Quantcast
Channel: NITRC CONN : functional connectivity toolbox Forum: help
Viewing all 6899 articles
Browse latest View live

Repetition of prepossessing steps in 2018b version

$
0
0
Dear Experts,

I am using conn2018b. I did processing steps with default pipeline for my resting state fMRI data. I entered 34 interleaved slice orders at the beginning manually.  After all pre-processing steps swaugrp240_rs.nii files were generated for all subjects.  A pop window asking to enter slice order from 1 to 91 (actual not the correct one). Though I entered the correct slice order, it was asking to enter 91 slice order. Then I selected interleave - middle top option. It started running again the preprocessing steps considering the swaugrp240_rs.nii file as initial file. Then I cancelled the preprocessing steps. I never experienced this repetition in the older version. I don't know where I am doing mistake. Could you please correct me if I am doing mistake anywhere? Please see the attached log file and have a look on the line [u][b][i]4744. [/i][/b][/u]

Thanks in advance
Ramesh

Pending jobs after SGE processing

$
0
0
Hi all

I've performed the Setup step using a SGE computing environment to separate into several jobs (n=20, total of 180 subjects). This runs smoothly, jobs are (apparently) merged, and I can move on to the Denoising step. However, once I want to start Denoising, Conn tells me there are still pending jobs, and starts merging again. This then takes ages, and Conn then saves, but jobs remain pending. I'm at loss as to why this happens. Has anyone run into similar issues?

Thanks for any pointers!

Eugenio

Reporting correct T- or Z-Statistics

$
0
0
Hi,

in typical fMRI analysis we report z- or T-scores that are reported in the whole brain statistics under SPM. In Conn, however, I am wondering how to correctly report T- or Z-scores and how I can obtain these? I read that using the T-scores under Plot Effects should not be used.

Thanks,
Stephanie

map surface to volume

$
0
0
Dear expert,

I have done a ROI surface-based connectivity analysis with CONN, using for every subjects their FreeSurfer native ROIs  .
After all the preprocessing I computed In the firstlevel folder BETA nifti files, that should be the surface functional connectivity maps for each ROIs I entered in the analysis.

Do you know how can I map these nifti files  to MNI volumetric space?

Thanks,
Lorenzo

Second-level covariate of interest with gPPI

$
0
0
Hi Alfonso!

I have a quick question with gPPI and covariates I would like to confirm the answer with you.

[u]My design:[/u]
1) I have a 4 conditions (2x2 factorial design) repeated measures. I am interested in the interaction term.
2) I have a behavioural covariate inserted as a second-level covariate (one value per subject).
3) I perform gPPI with one seed region. 

[u]What I am interested in:[/u]
I want to find if there are any voxels increasing their connectivity with my seed as a function of the behavioural covariate in the interaction contrast. I hypothesise that people with greater value in the covariate would show increased connectivity with the seed in this contrast than others.

[u]How I test this[/u]:
- In the subject effects, I selected 'All subjects' and my behavioral covariate and I set the contast [0 1].
- In the conditions, I selected my 4 conditions and I set my interaction contrast [-1 1 1 -1].
- In the seeds, I selected my seed.


Is this the combination of contrasts that correctly addresses the question?
Moreover, could you please write the complete GLM model of gPPI with the covariate? Would the model be this one (gppi: https://web.conn-toolbox.org/measures/seed-based) with an extra regressor? Which coefficient is being tested in the above-mentioned contrast? I suppose the coefficient of the covariate. The significance of this coefficient is tested for remaining variance (after removing the variance explained by the interaction only) right? 

Thank you,
Konstantina

Seed to voxel report

$
0
0
Hi,

I have done a seed to voxel analysis recently. I have a basic question about how to report the result. My seed was a hippocampus structural mask and I compared its functioncal connectivty with the whole brain between two condiions. The result was a netwrok that included several clusters. but I'm not sure how to report this result. For example if the netwrok contained a special cluster (e.g. DLPFC), can I say that hippocampus showed a significant change in connectivity with that cluster (DLPFC) in one condition compared with the other condition? Can I infer about the connectivity of one of the clusters separately or I can only infer about the whole netwrok?


Thanks a million,
Haleh

How to get values of functional connectivity’s strength

$
0
0
Dear Experts,

I used CONN version 18b.
I want to get values of functional connectivity's strength shown by attached file.
I can see YData from properties editor, but I can't copy YData.

Any suggestion would be highly appreciated.
Thank you so much,
Naoki Hirabayashi

Connectivity in Subject Space

$
0
0
Hi CONN Users,

I have reconstructed the structural image for each subject using Freesurfer. Also, I have an atlas for each subject and want to process connectivity using each subject's atlas. What would be the best way of doing this using CONN?

Thanks,
Tuner

RE: How much variance is removed after preprocessing/denoising?

$
0
0
Hello!

I just wanted to follow up on this question since I'm still unsure about it.

Thanks!
Kaitlin

Error in Preprocessing

$
0
0
I have received an error message I have not seen before in preprocessing of data in Conn version 18.b. I am running the standard preprocessing pipeline, and after the first step (Functional realignment and unwarp) completes, I get the following error:

ERROR DESCRIPTION:

Error using matlab.ui.control.UIControl/set
Invalid or deleted object.
Error in conn_disp (line 157)
set(CONN_h.screen.hlogstr,'value',[]);
Error in conn_msgbox (line 40)
if ok==0, conn_disp(char(txt)); end
Error in conn_setup_preproc (line 2578)
if dogui, hmsg=conn_msgbox({'Importing results to CONN project','Please wait...'},'');
Error in conn (line 1081)
ok=conn_setup_preproc('',varargin{2:end});
Error in conn_menumanager (line 120)
feval(CONN_MM.MENU{n0}.callback{n1}{1},CONN_MM.MENU{n0}.callback{n1}{2:end});
CONN18.b
SPM12 + DEM FieldMap MEEGtools
Matlab v.2017b
project: CONN18.b
storage: 661.7Gb available
spm @ C:\Matlab\spm12\spm12
conn @ C:\conn18b\conn

________

The 'u' prefixed .nii files are created but not imported into Conn, and the rp_.txt files are created. The processing then stops and does not continue. Any insight into what may be causing this error and how to proceed? Thanks!

Why denoised rfMRI data have normal density

$
0
0
Hi,

After performing essential preprocessing steps (slice timing, realignment, co-registration, segmentation, normalization, and smoothing) in SPM12, I loaded the normalized T1 image and also normalized (not smoothed) Functional images into CONN toolbox. Then, I defined the motion parameters (extracted by SPM12) in 1st level covariates. In the option settings, I selected only voxel-voxel analysis and "create confound-corrected time series". Finally, I press "Done" button to go for "Denoising" step. In the "Denoising" step, I considered the "White Matter (5P)", "CSF (5P)", and "Motion Parameters" as confounds. The band-pass filter was [0.01, 0.1]Hz and linear detrending was selected.
I have provided the density plot of my normalized rfMRI (loaded as functional images in CONN toolbox) and also density plot of its denoised version in CONN toolbox. They have completely different density and the denoised version shows a normal density. I know the connectivity density is normal (as shown in CONN toolbox) but I don't know why my denoised rfMRI time series show normal density?
In another try, I loaded my rfMRI data in CONN toolbox and carried out "Setup" step as described above. However, in "Denoising" step, I didn't select any confounds, any despiking, any detrending and band-pass filter [0.0001,200]Hz was applied. I expected to achieve rfMRI data showing density the same as or at least similar to my normalized rfMRI data. However, interestingly, the achieved rfMRI data again had normal density. I really don't know why? And what the CONN toolbox does when I have not selected any option for denoising?
I really am worried about my processed results and I don't know that can I trust on my denoised results?
I have considered "niftiDATA_Subject001_Condition000.nii' as denoised data.

Any response would be appreciated.

[img]https://www.dropbox.com/s/2rufoc5omuujw1u/Images.pdf?dl=0[/img].

Thanks,
Talesh

RE: Seed Based Analysis - Conceptual Question

$
0
0
Hi,
I'm bumping up my questions below in the hopes someone can offer clarification!
Kindly,
Paul

[i]Originally posted by Paul Cernasov:[/i][quote]Dear CONN Users,

Grad student here. I'm running Seed-based correlation analyses and I have several ROIs that I'm interested in probing with respect to my behavioral variables.

1) Am I correct in assuming that selecting "Any effect among ***" under Between-Sources Contrast displays results corrected for the number of seeds?

2) How is it possible to have significant results at a given height & cluster threshold when including two or more seeds in my Between-Sources contrast (any effect), but then a model with only seed (i.e. the one with significant connectivity from including multiple seeds) shows empty results? 

Cheers,
Paul[/quote]

Measure spatial overlap between networks

$
0
0
Hello,

I have two resting state networks of interest (using a seed to voxel approach), and I am wondering if there is a way to measure the magnitude of spatial overlap between the two networks (i.e., how spatially separate are these two networks?) Is there a way to quantify this using CONN?

Thanks!

How to remove first-order derivatives of realignment parameters in conn_batch_humanconnectomeproject's

$
0
0
[b]onDear CONN developers,[/b]

Thank you for your great toolbox, I really enjoy using it!

I have a problem regarding the editing of CONN_batch_humanconnectomeproject. Since HCP has included the motion parameters and its derivatives in its motion parameter text file, I didn't want to add first-order temporal derivatives to the the realignment confound parameters. So in the conn_batch_humanconnectomeproject I added:[quote][code]batch.Preprocessing.confounds.names={'White','CSF','scrubbing','realigment'};[/code][code]batch.Preprocessing.confounds.dimensions={5,5,6,12};[/code][code]batch.Preprocessing.confounds.deriv={0,0,0,0};[/code][/quote]
Unfortunately, after the processing of the data, the default setup won't get changed and CONN add first-order temporal derivatives of the realignments which results in 24 parameters of realignment. Also, even if "effect of rest" was not selected in the batch commands, after loading the results in the GUI, it was in the list of the confounds.

I know this problem can be solved by editing the GUI but due to the large sample size of HCP, I prefer using the batch commands.

Thank you!

Best,
Mohammad

RE: how to do functional preprocessing and denoising only with structural image preprocessed

$
0
0
Hello,

Did you figure out what the issue was? I'm running into a similar issue, although it gets through importing ROI data and then throws the error.

Ariana

CONN stalls while importing functional data

$
0
0
Hi all, 

I'm using the script below, it is based on the conn_batchexample_singlesubject.m Although it starts successfully, it stall while importing the functional data. When performing the same steps through the GUI, functional data import takes roughly 5 minutes; the script has been frozen for 1 hour now. 

Does any of you see my error? any help would be hugely appreciated!

Thank you

Eugenio

The CONN log is here:

CONN: RUNNING SETUP STEP
Step 1/7: Checking data completeness
spm @ /nan/ceph/network/system/el6/spm/spm-12-20181114
conn @ /home/k1632452/Documents/MATLAB/tools/conn
Checking if data files have been edited or moved
Step 2/7: Segmentation
Step 3/7: Importing conditions/covariates
Step 4/7: Importing functional data

It remains frozen at step 4/7

The script is below:

%% Preliminaries
% Addpath
addpath('/home/k1632452/Documents/MATLAB/tools/conn');

% Set data / results directory
subjDir = getenv('subj_dir'); % environment variables are set before the start of the script. this is to allow processing on a computer cluster
subjID = getenv('subj_id');
resultsDir = '/data/group/epilepsy/Abela/results/GSP1000';

% Select functional / anatomical volumes (already preprocessed externally)
func = cellstr(spm_select('FPListRec',subjDir,'^cwar'));
anat = cellstr(spm_select('FPListRec',subjDir,'^wmSub.*\.nii$'));

% Select tissue masks
gm = cellstr(spm_select('FPListRec',subjDir,'^mwp1'));
wm = cellstr(spm_select('FPListRec',subjDir,'^mwp2'));
csf = cellstr(spm_select('FPListRec',subjDir,'^mwp3'));

% Select regions of interest
roifiles = cellstr(spm_select('FPListRec','/data/group/epilepsy/Abela/data/ROI/Electrodes','ROI'));
roinames = cell(length(roifiles),1);
for roi = 1:length(roifiles)
[~,nam] = fileparts(roifiles{roi});
roinames(roi) = cellstr(nam);
end

% Select Volterra-expanded movement parametrs
vxmov = cellstr(spm_select('FPListRec',subjDir,'^vx'));

% Set erosion threshold for tissue masks
thresh = [0.5 0.5 0.5];

% Set TR (seconds)
TR = 2;

%% CONN New experiment
mkdir(resultsDir,['conn_' subjID]);
clear batch
batch.filename=fullfile(resultsDir,['conn_' subjID '.mat']);

%% CONN Setup
batch.Setup.nsubjects = 1;
batch.Setup.functionals{1}{1} = func;
batch.Setup.structurals{1}= anat;
batch.Setup.RT = TR;
batch.Setup.masks.Grey{1} = gm;
batch.Setup.masks.White{1}= wm;
batch.Setup.masks.CSF{1} = csf;
batch.Setup.binary_threshold = thresh;
batch.Setup.erosion_steps= [0 1 1];
batch.Setup.erosion_neighb= [1 1 1];
for count = 1:length(roifiles)
batch.Setup.rois.files{count}{1} =roifiles(count);
end
batch.Setup.conditions.names = {'rest'};
batch.Setup.conditions.onsets{1}{1}{1} = 0;
batch.Setup.conditions.durations{1}{1}{1}= Inf;
batch.Setup.covariates.names{1} = 'Friston24';
batch.Setup.covariates.files{1}{1}{1} = vxmov;
batch.Setup.outputfiles= [0 1 0 0 0 1];
batch.Setup.isnew=1;
batch.Setup.done =1;

%% CONN Denoising
batch.Denoising.filter = [0.009, 0.08]; % [0.01 0.1]
batch.Denoising.detrending= 1;
batch.Denoising.confounds.names{1} = 'Friston24';
batch.Denoising.confounds.names{2} = 'White Matter';
batch.Denoising.confounds.dimensions{2} = 5;
batch.Denoising.confounds.names{3}= 'CSF';
batch.Denoising.confounds.dimensions{3} = 5;
batch.Denoising.done=1;

%% CONN Analysis
batch.Analysis.analysis_number=1; % Sequential number identifying each set of independent first-level analyses
batch.Analysis.measure=1; % connectivity measure used {1 = 'correlation (bivariate)', 2 = 'correlation (semipartial)', 3 = 'regression (bivariate)', 4 = 'regression (multivariate)';
batch.Analysis.weight=1; % within-condition weight used {1 = 'none', 2 = 'hrf', 3 = 'hanning';
batch.Analysis.sources={}; % (defaults to all ROIs)
batch.Analysis.done=1;

%% CONN Run batch
conn_batch(batch);

RE: Error while importing ROI data

$
0
0
Hi Boris, 

I thin the error is in this line

mismatched number of scans and covariate datapoints ([2 1] vs. [800 16])

You seem to have loaded only two functional files, but your 16 covariates are 800 TR long.

This mismatch causes the error. Try reloading your functional files, and take care to add all the volumes you need.

Kind regards

Eugenio

RE: Extracting ROI info into correlation matrix

$
0
0
Hi, sorry for the late reply - I haven't tried, been stuck on more basic processing issues. Best to try Raphael directly, he can certainly help!

Best of luck, 

Eugenio

RE: Pending jobs after SGE processing

$
0
0
Nevermind, I solved it. This was something completely unrelated to CONN, but rather how our local SGE interacted with MATLAB.

Cheers

Eugenio

RE: PET-data

$
0
0
[i]Originally posted by Duncan Green:[/i][quote]Hi Andreas,

My lab and myself have been doing this for the past while. PET data obtained from various ROIs around the brain are entered in as second-level covariates and used as a correlation analysis with different ROI-ROI contrasts.

Regards,
-Duncan[/quote]

[color=#000000]Dear Duncan,[/color]

[color=#000000]I am at a point now where I want to perform this in my analysis, but I am not totally clear on how to implement it.[/color]

Do you input one (average) PET value per subject for each ROI?
And how do you obtain the value, given that the PET values are relative (no arterial data).

Best,
Andreas
Viewing all 6899 articles
Browse latest View live


<script src="https://jsc.adskeeper.com/r/s/rssing.com.1596347.js" async> </script>