Download presentation
Presentation is loading. Please wait.
Published byRoy Buck Murphy Modified over 7 years ago
1
Seed-based functional connectivity and its applications
I3T Workshop Seed-based functional connectivity and its applications Xiangyu Long Postdoctoral fellow Developmental Neuroimaging Lab February 15th, 2017
2
Acknowledgement China: Dr. Yufeng Zang; Germany: Dr. Arno Villringer and Dr. Daniel Margulies; Canada: Dr. Catherine Lebel;
3
Last but not least: Internet
Before we start… Mac OS or Linux system Matlab: SPM Nifti Tools BrainNet Viewer REST toolbox GRETNA AFNI Last but not least: Internet
4
Outline Background The applications How to perform Parcellation analysis Graph-theory based analysis
5
Background Why use seed-based functional connectivity (seed-FC) analysis? Straight forward; Easy to learn and understand; Easy to implement; Fast; Easy to interpret;
6
Background Margulies et al.,2010
7
Background Resting-state fMRI (rs-fMRI)
No need to perform any tasks during scanning, but are instructed to remain awake; Eyes closed: might fall asleep, less extra stimulation; Fixation on screen: more extra stimulation, less head motion; Infant: fall asleep; Van Den Heuvel et al ., 2010
8
Background Low frequency fluctuations (LFFs, Hz) of blood-oxygen level dependent (BOLD) signals is related to spontaneous neuronal activity. The cardiac cycle The respiratory cycle low-frequency fluctuations Beckmann et al;., 2005
9
Background Pearson correlation coefficient, p<0.05 TR BOLD
10
Background Functional networks Buckner et al;., 2013
11
FC can be positive and negative.
Background FC can be positive and negative. Fox et al;., 2005
12
FC could be indirect connection.
Background FC could be indirect connection. r(A,B) = 0.5 A B C Solution: Partial correlation
13
FC could be directional: effective connectivity.
Background FC could be directional: effective connectivity. Dynamic Causal modelling Granger causality A B
14
FC doesn’t mean structural connectivity.
Background FC doesn’t mean structural connectivity. White matter fibers Honey et al;., 2008
15
FC in the frequency domain: The Coherence Coefficient
Background FC in the frequency domain: The Coherence Coefficient Sun et al., 2003
16
FC analysis could be static and dynamic.
Background FC analysis could be static and dynamic. Static FC: r=0.3 Dynamic FC Sliding windows Hindriks et al., 2015
17
Background Types of FC analysis Voxel-wise ROI-wise
Vogel et al;., 2010
18
Types of the region-of-interest (ROI)
Background Types of the region-of-interest (ROI) Sphere Anatomical regions Activations Whole brain coordinates Atlas or manual Comparisons etc.
19
Signals from white matter and cerebral spinal fluid; Global signal;
Background Noise signals: Head motion; Signals from white matter and cerebral spinal fluid; Global signal; Power et al., 2014
20
Background Fisher’s z transform:
21
Background The Brodmann areas: Brodmann ., 1909
22
Background The Automated Anatomical Labeling template: widely used for the graph theory analysis Tzourio-Mazoyer., 2002
23
The Human Brainnetome Atlas:
Background The Human Brainnetome Atlas: Fan et al;., 2016
24
How do we implement the functional connectivity anaysis in neuroscience study?
25
From task activation to functional network
The applications From task activation to functional network Biswal et al;., 1995
26
Functional network development/plasticity
The applications Functional network development/plasticity Sherman et al;., 2014
27
Functional network changes between populations
The applications Functional network changes between populations Castellanos et al;., 2008
28
Functional network and behavior/age/gender etc.
The applications Functional network and behavior/age/gender etc. Mennes et al;., 2012
29
Investigating neuroanatomy
The applications Investigating neuroanatomy Margulies et al;., 2009
30
Graph theory based network analysis:
The applications Graph theory based network analysis: Bullmore et al;., 2009
31
Background Today’ workshop: Basic FC analysis Static, undirectional FC; Voxel-wise, ROI-wise; Preliminary applications: parcellation and graph;
32
How to conduct a resting-state fMRI functional connectivity study?
33
Set up the T1 and EPI sequence; Prepare a moderate visual stimuli:
How to perform Overview: Invite participants; Set up the T1 and EPI sequence; Prepare a moderate visual stimuli: A cross on screen An inscape movie etc.; Vanderwal et al., 2015
34
How to perform Scan; Transfer the data to your local hard drive; Pre-processing the data; Select the seeds; Generate FC maps/matrix; Statistical analysis; Discuss the results;
35
Pre-processing the data:
How to perform Pre-processing the data: T1 image segmentation: grey matter, whiter matter and cerebral spinal fluid; Slice timing; Head motion correction; Sptial normalization to a standard space (MNI); Nuisance signals removal: head motion, average signals within WM and CSF, scrubbing, ICA-fix, etc., Power et al;., 2014
36
How to perform Band-pass filtering (0.01 to 0.1 Hz); Linear trend removal; Spatial smoothing (~ 6mm FWHM);
37
How to perform AFNI: (Linux/Mac OS)
38
How to perform AFNI commands:
3dcalc -a subj01.nii –b subj02.nii -expr ‘a+b’ -prefix Corr_example_Z.nii Command input options Output, can be any name
39
How to perform Band-pass filtering:
3dFourier -lowpass 0.1 -highpass retrend –prefix rest.nii.gz rest_filt.nii.gz Linear trend removal: 3dTstat -mean -prefix ${rest}_filt_mean.nii.gz rest_filt.nii.gz 3dDetrend -polort 2 -prefix rest_dt.nii.gz rest_filt.nii.gz 3dcalc -a rest_filt_mean.nii.gz -b rest_dt.nii.gz -expr 'a+b' -prefix rest_pp.nii.gz Covariates regression: 3dmaskave –mask csf_mask.nii.gz -quiet rest_pp.nii.gz > csf.1D 3dmaskave –mask wm_mask.nii.gz -quiet rest_pp.nii.gz > wm.1D 3dmaskave –mask global_mask.nii.gz -quiet rest_pp.nii.gz > wm.1D 1dcat csf.1D wm.1D global.1D headmotion.1D > covs.1D 3dDetrend -prefix rest_proc.nii.gz -vector covs.1D -polort 1 rest_pp.nii.gz
40
Q&A So far, any questions?
41
How to perform Let’s create functional connectivity maps!
42
How to perform Quick & (not so) Dirty: Online FC tool: Neurosynth
43
How to perform AFNI: InstaCorr
44
How to perform AFNI: InstaCorr
45
Select the coordinates:
How to perform Traditional way: Select the coordinates: Greicius et al;., 2003
46
How to perform
47
How to perform
48
How to perform
49
How to perform Click on the AFNI interface
50
How to perform Click middle button or Option+click in Mac OS
51
How to perform Underlay: T1.nii Overlay: PCC_ROI.nii
52
How to perform Get the averaged time series within the ROI:
3dmaskave -quiet -mask PCC_ROI.nii rest_example.nii.gz >>seed.1D Generate the FC map: 3dfim+ -input rest_example.nii.gz -ideal_file seed.1D -mask BrainMask.nii -out Correlation -bucket Corr_example.nii Fisher’s z transform: 3dcalc -a Corr_example.nii -expr 'log((1+a)/(1-a))/2' -prefix Corr_example_Z.nii
53
A region from a atlas as the ROI: e.g., AAL atlas,
How to perform A region from a atlas as the ROI: e.g., AAL atlas, 3dcalc -a AAL_61x73x61.nii -expr 'within(a,10.5,11.5)' -prefix aal11.nii 3dresample -prefix AAL_epi.nii -inset AAL_T1.nii -master func.nii
54
How to perform After generate the FC maps: Interact with other measurements; Within group statistic; Between group statistic;
55
How to perform Interact with other measurements:
3dfim+ -input all_FC.nii.gz -ideal_file age.1D -mask BrainMask.nii -out Correlation -bucket FC_Age.nii …….. Age: ……
56
How to perform Within and between group statistics: e.g., t-tests
57
How to perform Check the results: ROI analysis
1) Label the clusters within the statistic maps: 3dclust -quite -savemask rois.nii ttest+orig. 2) Extract the ROI: 3dcalc -a rois.nii -expr 'within(a,2.9,3.1)' -prefix cluster3.nii 3) Calculate the FC value within the ROI: 3dmaskave -quite -mask cluster3.nii subject01fc.nii >> avgvalue.txt
58
How to perform Control FC value Patient
59
Q&A So far, any questions?
60
Parcellation analysis
Functional connectivity-based pacellation: Assume the neighbor voxels that have the similar functional connectivity patterns belong to the same function area; Create the study specific ROIs instead of using templates; Further FC analysis, subregions comparion, etc.
61
Parcellation analysis
Nebel et al., 2012.
62
Parcellation analysis
Paradigm: … … Functional connectivity maps k = 3 Similarity matrix
63
Parcellation analysis
Choose a ROI, e.g., AAL regions; Make sure the template is in the same space with functional images; Create voxel seeds within the ROI: (by NIFTI tools) a = load_nii(’ROI.nii'); b = find(a.img>0); c = a; for i = 1 : length(b) d = zeros(size(a.img)); d(b(i)) = 1; c.img = d; number = ['000',num2str(i)]; number = number(end-3:end); save_nii(c,['voxel_',number,'.nii']); end Calculate the FC map of each voxel; 59
64
Parcellation analysis
Calculate the similarity of each pair of the FC maps; Spatial correlation; eta2; Run classification on the similarity matrix; K-means: kmeans; Hierarchy clustering: linkage and dendrogram; Cohen et al;., 2009
65
Parcellation analysis
Number of clusters: Overlap between two datasets: Dice’s coefficient X Y
66
Graph theory-based analysis: Whole brain FC analysis;
How to perform Graph theory-based analysis: Whole brain FC analysis; Gnerally use regions instead of voxels; Small-world property; Brain regions FC values
67
Templates: AAL_90_3mm.nii or AAL_61x73x61_116.nii;
How to perform What we need: Templates: AAL_90_3mm.nii or AAL_61x73x61_116.nii; ROIs: numbered within the nifti file; Data: rest_example.nii.gz Bullmore et al;., 2009
68
How to perform Create the connectivity matrix:
3dNetCorr -inset rest_example.nii.gz -in_rois AAL_90_3mm.nii -fish_z -mask BrainMask.nii -prefix aal_fc_matrix Then switch to Matlab: conmat = ach_readnetcc(‘aal_fc_matrix_000.netcc’); save(‘conmat’,’conmat’); save('subj01.edge', 'conmat','-ASCII','-DOUBLE','-TABS');
69
How to perform View the connections: run BrainNet in Matlab;
70
How to perform
71
How to perform Surface:
72
How to perform Node:
73
How to perform “Load File” interface:
74
How to perform
75
How to perform
76
How to perform
77
How to perform Open Gretna in Matlab;
78
How to perform Select conmat.mat; Network Type: weighted;
Network Member: positive; Threshold Type: similarity; Select all network metrics; Select Output directory; Press RUN.
79
How to perform
80
Thank you for your time!
Similar presentations
© 2025 SlidePlayer.com Inc.
All rights reserved.