A human brain atlas derived via n-cut parcellation of resting-state and task-based fMRI data
Magnetic Resonance Imaging
The growth of functional MRI has led to development of human brain atlases derived by parcellating resting-state connectivity patterns into functionally independent regions of interest (ROIs). All functional atlases to date have been derived from resting-state fMRI data. But given that functional connectivity between regions varies with task, we hypothesized that an atlas incorporating both resting-state and task-based fMRI data would produce an atlas with finer characterization of
... regions than an atlas derived from resting-state alone. To test this hypothesis, we derived parcellation atlases from twenty-nine healthy adult participants enrolled in the Cognitive Connectome project, an initiative to improve functional MRI's translation into clinical decision-making by mapping normative variance in brain-behavior relationships. Participants underwent resting-state and task-based fMRI spanning nine cognitive domains: motor, visuospatial, attention, language, memory, affective processing, decision-making, working memory, and executive function. Spatially constrained n-cut parcellation derived brain atlases using (1) all participants' functional data (Task) or (2) a single resting-state scan (Rest). An atlas was also derived from random parcellation for comparison purposes (Random). Two methods were compared: (1) a parcellation applied to the group's mean edge weights (mean), and (2) a two-stage approach with parcellation of individual edge weights followed by parcellation of mean binarized edges (two-stage). The resulting Task and Rest atlases had significantly greater similarity with each other (mean Jaccard indices JI= 0.72-0.85) than with the Random atlases (JI=0.59-0.63; all p<0.001 after Bonferroni correction). Task and Rest atlas similarity was greatest for the two-stage method (JI=0.85), which has been shown as more robust than the mean method; these atlases also better reproduced voxelwise seed maps of the left dorsolateral prefrontal cortex during rest and performing the n-back working memory task (r=0.75-0.80) than the Random atlases (r=0.64-0.72), further validating their utility. We expected regions governing higher-order cognition (such as frontal and anterior temporal lobes) to show greatest difference between Task and Rest atlases; contrary to expectations, these areas had greatest similarity between atlases. Our findings indicate that atlases derived from parcellation of task-based and resting-state fMRI data are highly comparable, and existing resting-state atlases are suitable for task-based analyses. We Image Acquisition and Preprocessing 2.3.1. Image Acquisition-Imaging data were acquired using a Philips 3T Achieva Xseries MRI scanner (Philips Healthcare, Eindhoven, The Netherlands). Anatomic images were acquired with a MPRAGE sequence (matrix = 256 × 256, 220 sagittal slices, TR/TE/FA = shortest/shortest/8°, final resolution =0.94 × 0.94 × 1 mm 3 resolution). Functional images for early participants (001-050) were acquired using an 8-channel head coil with an echo planar imaging (EPI) sequence (TR/TE/FA = 2000 ms/30 ms/90°, FOV=240 × 240 mm, matrix = 80 × 80, 37 oblique slices parallel to orbitofrontal cortex to reduce sinus artifact, interleaved ascending slice acquisition, slice thickness = 4 mm, final resolution 3.0 × 3.0 × 4.0 mm 3 ). For these subjects, one session's resting-state scan was acquired with 3-mm slice thickness to be consistent with data acquired for other BIRC studies. Functional images for later participants (051-079) were acquired using a 32-channel head coil with the following EPI sequence parameters: TR/TE/FA = 2000 ms/30 ms/90°, FOV = 240 × 240 mm, matrix = 80 × 80, 37 oblique slices, ascending sequential slice acquisition, slice thickness = 2.5 mm with 0.5 mm gap, final resolution 3.0 × 3.0 × 3.0 mm 3 . James et al.