Abstract:Numerous brain imaging studies revealed that human behavior and cognition,which included perception,attention,memory and personality,are complex processes.The neural correlates underlying these fundamental cognitive functions are not associated with a particular brain region,but closely related with our human brain functional networks.In recent years,the development of new brain connectivity technologies for mapping the whole brain functional networks advanced the knowledge of comprehensive neuronal circuits and systems.For instance,we can examine the architecture of these functional networks by a wide variety of graph theory tools.Previous studies suggested that brain networks were modular and hierarchically organized,and they consisted of blocks or subnetworks that were particularly densely connected,but not spatially depended.Recently,functional magnetic resonance imaging (fMRI) studies uncovered that brain functional connectivity was inter-subject different.Individual differences in human behavior and cognition might associate with inter-subject variability in cortical functional systems.Two major sources of these variances might contribute to inter-subject variability in brain functional networks:the spatial distribution variability and the connection strength variability.Specifically,different subjects would show variable functional connectivity locations in connected to one same region of interest.In addition,individual differences in cortical functional networks were heterogeneous,with significantly higher variability in heteromodal association cortex (fronto-parietal control and attentional networks) and lower variability in unimodal cortices (sensory-motor and visual systems).Therefore,it is necessary and crucial to parcellate the individual brain functional networks,if we would like to conduct the application of imaging research findings.In compared to the task fMRI,resting-state fMRI has the advantage of implementation in clinic and high reliability of functional parcellation.Thus,resting-state fMRI is widely used to build the cortical functional networks.In the present paper,we first described several ways for parcellating functional networks,which is based on group's resting state fMRI data.They included cluster analysis and independent component analysis (ICA).With clustering approach,previous study found that human brain could be parcellated into 7 or 17 large-scale cerebral networks.Then,the multigraph K-way clustering method was proposed to identify the network node and find that a total of 100,200 and 300 subunits could represent the whole brain.Moreover,another researcher generated a functional brain atlas via spatially constrained spectral clustering including.ICA,a data driven method,was another widely applied approach to explore the brain functional networks.Each spatial component was independent,and represent a source of signal.However,it was still a challenge to decide the number of spatial component.Finally,Gordon and his colleagues came up with a technique to identify the boundary to define the represent putative cortical areas,which could be better after combining Wig et al's snowball sampling.Next,we illustrated the methods for parcellating individual cerebral functional networks.With resting-state fMRI data,researchers from Massachusetts General Hospital developed an approach to accurately map individual cortical functional organization.Crucially,in multiple iterative processes,Yeo's functional atlas and inter-individual variability were employed to direct the functional networks,which achieved highly reproducible within subjects and effectively captured the intra-subjects' difference.In addition,a combined method from Gordon's boundary approach and Wig's snowball sampling could also be employed to parcellate functional organization at individual level.Finally,we thought that the individual difference in functional networks might drive from cortical expansion during human evolution and necessary in high-level cognitive function.With the character of functional connectivity,we should pay much attention to individual difference in functional organization in the aspects of personalized medicine and the knowledge of brain's function.

Functional Parcellation of fMRI data using multistage k-means clustering

Clustered Cnmf for Fmri Data Analysis

Clustering Analysis for Fmri Dataset Based on ISODATA Algorithm

Multi-Session Parcellation of the Human Brain Using Resting-State Fmri.

A fusion approach based on multi-clustering to explore brain functional connectivity network

Which fMRI clustering gives good brain parcellations?

Functional parcellation of the hippocampus by semi-supervised clustering of resting state fMRI data

Parcellation of Fmri Datasets with ICA and PLS--a Data Driven Approach.

Exploring Functional Connectivity Networks in fMRI Data Using Clustering Analysis.

Intrinsic Functional Connectivity Pattern Based Brain Parcellation Using Normalized Cut

A Supervoxel-Based Method for Groupwise Whole Brain Parcellation with Resting-State Fmri Data

Individual parcellation of resting fMRI with a group functional connectivity prior

Concurrent brain parcellation and connectivity estimation via co‐clustering of resting state fMRI data: A novel approach

From Groupwise to Individual Brain Functional Networks Parcellation and Application

Analysis of FMRI data using an integrated principal component analysis and supervised affinity propagation clustering approach.

Using deep clustering to improve fMRI dynamic functional connectivity analysis

Frequency Clustering Analysis for Resting State Functional Magnetic Resonance Imaging Based on Hilbert-Huang Transform.

Functional Clustering: Identifying Strongly Interactive Brain Regions in Neuroimaging Data

A Hybrid Point-Sets Clustering Approach to Identification of Resting State Functional Network.

Parcellating Whole Brain for Individuals by Simple Linear Iterative Clustering

Artificial Bee Colony Clustering with Self-Adaptive Crossover and Stepwise Search for Brain Functional Parcellation in Fmri Data.