Quantcast
Channel: NITRC CONN : functional connectivity toolbox Forum: help
Viewing all 6914 articles
Browse latest View live

importing second level covariates

$
0
0
Hello
Is it possible to import multiple second level covariates at once from a text file? This function in the CONN gui seems to only allow for the selection of one covariate at a time

BOLD Signal Falling to Zero: What On Earth Is Happening Here?

$
0
0
I'm getting a very odd result: when I plot the ROI BOLD series after denoising, I get a very strange artifact: the BOLD signal will sometimes simply drop to 0, without any variation for seemingly random periods of time. If you imagine a normal, noisy looking waves, with large sections clipped out, you have a sense of what's going on.

This seems to occur in datasets at random: some are extremely vulnerable to it, while others don't show this artifact at all.

All the best
~Thomas

Questions about Harvard-Oxford Subcortical parcellation used for Conn toolbox

$
0
0
Hello
I have some questions regarding the default atlas used in CONN. I am trying to identify certain subcortical structures using the HarvardOxfrod_Subcortical.xml file (attached in this email for reference) and am finding that the atlas.nii file contained within the ROI files of the CONN toolbox does not match up with the label index of the file. Could you please provide the correct Harvard-Oxford lookup table for subcortical structures?

RE: Motion Parameters

$
0
0
Dear Boshra,

I would say that you are correct, and that you can either use the 12 parameters from the HCP dataset and remove the derivatives, or use only the 6 parameters without derivatives and then enable CONN calculation of derivatives, both should lead to the same results.

Hope this helps,
Best regards,
Stephen Larroque

Using CONN preprocessed and denoised data to do ICA in GIFT

$
0
0
Dear CONN users,

I successfully preprocessed and denoised my resting-state data in CONN. But when I try to import these preprocessed and denoised data (i.e., niftiDATA*_Condition000.nii) to GIFT to run ICA, I keep getting the following error at the very beginning (when GIFT tries to create a brain mask from the first files of each subject):

[color=#ff0000][b]Group ICA Error Information:[/b][/color]
[color=#ff0000][b][/b][/color]
[color=#ff0000][b]Error using icatb_createMask (line 176)[/b][/color]
[color=#ff0000][b]No voxels found. Error in creating Mask[/b][/color]

I was wondering if this is due to changing the global BOLD signal or averaging BOLD signals during the denoising process (after reading posts this post: https://www.nitrc.org/forum/message.php?msg_id=19383  and "[u]no significant voxels[/u]" problem). 

In GIFT, the first couple of steps are :

1) Creating mask file from functional data
2) Performing data intensity normalization 
3) PCA Type : Expectation Maximization<span id="_plain_text_marker"> </span>

<span>Has anyone used the denoised CONN data in GIFT to run ICA successfully? </span>
<span>Any help is highly appreciated!
</span>

dynamic ICA

$
0
0
Hello,

I tried to use dynamic ICA but I do not really understand it. (Neither the method nor how to interpret the results given by conn). Are there any tutorials or paper I may read?

Best regards,
Eli

Interpreting the Contrasts for ROI analysis

$
0
0
Dear Conn experts ,

I am doing ROI-ROI analysis  in which i am comparing two groups ( Group A and Group B) , Could you please let me know how to interpret the results of  contrast Group A > Group B :  [1 -1] , Does it imply Average connectivity ( increase/decrease) in ROI's of Group A >  Average Connectivity in Group B ?.


Thanks
Vasudev

multiple datasets per subject

$
0
0
Hello,

I have a relatively complex design and I'd really appreciate your inputs on whether I have defined the project correctly. In one fMRI session, I have a 10 minutes resting state pre stimulation, 10 minutes stimulation (i.e. task with block design), and then another 10 min resting state post stimulation. I want to investigate the changes in connectivity from each step to the next. I also need to look into these connectivity changes for whole brain and cerebellum separately, since I am using SUIT corrected cerebellar volumes. Finally, the task portion itself has rest blocks (20s) and stimulation blocks (24s). So to be able to compare all possible changes in connectivity, I thought I should define 8 datasets for each subject:

2 resting state pre stimulation (1 for whole brain & 1 for suit cerebellum)
2 task-rest blocks (1 for whole brain & 1 for suit cerebellum)
2 task-stimulation blocks (1 for whole brain & 1 for suit cerebellum)
2 resting state post stimulation (1 for whole brain & 1 for suit cerebellum)

Then, the contrasts on the second level would be defined accordingly. For example, for comparing connectivity in task-stimulation block whole brain with post resting state whole brain the contrast will be: [0 0 0 0 1 0 -1 0]

Am I on the right track? Or is there a better way to do this? I really appreciate any recommendations that you might have. 
Thanks!

RE: Cluster extent thresholding of first-level connectivity maps

$
0
0
Dear Stephen,

Thank you very much for this exhaustive response!

For the voxel-wise correction, I checked a box in CONN set-up options to create first level seed-to-voxel FDR corrected p maps. Then I threshold them at 0.05 and use them to mask the corresponding first level r-maps. That way I get r maps at p=0.05 FDR corrected. Is your approach the same?

The problem is that I still get small clusters (k<5) of highly significant voxels, partially in a lesion site, so they are most probably meaningless. In order to clear my image from them, I wrote a function that applies a cluster extent threshold at a given k. Maybe someone will need it, so I put it below:


[code]function [no_small_cluster_matrix] = cluster_size_threshold(img,cluster_threshold)
%given an .nii activation/connectivity map (img), this function provides a
%cluster extend threshold at a given number of voxels (cluster_threshold)[/code][code]% example: cluster_size_threshold(pFDRcorrected_r_map.nii, 10)[/code][code]% output: a .nii file with the same name as input image and a suffix: k>cluster_threshold (eg. pFDRcorrected_r_map_k>10.nii)[/code][code]spm_struct = spm_vol(img);
matrix = spm_read_vols(spm_struct);[/code][code]I = find(matrix); %find non-zero voxels
[x,y,z]=ind2sub(spm_struct.dim, I);
L = [x y z ]'; %%% locations in voxels
clusterindex = spm_clusters(L); %finds indices of clusters[/code][code][cluster_size,cluster_number]=hist(clusterindex,unique(clusterindex));
big_cluster_indices = cluster_number(cluster_size>cluster_threshold);[/code][code]without_small_clusters_I = I(ismember(clusterindex, big_cluster_indices));[/code][code]template = zeros(spm_struct.dim);
lin_no_small_cluster_matrix = template(:);
lin_no_small_cluster_matrix(without_small_clusters_I)=1;
no_small_cluster_binary_matrix = reshape(lin_no_small_cluster_matrix, spm_struct.dim);[/code][code]no_small_cluster_matrix = no_small_cluster_binary_matrix.*matrix;[/code][code]no_small_cluster_struct_file = spm_struct;
no_small_cluster_struct_file.fname = [spm_struct.fname '_k>' num2str(cluster_threshold)'
no_small_cluster_struct_file = spm_create_vol(no_small_cluster_struct_file);
spm_write_vol(no_small_cluster_struct_file,no_small_cluster_matrix);
end[/code]
All the best, Kasia

low signal intensity issue

$
0
0
Hello!

I am running CONN Toolbox and examining connectivity using ROIs. We have signal dropout in an ROI that includes the OFC.

I know that you can mask to grey matter in the CONN toolbox, but I was hoping there might be another method of dealing with signal dropout (e.g., excluding signals with low intensity)?

Any help would be greatly appreciated.

Best,
Nicole

conn model set up

$
0
0
Greetings experts from down under 

i have run a data set with two time points one year apart in 12 patients and reached to setting up the model. Although i have gone through the previous posts on this topic, i am still clueless as to how this set up should be. I assumed that since there are no conditions in this and all are rest it should be rest for the first and second sessions.

(I must admit that i could not find how to change the condition with options like onset and duration as seperate columns even if i wanted to, but thats a different question), but i as, assumed its not the right thing to do.

Can some one advise me on this please.

One more question on the topic which i raised last time and i did not get any response is  if i want to compare the normal controls with these two time points data, is the design going to be different and what if the normal controls and patients are unequal in numbers, would be more different 

best regards
Joga

are motion regressors overkill with HCP FIX data

$
0
0
My student recently completed an analysis of HCP data using the FIX clean data following the conn_batch_humanconnectomeproject.m script. The script imports the motion regressors, and so conn uses them as confounds. A reviewer of the paper questioned whether it was appropriate to also include motion regressor confounds in the analysis, since the data have also been cleaned with FIX. I didn't see any problems with including them, as it seemed like the conservative approach, but I was wondering if anyone has looked into whether include motion regressor confounds might hurt the results.

Thanks

CONN crossover analysis

$
0
0
I am struggling to set up a contrast for what appears to be (at least for me) a complex analysis.  We have a one year crossover study with two groups:
Group one received drug for the first 3 months, placebo for the next 3 months, and was then on open label drug for the last 6 months.
Group two received placebo for the first 3 months, drug for the next 3 months, and was then on open label drug for the last 6 months.
We collected rsFMRI data at baseline, 3 months, 6 months, and 12 months.  
I set up conditions of: rest, On Drug, Off Drug, session 1, session 2, session 3, and session 4 (see attached: Image A). 
I set up second-level covariate groups of Drug first and Placebo first (Image B).  
We set up a weighted contrast model to compare these two groups: [-1.25 0.75 -1.25 1.75;-1.05 -1.05 0.1 2] but I am not sure how to set up the second-level analysis in CONN to use our weighted model to analyze both groups together.  As I have it set now it appears to be running both contrast models for both groups, so I am not sure how it is calculating a single p-value for individual ROI-to-ROI seed and target analysis (Image C).
If properly set up is the Intrinsic Connectivity V-to-V analysis the equivalent of a Flex. Factorial Analysis in SPM?
Thanks for any help.

ICA versus seed-based analysis

$
0
0
Dear All
Dear Alfonso

This question is a bit conceptual: what is the compatibility/ the common way of using ICA and seed-based method together? In practical terms, is it justified to take e.g. a significant ICA-cluster as seed? Or combining it with a priori seed from Conn (as example, the ICA gives an ACC cluster and you further take PCC as seed, or even ACC as well)? I am also thinking of using behavioral ratings with extracted values.

Some comments on that will be appreciated.
Best,
Lucas

Voxel-toVoxel analysis within a mask

$
0
0
Hi Alfonso,

I am interested in running a voxel-to-voxel analysis within a region of interest. Specifically, I'd like to compute the average correlation between a given voxel and the rest of the voxels within a mask; sort of a local coherence measure. The ILC analysis seems to do this within a kernel, is it possible to define a custom mask instead of that? My mask is highly unlikely to include anticorrelated regions so it shouldn't matter that much if I use average or RMS.

Thank you very much in advance,

Bengi

different number of slices in a within subject pre/post design

$
0
0
Hi all,

I'm trying to use conn to process some resting state data, and I've run into an issue that I was hoping you might be able to give me some advice on.

The study is a within subject placebo controlled drug challenge. Normally, there are 154 time points and 30 slices. For one participant on the drug visit, there are 154 time points, but only 26 slices (some error in the scanning! the placebo visit has the normal numebr of time points/slices) – resulting in an error during slice timing correction. Unfortunately, we've only got a very small sample, so can't afford to exclude this participant. Is there anything you'd suggest doing as an alternative?

Despite this not toolbox specific issue, conn's been brilliant!

Thanks,

A

decreased temporal SNR after preprocessing and denoising

$
0
0
Hi Alfonso,
   I preprocessed a resting state fMRI study, pretty much using the defaults (I am attaching the conn_batch script, just in case). Before preprocessing I computed a tSNR of about 96 using wau4D image or swau4Dimage.Unexpectedly, after preprocessing, I computed tSNR using the niftiDATA...subj..nii images stored in the /connproject/results/preprocessing folder, and I got a much lower result (about 35). To do that I had to rescale the niftiDATA..subj.nii as they were demeaned. I've been trying different thresholds in the art_threshold field in batch, with similar results. I am sure there is something I am doing wrong but I don't know what it is. Below there is the script to compute tsnr, but I don't think it's wrong.

Any help or suggestions would be appreciated.

Thanks

#!/bin/bash
cd /nfs/ep2/rest/DBtests/jobfiles/
for test in $( ls -d nifti*test ) ; do
cd /nfs/ep2/rest/DBtests/jobfiles/$test/connproj${test}/results/preprocessing/
for subj in $( ls -d niftiDATA*.nii ); do
echo $subj
min='fslstats $subj -R | awk '{print $1 }''
max='fslstats $subj -R | awk '{print $2 }''
scaling='echo "scale=5; 255.0 / ( $max - $min )" | bc'
fslmaths $subj -sub $min -mul $scaling ${subj}_scaled
fslmaths ${subj}_scaled -mas /nfs/ep2/rest/DBtests/nifti2/anatclear/epc139/ewc1c3Danat ${subj}_masked_scaled
fslmaths ${subj}_masked_scaled -Tmean mean${subj}
fslmaths ${subj}_masked_scaled -Tstd std${subj}
fslmaths mean${subj} -div std${subj} tSNR${subj}
echo "$test $subj" >> /nfs/ep2/rest/DBtests/tSNRall.txt
fslstats tSNR${subj} -M >> /nfs/ep2/rest/DBtests/tSNRall.txt
done
done

Single 2nd Level ROI-ROI Analysis with Some Missing ROI's

$
0
0
Hello,

I am working with functionally defined ROI's at the individual subject level using 4 different functional localizer scans. As you probably know, when doing this, sometimes an individual won't activate in a given region (say the visual word form area), thereby not creating a function ROI, but will in all the other ROI's. I am wondering if there is a way to still do a single 2nd level ROI-ROI analysis including those subjects? Someone told me I could enter "dummy" ROI's when setting up the project, then at the 2nd level input some vector with zeros for those subjects, so they are not included at all in a given analysis. However, that is not what I'm wanting. Here is a very simple example:

Say you have 3 subjects and creating 3 functionally defined ROI's for each (say maybe VWFA, TPJ and IFG).

You are able to define all 3 ROI's for subjects 1 and 2. However, for subject 3 only TPJ and IFG work.

Is there a way in a single network analysis of all three ROI's to get the connectivity between the TPJ and IFG including all 3 subjects, but for say the VWFA to TPJ, have it omit subject 3 and calculate it using only the first 2 subjects? Would this even be statically valid?

Or if I want to look at all three ROI's do I have to omit subject 3 since they are missing one of them?

I kindly thank you in advance,
Benson Stevens

RE: Error at 1st level: incorrect # ROIs

$
0
0
Hi Suzanne, 

Did you ever find a way to resolve this error? I am getting the same error at this exact step. I have tried re-running the denoising step as the prompt suggested, but still receive the same error. I tried to re-ran preprocessing on the participant with their raw data and tried to denoise again but still have the same error. 

If you or any other conn users know how to resolve this issue, please let me know. I would really appreciate it. 


Best,
Nicole

Voxel-to-voxel contrast

$
0
0
Hello all, 

I'm running a intrinsic connectivity contrast but I get this weird maps (attached). Do you have any tips? 

Thanks 

Davide
Viewing all 6914 articles
Browse latest View live