Acquisition
The MRI data were acquired using a 3-Tesla MRI scanner (DISCOVERY MR 750, GE Healthcare, Waukesha WI, United States) at the Center for Information in Medicine of University of Electronic Science and Technology of China. Subjects were scanned in a supine, head-first position. Cushions were placed on both sides of the head to decrease head motion.
High-spatial-resolution structural images were acquired using a T1-spoiled gradient recalled echo pulse sequence. The scan parameters were as follows: repetition time (TR), 5.976 ms; echo time (TE), 1.976 ms; flip angle, 9°; matrix, 256 × 256; slice thickness, 1 mm (no gap); field of view (FOV), 25.6 cm × 25.6 cm; and slice number, 154.
A gradient-echo echo-planar imaging sequence was used to collect resting-state fMRI data. The participants were instructed to lying quietly and close their eyes without falling asleep during scanning. A period of 510 s resting-state functional images were also collected for each subject. The scan parameters were as follows: TR, 2000 ms; TE, 30 ms; flip angle, 90°; matrix, 64 × 64; FOV, 24 cm × 24 cm, slice thickness, 4 mm (no gap); slice number, 35 (each volume). In total, 255 volumes were acquired.
Data Preprocessing
Image pre-processing was based on the SPM12 (Statistical Parametric Mapping 12) toolbox1. For resting-state fMRI data, the first five scans of each subject were discarded for magnetization equilibrium. The remaining 250 images were slice-time and head-motion corrected. One pilot was excluded from the analyses due to excessive head motion (a maximum displacement exceeding 2 mm in any direction or a maximum spin exceeding 2°). Demographic characteristics of the two groups are summarized in Table 1. Head motion differences between the groups were compared. In addition, the head motion of each participant was analyzed using the following formula:
Table. Demographic characteristics of the two groups. (See the original article for the full table.)
M is the time point number; i represents a certain time point; x1, y1, and z1 are the translations at certain time point; x2, y2, and z2 are the rotations at certain time point.
For the structural scans, the images were segmented into white matter, gray matter, and cerebrospinal fluid. A specific template of this study for normalization was created using the DARTEL toolbox. Subsequently, the functional images were coregistered to the structural images and transformed into standard MNI space (3 mm × 3 mm × 3 mm). Finally, the images were smoothed using 8 mm full-width at half maximum Gaussian kernel.
Group-ICA
Group ICA is a data-driven method that can identify intercorrelated slow fluctuations in the fMRI signals. GIFT software (version 1.4b), run under MATLAB 2013b, was used to perform spatial ICA (Calhoun et al., 2001). The minimum description length criterion was used to estimate the dimensions of the datasets from the 49 subjects. Thirty-five independent components were ultimately determined. Data from all subjects were concatenated and principal component analysis was used for these datasets to reduce the temporal dimension. The data were decomposed using the Informax algorithm by independent component estimation. This operation was repeated 20 times in ICASSO. The cluster size was set as 16 to get the stable run. Finally, individual subject components were back-reconstructed into single-subject space and the intensity values were scaled to Z scores.
A two-step process was used to select the component that best matched the DMN. First, a frequency filter was applied to exclude any components in which high-frequency signal (>0.1 Hz) constituted ≥50% of the power. Then, the DMN component was selected using an automated template-matching procedure combined by four expert inspection. The components that exhibited close relationship with the DMN template (Luo et al., 2011) and low spatial overlap with known vascular, ventricular, motion and other susceptibility artifacts, were selected.
A two-sample t-test was conducted to quantitatively compare the DMN between pilots and controls (GRF correction, p < 0.05, with an initial height threshold of p < 0.001), controlling for age and head motion effects. This analysis was performed within the mask, which resulted from the one-sample t-test of the group (p < 0.001).
Functional Connectivity Analysis
Brain regions that exhibited significant group differences were selected as seeds to further calculate the functional connectivity map. The smoothed images were further regressed using six head motion parameters, global mean signal, white matter signal, cerebrospinal fluid signal, and linear drift signal (Dong et al., 2018). The resulting images were then band-pass filtered (0.01–0.08 Hz) (Fox et al., 2005). Pearson correlation coefficients between time courses of the seeds and of the remainder of the voxels of the entire brain were calculated and Fisher Z transformed. Group differences were obtained using the two-sample t-test (GRF corrected, p < 0.05, with an initial height threshold of p < 0.001) which was performed within the mask obtained from one-sample t-test (FDR corrected, p < 0.05), controlling for age and head motion effects.
Finally, the relationship between the Z value indexes of functional properties and the total flight hours were determined by calculating the Kendall correlation coefficient.
CC BY 4.0. "Altered Default Mode Network Dynamics in Civil Aviation Pilots" by Chen X, Xu K, Yang Y et al., Frontiers in neuroscience (2019). https://doi.org/10.3389/fnins.2019.01406. Licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). Peer-reviewed and open access; full text from PubMed Central (PMC6971098). Headings and layout adapted for reading; figures and tables are in the original.