Subject sample and neuropsychological assessment
Thirteen HCs and fourteen PGs were recruited (all right handed males; drugs free; mean age 35.96±9.56; no schooling differences).
In these subjects,
pathological gambling trend was measured by the South Oaks Gambling Screen (SOGS) and Gambling Attitudes and Beliefs Survey (GABS).
The severity and type of obsessive-compulsive symptoms was measured by the Yale-Brown Obsessive Compulsive Test (YBOCS),
anxiety by the State-Trait Anxiety Inventory (STAI),
severity of depression by Beck Depression Inventory (BDI),
personality/behavioral construct of impulsiveness by Barratt Impulsiveness Scale (BIS) and perception of stress by the Perceived Stress Scale (PSS).
MRI data acquisition
All subjects underwent brain MRI scan using a 1,5T magnet (Signa HDxt; GE Medical System,
Milwaukee,
Wisconsin,
USA); an eight-channel brain phased array coil was used for radiofrequency transmission and reception.
Foam pads was placed on both sides of the head,
within the head coil,
to limit head motion during the scan.
All participants were explicitly instructed on not moving during the MR scan.
Structural images were obtained via a T1-weighted sagittal three-dimensional (3D) Fast SPoiled GRadient-echo (FSPGR) prepped inversion recovery (acquisition matrix 256 x 256; slice thickness 1,2 mm; TR 12.4 ms; TE 5 ms; IT 450 ms; FA 20; parallel imaging method: Array coil Spatial Sensitivity Encoding,
ASSET).
Resting state functional magnetic resonance imaging (fMRI) data were acquired with a two-dimensional (2D) axial T2*-weighted gradient-echo Echo-Planar (EP) pulse sequence parallel to the anterior commissure–posterior commissure (AC–PC) line over the entire brain (acquisition matrix 64 x 64; 33 slices; slice thickness 3 mm; gap 1 mm; TR 3000 ms; TE 60 ms; FA 90).
All participants were asked to quietly rest in the scanner with their eyes open and not to think of anything specific.
A ten minutes (200 volumes) fMRI scan was performed on each participant.
Scan parameters were consistent for all imaging sessions.
Data analysis
All the preprocessing was performed using FSL’s recommended preprocessing pipeline from FMRIB’s Software Library (FSL version 5.0 - https://fsl.fmrib.ox.ac.uk/fsl/fslwiki).
Preprocessing included motion correction,
removal of non-brain structures,
high pass temporal filtering with sigma = 100.0 s,
pre-whitening and global spatial smoothing using a filter with a 6 mm kernel.
After the pre-processing the functional scans were aligned to the high-resolution template EPI scan using non-linear registration with 12 degrees of freedom as implemented in FLIRT,
followed by nonlinear (FNIRT) warping [6-7].
We used FSL’s MELODIC software for probabilistic independent component analysis [8].
The multisession temporal ICA concatenated (Concat-ICA) approach,
as recommended for resting state data analysis [9-10],
allowed the inputting of all subjects from the two groups in a temporally concatenated fashion for the ICA analysis.
Concat-ICA yielded different components without the need for specifying any explicit time series model.
A total of 40 independent components (IC maps) were extracted.
A mixture model approach was used to perform the inference on estimated maps.
An alternative hypothesis test based on fitting a Gaussian/gamma mixture model to the distribution of voxel intensities within spatial maps [9] was used to threshold the IC maps.
The analysis for the differences between groups was carried out using an FSL dual regression technique that allows for voxel-wise comparisons of resting-state fMRI [11-12].
Non-parametric permutation based inference analysis [13] was performed.
For each analysis we ran 10000 randomized permutations,
while threshold-free cluster enhancement (TFCE) [14] was used for statistical inference to validate the likelihood of extended areas of signal,
which also takes into account information from neighboring voxels.
TFCE enhances cluster-like structures but the image remains fundamentally voxelized.
This cluster enhancement make the analysis more sensitive than the voxel-wise thresholding approach alone.
Correction for multiple comparisons across space was applied assuming an overall significance of p < 0.05 using permutation testing and TFCE.