A Novel Feature-Map Based ICA Model for Identifying the Individual, Intra/Inter-Group Brain Networks across Multiple fMRI Datasets.

Chunqi Chang,Hongjie Yan,Nizhuan Wang,Weiming Zeng,Yuhu Shi

doi:10.3389/fnins.2017.00510

Abstract

Independent component analysis (ICA) has been widely used in functional magnetic resonance imaging (fMRI) data analysis to evaluate functional connectivity of the brain; however, there are still some limitations on ICA simultaneously handling neuroimaging datasets with diverse acquisition parameters, e.g., different repetition time, different scanner, etc. Therefore, it is difficult for the traditional ICA framework to effectively handle ever-increasingly big neuroimaging datasets. In this research, a novel feature-map based ICA framework (FMICA) was proposed to address the aforementioned deficiencies, which aimed at exploring brain functional networks (BFNs) at different scales, e.g., the first level (individual subject level), second level (intragroup level of subjects within a certain dataset) and third level (intergroup level of subjects across different datasets), based only on the feature maps extracted from the fMRI datasets. The FMICA was presented as a hierarchical framework, which effectively made ICA and constrained ICA as a whole to identify the BFNs from the feature maps. The simulated and real experimental results demonstrated that FMICA had the excellent ability to identify the intergroup BFNs and to characterize subject-specific and group-specific difference of BFNs from the independent component feature maps, which sharply reduced the size of fMRI datasets. Compared with traditional ICAs, FMICA as a more generalized framework could efficiently and simultaneously identify the variant BFNs at the subject-specific, intragroup, intragroup-specific and intergroup levels, implying that FMICA was able to handle big neuroimaging datasets in neuroscience research.

Highlights

Blood oxygen level dependent (BOLD) functional magnetic resonance imaging has been used as an effective neuroimaging tool to study functional connectivity, which can reveal the neural correlates of cognitive processes, among multiple cortical brain regions (Biswal et al, 1995, 1997; Kawashima et al, 2000; Greicius et al, 2003; Yang et al, 2014; Shi et al, 2015b)
The 12 sources determined by feature-map based ICA model (FMICA) at intragroup level were displayed in Figure 2, which were highly approximate to the simulated ground truth sources
As described in Section Some Key Points in FMICA Implementation, the ICASSO method was used to determine the optimal order for the intergroup-level analysis based on the stability measure of the estimated components for each order, as shown in Figure session 2 (S2), demonstrating that the estimated components had the highest mean/median stability and relatively small values of standard deviation (STD) and inter-quartile range (IQR), when the order was equal to 57

Summary

Introduction

Blood oxygen level dependent (BOLD) functional magnetic resonance imaging (fMRI) has been used as an effective neuroimaging tool to study functional connectivity, which can reveal the neural correlates of cognitive processes, among multiple cortical brain regions (Biswal et al, 1995, 1997; Kawashima et al, 2000; Greicius et al, 2003; Yang et al, 2014; Shi et al, 2015b). Since no prior knowledge on the spatial or temporal pattern prior of the BFNs is required, the data-driven methods are more widely used in functional connectivity study. Examples of such datadriven methods include spatial ICA (McKeown et al, 1998) and temporal ICA (Biswal and Ulmer, 1999), assuming the spatial and temporal independence, respectively, while probabilistic ICA (PICA) carries out a probabilistic modeling, to achieve an asymptotically unique decomposition of the fMRI data (Beckmann and Smith, 2004). Other ICA methods for fMRI data analysis include an approach making use of spatial regularity of sources (Valente et al, 2009), and the models combining the sparsity and the mutual independence of components (Calhoun et al, 2013; Wang et al, 2013, 2015), to improve the accuracy of the estimated brain sources

Results

Discussion

Conclusion