ICN_Atlas: Automated description and quantification of functional MRI activation patterns in the framework of intrinsic connectivity networks

• Published in: NeuroImage (Orlando, Fla.) Vol. 163; pp. 319 - 341
• Main Authors: Kozák, Lajos R., van Graan, Louis André, Chaudhary, Umair J., Szabó, Ádám György, Lemieux, Louis
• Format: Journal Article
• Language: English
• Published: United States Elsevier Inc 01.12.2017; Elsevier Limited; Academic Press
• Subjects: Activation; Atlases as Topic; Automation; Brain; Brain - physiology; Brain atlas; Brain mapping; Brain Mapping - methods; Data analysis; Downloading; EEG; Electroencephalography; Epilepsy; Functional characterization; Functional magnetic resonance imaging; Generalized linear models; Humans; Image Interpretation, Computer-Assisted - methods; Intrinsic connectivity networks; Localization; Magnetic Resonance Imaging - methods; Medical imaging; Nerve Net - physiology; Neural networks; Neural Pathways - physiology; NMR; Nuclear magnetic resonance; Principal components analysis; Resting-state networks; Seizures; Sleep; Variance analysis; Canada; Montreal Quebec Canada; Intrinsic connectivity networks; Brain atlas; Functional magnetic resonance imaging; Data analysis; Functional characterization; Resting-state networks
• ISSN: 1053-8119; 1095-9572
• DOI: 10.1016/j.neuroimage.2017.09.014

Generally, the interpretation of functional MRI (fMRI) activation maps continues to rely on assessing their relationship to anatomical structures, mostly in a qualitative and often subjective way. Recently, the existence of persistent and stable brain networks of functional nature has been revealed; in particular these so-called intrinsic connectivity networks (ICNs) appear to link patterns of resting state and task-related state connectivity. These networks provide an opportunity of functionally-derived description and interpretation of fMRI maps, that may be especially important in cases where the maps are predominantly task-unrelated, such as studies of spontaneous brain activity e.g. in the case of seizure-related fMRI maps in epilepsy patients or sleep states. Here we present a new toolbox (ICN_Atlas) aimed at facilitating the interpretation of fMRI data in the context of ICN. More specifically, the new methodology was designed to describe fMRI maps in function-oriented, objective and quantitative way using a set of 15 metrics conceived to quantify the degree of ‘engagement’ of ICNs for any given fMRI-derived statistical map of interest. We demonstrate that the proposed framework provides a highly reliable quantification of fMRI activation maps using a publicly available longitudinal (test-retest) resting-state fMRI dataset. The utility of the ICN_Atlas is also illustrated on a parametric task-modulation fMRI dataset, and on a dataset of a patient who had repeated seizures during resting-state fMRI, confirmed on simultaneously recorded EEG. The proposed ICN_Atlas toolbox is freely available for download at http://icnatlas.com and at http://www.nitrc.org for researchers to use in their fMRI investigations. [Display omitted] •fMRI activation maps are usually described in relation to anatomical structures.•ICN_Atlas describes fMRI data in the framework of intrinsic connectivity networks (ICN).•ICN_Atlas output: ICN ‘engagement’ (spatial and activation strength) metrics.•ICN_Atlas shows high ICN engagement reliability on test-retest resting-state fMRI data.•Method demonstrated (1) on parametric task-modulation fMRI data set shows differential engagement of ICN with respect to task difficulty, and (2) on epileptic seizure EEG-fMRI data shows compelling evolution of ICN engagement patterns.