Functional connectivity analysis of fMRI data using parameterized regions-of-interest

Influence of ROI selection on resting state functional connectivity: an individualized approach for resting state fMRI analysis

The differences in how our brain is connected are often thought to reflect the differences in our individual personalities and cognitive abilities. Individual differences in brain connectivity has long been recognized in the neuroscience community however it has yet to manifest itself in the methodology of resting state analysis. This is evident as previous studies use the same region of interest (ROIs) for all subjects. In this paper we demonstrate that the use of ROIs which are standardized across individuals leads to inaccurate calculations of functional connectivity. We also show that this problem can be addressed by taking an individualized approach by using subject-specific ROIs. Finally we show that ROI selection can affect the way we interpret our data by showing different changes in functional connectivity with aging.

Functional connectivity-based parcellation of amygdala using self-organized mapping: a data driven approach

The overall goal of this work is to demonstrate how resting state functional magnetic resonance imaging (fMRI) signals may be used to objectively parcellate functionally heterogeneous subregions of the human amygdala into structures characterized by similar patterns of functional connectivity. We hypothesize that similarity of functional connectivity of subregions with other parts of the brain can be a potential basis to segment and cluster voxels using data driven approaches. In this work, self-organizing map (SOM) was implemented to cluster the connectivity maps associated with each voxel of the human amygdala, thereby defining distinct subregions. The functional separation was optimized by evaluating the overall differences in functional connectivity between the subregions at group level. Analysis of 25 resting state fMRI data sets suggests that SOM can successfully identify functionally independent nuclei based on differences in their inter subregional functional connectivity, evaluated statistically at various confidence levels. Although amygdala contains several nuclei whose distinct roles are implicated in various functions, our objective approach discerns at least two functionally distinct volumes comparable to previous parcellation results obtained using probabilistic tractography and cytoarchitectonic analysis. Association of these nuclei with various known functions and a quantitative evaluation of their differences in overall functional connectivity with lateral orbital frontal cortex and temporal pole confirms the functional diversity of amygdala. The data driven approach adopted here may be used as a powerful indicator of structure-function relationships in the amygdala and other functionally heterogeneous structures as well.

Task-driven ICA feature generation for accurate and interpretable prediction using fMRI

Functional Magnetic Resonance Imaging (fMRI) shows significant potential as a tool for predicting clinically important information such as future disease progression or drug effect from brain activity. Multivariate techniques have been developed that combine fMRI signals from across the brain to produce more robust predictive capabilities than can be obtained from single regions. However, the high dimensionality of fMRI data makes overfitting a significant problem. Reliable methods are needed for transforming fMRI data to a set of signals reflecting the underlying spatially extended patterns of neural dynamics. This paper demonstrates a task-specific Independent Component Analysis (ICA) procedure which identifies signals associated with coherent functional brain networks, and shows that these signals can be used for accurate and interpretable prediction. The task-specific ICA parcellations outperformed other feature generation methods in two separate datasets including parcellations based on resting-state data and anatomy. The pattern of response of the task-specific ICA parcellations to particular feature selection strategies indicates that they identify important functional networks associated with the discriminative task. We show ICA parcellations to be robust and informative with respect to non-neural artefacts affecting the fMRI series. Together, these results suggest that task-specific ICA parcellation is a powerful technique for producing predictive and informative signals from fMRI time series. The results presented in this paper also contribute evidence for the general functional validity of the parcellations produced by ICA approaches.