Quantcast
Channel: NITRC CONN : functional connectivity toolbox Forum: help
Viewing all 6884 articles
Browse latest View live
↧

unmatched number of files after preprocessing

November 4, 2021, 5:20 am
$
0
0
Hi all!

My data has 3 sessions, and I found that after preprocessing (using default pipeline), the number of output files are not the same for 3 folders. Say, for each subject folder in session1, totally 31 functional files were generated. But for session 2 and 3, the number was 13. I wonder if this is correct and the reason behind it.

So I wanted to know whether the above case is a result of any internal and/or unknown error ( as I am very new to neuroimaging processing), I re-run the data and things remain unchanged. Also, for the 3rd time, I imported only 2 subjects with 3 sessions instead (originally there were 6 subjects, 3 sessions), and found that output number of processed anatomical images decreased to 18 files (for 6 subjects, the number was 21), though the number of functional data remain the same(session1=31, session2&3=13).

Many thanks in advance for your help!

Sheungling
↧
Search

RE: new ROIs from file

November 4, 2021, 7:58 am
$
0
0
[color=#000000]Dear Sven,[/color]

[color=#000000]To add new ROIs in the [i]Setup.ROIs[/i] tab click on the empty space in the 'ROIs' list (or click on the '+/new' button), then give your new ROI a name, select your NIFTI files in the list towards the right, select the subjects that this ROI files should be associated with in the 'subjects' list, and last click on 'import' to load that ROI. [/color]

[color=#000000]If your dataset is already processed/analyzed, then click on 'Done' sequentially on each of the Setup and Denoising steps, selecting each time the option that reads 'do not overwrite (skip already processed subjects/ROIs)' so that only the new ROIs information is being updated, and that's it, those new ROIs will be present now in the 'ROIs/seeds' list to be used when defining any new first-level analysis or editing an existing one (e.g. to add these new ROIs to an existing analysis)[/color]

[color=#000000]Hope this helps[/color]
[color=#000000]Alfonso[/color]

[color=#000000] 
[/color][i]Originally posted by Sven HALLER:[/i][quote]Dear all

I am relatively new to CONN - but I like it :-)

I did the standard processing of an fMRI dataset, which worked fine

now, I would like to do a ROI analysis of specific (deep brain structure) ROIs that are not in the included atlas
I have the ROI file as .nifty file

How do I integrate new ROIs into the CONN toolbox ?

Thanks
Sven[/quote]
↧
↧

RE: Second Level Analysis' Output Files

November 4, 2021, 8:14 am
$
0
0
[color=#000000]Hi Mor,[/color]

[color=#000000]The only difference is in the context of repeated measure analyses (when you are evaluating multiple conditions and your 'between-conditions' contrast is a matrix, instead of simply a vector). In those cases, SPM approach uses univariate statistics combined with reML covariance estimation to deal with repeated measures, while CONN approach uses multivariate statistics to deal with repeated measures, so they will produce (slightly) different statistics (see https://www.nitrc.org/forum/message.php?msg_id=25437 for a few more details)[/color]

[color=#000000]Hope this helps[/color]
[color=#000000]Alfonso[/color]
[i]Originally posted by mor_fine:[/i][quote]Hi Alfonso,
Thank you for your answer.

I have a follow up question; 
What is the difference between the files "spmT_": T-statistics map (univariate analysis, generated by SPM), and "spmF_mv": T/F-statistic map (multivariate analyses, generated by CONN)?

Regards,
Mor[/quote]
↧

RE: REX results

November 4, 2021, 8:23 am
$
0
0
[color=#000000]Dear Pedro,[/color]

[color=#000000]In general contrasts are always estimated 'at the zero-level' of all of the control variables so centering allows you to choose that 'zero-level' to be something meaningful/representative (the average values of those control variables in your sample). In practice, in many common scenarios centering or not centering your control covariates produces exactly the same analysis results (e.g in ANCOVA-type analyses, in multiple-regression analyses like in your example, etc.) but I still recommend centering as a "good practice" measure in general for those few cases when it does matter (e.g. when looking at ANOVA simple main effects, or main effects in the presence of interactions, etc.)[/color]

[color=#000000]Hope this helps[/color]
[color=#000000]Alfonso[/color]
[i]Originally posted by Pedro Henrique Rodrigues Silva:[/i][quote]Dear Alfonso,

Thank you very much for the amazing response.

I was checking the forum and I read the following: " when using these as control covariates it is recommended to center them" (https://www.nitrc.org/forum/message.php?msg_id=33882). In my multiple regression model with Age (continuous variable), Sex (categorical variable), and Education (ordinal variable with seven levels --> each level added as one column), should I centering all these variables, including the cognitive score?


Thank you very much[/quote]
↧

RE: REX data issue

November 4, 2021, 8:32 am
$
0
0
[color=#000000]Dear Haixia[/color]

[color=#000000]The effect-sizes in those plots represent the 'adjusted means' of the connectivity strength levels for each condition. Those 'adjusted' means are computed as the regressor coefficients of your second-level model, and they represented the estimated average connectivity strength at the zero-level of your control covariates. If the covariates are centered those adjusted means (in the REX plots) should be identical to the standard means (in the violin plots), so it is strange that they differ. If you send me the REX.mat file (or the SPM.mat file) in the second-level results folder I will be happy to take a quick look to the design there to see if I can spot what the issue driving those differences may be here. [/color]

[color=#000000]Best[/color]
[color=#000000]Alfonso[/color]
[i]Originally posted by haixia zheng:[/i][quote][color=#000000]Dear Alfonso,[/color]
[color=#000000]It was a great pleasure to see you via zoom! [/color]
[color=#000000]You were right about the covariates. For some reason, the value of 'sex' was 0 for all subjects. I have now centered age and BMI, also entered corrected value for 'sex'. Now the bar plot generated by REX tool and the violin plot in R looks more alike. However, it is still not quit the same (Please see the attached jpg). The effect size in the plot from REX looks much bigger than the violin plot. I wonder how the value of y axial ('effect size') in the REX bar plot was calculated? [/color]
[color=#000000]Thank you so much for your help![/color]
[color=#000000]Haixia [/color]
[i]Originally posted by Alfonso Nieto-Castanon:[/i][quote][color=#000000]Dear Haixia,[/color]

[color=#000000]Sorry for the late reply, there are a couple of potential issues with the original second-level analysis, could you please double-check the following:[/color]

1) the lack of observable effects in the extracted signal could indicate a problem with your age/sex/BMI covariates, when using these as control covariates it is recommended to center them (subtract the average) so that the "zero-level" of those covariates is meaningful. Could you please double-check if that is the case (are your age/sex/BMI covariates centered)?

and 2) the big red 'warning!' label in your second-level tab example image also potentially indicates a problem with your covariates (e.g. perhaps the AllSubjects term contains some 0 values? or perhaps some of your control covariates contain all 1 values?), could you please check your second-level covariates to see if that is the case? (in case of doubt please send me your conn_*.mat file I will quickly check both of these things)

[color=#000000]Thanks[/color]
[color=#000000]Alfonso[/color]

[i]Originally posted by haixia zheng:[/i][quote]Dear Alfonso,
I used 'Plot effects' button and enable REX gui to extract the data. I have tried 'import data' button too, that gave me same data as the ones from REX gui. Placebo/200/600 are three different within-subject conditions. Because this was a cross-over design, 20 subjects visited 3 times for different doses. Please see the attached PDF file which shows the screenshots for all the settings.
[color=#000000]Thank you very much for your help.[/color]
[color=#000000]Haixia Zheng[/color]
[i]Originally posted by Alfonso Nieto-Castanon:[/i][quote][color=#000000]Dear Haixia Zheng[/color]

[color=#000000]Could you please provide a few more details on how exactly the data was extracted using REX (e.g. did you use the 'import data' button and then exported the resulting covariates to a file, did you use the 'export mask' button and then used rex manually to extract the original data? ...) and also abut your second-level analyses (e.g. are the placebo/200/600 three different subject groups or are they three different within-subject conditions?)[/color]

[color=#000000]Thanks[/color]
[color=#000000]Alfonso[/color]
[i]Originally posted by haixia zheng:[/i][quote]Dear Alfonso and Community members,
I did some seed-to-voxel analyses using CONN and identified a significant cluster. CONN REX showed a nice bar plot. But when I extracted the data from CONN REX and plotted it in R, it looks very different from the plot gave by CONN REX (please see the attached). I wonder how the plot given by CONN REX was generated, and why it looks so different from the R plot using the data extracted from CONN REX. 
Also, I did the statistical test in R using these extracted data, there was no significant effect at all. Did I do anything wrong? 
Any input or guidance is much appreciated!
Thank you[img]https://www.nitrc.org/Users/hzheng/Documents/IBU/Manuscript_VIA/REXdata.jpg[/img][/quote][/quote][/quote][/quote][/quote]
↧
↧

RE: Error using load: during pre processing but referring to a Second Level (SBC) analysis

November 4, 2021, 8:51 am
$
0
0
[color=#000000]Dear Renzo,[/color]

[color=#000000]The error seems to indicate that CONN is still trying to merge the results from the original first-level analysis that were run in your cluster or HPC environment back into your project. First save your CONN project and load it again to see if you still see a 'pending jobs' bright label in the top bar or you get some warning message about pending jobs finished that need to be merged. If you still see any of those, then go to the menu "HPC options. Job history" and delete any last jobs that do not appear as "finished" and then try saving/loading again. Once you see none of those warning/labels you should be fine to proceed processing your data[/color]

[color=#000000]Hope this helps[/color]
[color=#000000]Alfonso[/color]
[i]Originally posted by Renzo Torrecuso:[/i][quote]Dear CONN experts,
At a given moment after performing several Second Level analysis(LC, RC, GC etc) I got this message when I tried SBC.
Then I reloaded all over again my 120 subjects and still got this problem.
Could anybody shed some light please?

All best
Renzo Torrecuso



ERROR DESCRIPTION:

Error using load
Unable to read file '/data/pt_02419/CONN projects/LZ_snr/results/firstlevel/SBC_01/resultsROI_Subject120_Condition001.mat'. No such file or directory.
Error in conn_process (line 4381)
t=load(filename,'Z','regressors','names','names2','xyz','SE','DOF');
Error in conn_projectmanager (line 205)
conn_process(psteps{n}{:});
Error in conn (line 885)
conn_projectmanager('updateproject',fromgui);
Error in conn_jobmanager>conn_jobmanager_gui/conn_jobmanager_update (line 1438)
conn('load',filename);
Error in conn_jobmanager>conn_jobmanager_gui/conn_jobmanager_update (line 1351)
conn_jobmanager_update('finish');
Error in conn_jobmanager>@(varargin)conn_jobmanager_update('refresh') (line 1261)
handles.timer=timer('name','jobmanager','startdelay',1,'period',10,'executionmode','fixedspacing','taskstoexecute',inf,'busymode','drop','timerfcn',@(varargin)conn_jobmanager_update('refresh'));
Error in timer/timercb (line 30)
feval(val{1}, obj, eventStruct, val{2:end});
Error in timercb (line 13)
timercb(t, varargin{2:end});
Error in conn_jobmanager>conn_jobmanager_gui (line 1272)
waitfor(handles.hfig);
Error in conn_jobmanager (line 467)
info=conn_jobmanager_gui(info,files,filedates,varargin{:});
Error in conn_setup_preproc (line 888)
conn_jobmanager(info);
Error in conn (line 1143)
ok=conn_setup_preproc('',varargin{2:end});
Error in conn_menumanager (line 121)
feval(CONN_MM.MENU{n0}.callback{n1}{1},CONN_MM.MENU{n0}.callback{n1}{2:end});
CONN20.b
SPM12 + DEM FieldMap MEEGtools
Matlab v.2017b
project: CONN20.b
storage: 4571.3Gb available
spm @ /data/hu_torrecuso/Documents/MATLAB/spm12
conn @ /data/hu_torrecuso/Documents/MATLAB/conn[/quote]
↧

RE: Different voxel dimension and FOV

November 4, 2021, 8:53 am
$
0
0
[color=#000000]Dear Ramesh,[/color]

[color=#000000]Not sure about VBM but for CONN resting state analyses differences in voxel-size or FOV across subjects are perfectly fine (typically during preprocessing your data will be resampled to a common resolution)[/color]

[color=#000000]Best[/color]
[color=#000000]Alfonso[/color]
[i]Originally posted by Ramesh Babu MG:[/i][quote]Dear Experts,

In my project work, all participants structrual MRI scans have a voxel dimensions of 1x1x1 with FOV 256x256x175, but one scane dimension is 0.8x0.8x2 with FOV 288x288x175. Functional MRI demensions for all the scans is same.
Can I include ths scan with other scan or should I exclude it for resting state analysis and also for VBM analysis?[/quote]
↧

Surface group analysis error

November 4, 2021, 11:25 am
$
0
0
Hi,

I completed 1st-level surface-based analysis (all those results look sensible) and attempting to run Second-level (Group) Analysis. Whenever I try to run Seed-to-Voxel analysis, it takes several minutes to compute, with the message "Updating non-parametric statistics. Please wait."  Volume-based analyses take a few seconds compute, but perhaps surface based analyses need more time in general.

Now to the error: After several minutes the analyses finishes, a sensible group result is displayed on glass brain and can be also explored using the Display&Print options, but I also receive the following error:

Error using reshape
Product of known dimensions, 5124, not divisible into total number of elements, 6972.

Error in conn(line 10951)

t1=reshape(S1,size(CONN_gui.refs.surf.defaultreduced(1).vetrices,1),2,1,[]);

(see attachment for whole message)

It appears that the results are not saved on disk in a readable form - I cannot open any of the results nii files, and I get no visualization of the results in the preview window for ROIs (Seeds) that were already computed and instead get the error again.

My ROIs are fsaverage parcels plus hippocampus (anterior, posterior, whole). One thought I had was that it had difficulty with the mixed surface and subject-space ROIs, but the connectivity of anterior and posterior hippocampus to the whole cortex is of main interest to me (I cannot drop the hippocampus ROI) and I can see the results actually computed and displayed right after a seed-based analysis is run, they just refuse to get saved.

Thoughts?
↧
Search

Failed to load file (line 837)

November 5, 2021, 11:13 am
$
0
0
Hello,

I have received the following erro when attempting to open my CONN project. I previously opened it without a problem, grabbed some data, and then closed it normally, so I am unsure what may have caused this error.

Error using conn (line 837)
Failed to load file /project2/skeedy/BSNIP2_2020/BSNIP2_2020_conn - Copy.mat.

Error in conn (line 4427)
conn('load',filename,true);
Error in conn_menumanager (line 120)

Thank you,
Mikey
↧
↧

have any condition associated with data from session 2

November 7, 2021, 1:25 am
$
0
0
Dear community members:
when i have completed the preprocessing ,click the "done" button,and a error message present:
"Subject 1 does not have any condition associated with data from session 2.
Subject 1 does not have any condition associated with data from session 3.
...."
I have add 51 subjects and have 4 session.who can help me solve this problem.Thank you wery much
↧

RE: new ROIs from file

November 7, 2021, 1:35 am
$
0
0
when i have completed preprocessing and clicked  the 'done' button , an error message present:
"Subject 1 does not have any condition associated with data from session 2."
i don't know what this. could you please help me to solve this?
i have 51 subjects and have 4 sessions per subject. 
thank you for your attention
↧

Conn paired image import

November 7, 2021, 5:28 pm
$
0
0
Hi everyone,
I have several subject preoperative and postoperative rsfmri. I can import these two different functional image in the different "section" in Conn. But I cannot figure out how to import different structual image (T1) for different sections. Because I draw individual T1-masked for each bold series, I cannot use the same T1 structual for all sections.

Does anyone know how to import multiple T1-structual image in Conn? Thank you very much!
↧

Cross-team replication

November 8, 2021, 6:21 am
$
0
0
Dear CONN-experts,

we are currently performing a cross-site replication of connectivity results within a large clinical study, which was conducted at two different university medical centres. We started by analysing connectivity at site A. The goal is now to independently replicate those results within the patient sample of site B. To first test, whether our analysis protocols are fully harmonized, we have tried to replicate the original results from site A by the team of site B using the data of site A. We expected the results to be 100% identical as we had a very detailed analysis protocol from site A and used the same conn version (18a) at both sites.
The results of the cross-team replication are indeed very similar to those from site A, but not 100% identical. Peak voxels are identical, but T- and p-values vary slightly. The same is true for the cluster extent. In one analysis this even leads to two clusters to be non-significant at site B, even though they were significant at site A. We are now wondering what may have caused those small differences? Do you think such differences may have occured due to different MATLAB versions (R2012b at site A and R2019ba at site B)? Are there any estimation or rounding procedures, which may lead to such slightly differing results? Is there an opportunity to obtain approximate values for those estimations to compare them across sites?

Thank you very much in advance!

Best
Fabian
↧
↧

CONN minimum cluster sizes

November 8, 2021, 10:01 am
$
0
0
Hello there,

I was asked by reviewers to provide the minimum number of contiguous voxels required in my resting-state functional connectivity using CONN, but I could find this information in CONN, it seems like only when there was significant effect, the minimum cluster size would be reported for the identified cluster with the smallest size among all the identified clusters in that resulting map. I am wondering where could I find the minimum cluster-size in CONN for my current data analysis?

Any help would be appreciated!!!!

Best wishes,
Meichao.
↧

edges counted twice in gPPI

November 8, 2021, 5:13 pm
$
0
0
Hello,

We are running an ROI-ROI, gPPI analysis and have noticed that each edge/connection is being counted twice, and we hoped to get some advice on this. When doing the gPPI analysis, both terms' edges (i.e. both node A time course * task predicts node B time course and B * task predicts A time course) go into the group-level analysis; so, for example, when looking at 27 ROIs, the result should be 351 connectors, but instead we get double that (so 702). This really limits the ability to run this kind of analysis at a network level.

Is there any way to resolve this issue, or perhaps integrate a number of potential solutions (eg, add options to include the mean of the edges as one edge, or the max)? Given that integrating task-based analyses and network-level analyses could really be an important way forward in fMRI, we think that offering ways to address this limitation could be really useful for many groups.

Thank you for your help!
Laura
↧
Search

CONN workshop, Nov 22 - Dec 20

November 9, 2021, 5:27 am
$
0
0
Hi everyone,

Just a heads up to let you know that the next online [url=https://web.conn-toolbox.org/workshops]CONN workshop [/url]will be starting in a couple of weeks (on Monday November 22th) and there is still time to subscribe if you are interested. The workshop will take place over five consecutive Mondays from 9am to 5pm EST (UTC−05:00), starting on Nov 22th and ending on Dec 20th 2021

As always, the workshop offers intensive hands-on and highly interactive courses covering all aspects of functional connectivity analyses in CONN, and both beginners and advanced users are welcome. Classes will take place over Zoom on Mondays for five consecutive weeks, and there will be a bit of homework between consecutive weeks and some office hours available to help with that.

For more details and registration see [url=https://education.martinos.org/home/using-the-conn-toolbox-for-functional-connectivity-analysis/]https://education.martinos.org/home/using-the-conn-toolbox-for-functional-connectivity-analysis/[/url]

Best
Alfonso
↧

RE: Average within / between network connectivity

November 9, 2021, 7:10 am
$
0
0
Hi Olivier,

I got the same message and then realized that I'd set up my second level as a Seed-to-Voxel analysis. Once I switched to ROI-to-ROI, it ran fine. I hope this helps!
Karli
↧
↧

Strange segmentation results

November 9, 2021, 3:20 pm
$
0
0
Dear Conn experts,

I have found some strange segmentation results, and would like some advice on how to best proceed.

(i) in a subject the CFS voxels are only 110 voxels, and after erosion there was none, which lead to an error during denoising. I solved it by changing the erosion settings so that the binary threshold is 0 instead of 0.5 (only for that subject), and thus keeping all 110 voxels. That seemed to solve the problem, but is it the best way to solve it?

(ii) for a couple of subjects, it seems like the segmentation misclassified the CFS to contain grey matter, and thus also underestimated the amount of grey matter (the white matter looked normal). Looking at the original T1, it seems fine, but overall a bit brighter than other T1s. What can be done to rectify this? I am worried that the compcor in the CFS might catch signal from grey matter and thus have a negative effect on the results. If it is not possible to fix the segmentation, could one simply not use the CFS to create covariates (The white matter seems fins and might be enough)? What detrimental effects would a suboptimal grey matter segmentation have on the results?

I have attached the CFS (left image) and grey matter (right image) of a subject.

Any advice would be much appreciated.

Best,
Fredrik
↧

RE: grey matter denoising problem

November 9, 2021, 11:03 pm
$
0
0
[color=#000000]Hi Alfonso,[/color]

[color=#000000]Can you please explain more on what "low confidence on the tissue-mask segmentation" might mean and is there a way to fix this problem? Thank you![/color]

[color=#000000]Best,[/color]
[color=#000000]Yasemin[/color]

[i]Originally posted by Alfonso Nieto-Castanon:[/i][quote][color=#000000]Hi Yasemin,[/color]

[color=#000000]I believe that indicates a problem in the normalization/segmentation step for those subjects, as, if I am interpreting correctly, what those images are displaying is the gray-matter masks generated during normalization, and the 'disappearing' subject as well as at least one other subject with relatively low values in those images is indicating that the posterior probability maps values are relatively low, indicating low confidence on the tissue-mask segmentation[/color]

[color=#000000]Hope this helps[/color]
[color=#000000]Alfonso[/color]

[i]Originally posted by Candan Yasemin Yazicioglu:[/i][quote]Hi,

One of my subjects do not show an image after denoising even though it seems perfectly fine before the denosing step (looking at structural data etc). Can you help me with this issue? Attached you can see the subject with no image. I believe it is a problem with denoising so is this step adjustable like preprocessing?

Thank you!

Best,
Yasemin[/quote][/quote]
↧

CONN normalisation problem (default pipeline)

November 10, 2021, 12:43 am
$
0
0
Hi Alfonso,

I am pre-processing some resting-state data with the default CONN pre-processing pipeline and experiencing some problems with the normalisation step. The pre-processed output (structural and functional) showing misalignment with the MNI template, with the MNI template mostly missing some frontoparietal brain areas. Non-linear warping seems impossible in this dataset since we do not have fieldmaps.

The attached file shows an example of this based on a single participant, but a similar pattern exists in most participants in this dataset. Using a brain-extracted version of the T1 image (extracted in FreeSurfer) seems to slightly improve the structural output, however, this did not improve the functional output. The attached document shows some examples of attempts to resolve this problem (e.g., by pre-processing in SPM), which have not been successful.

Do you have any recommendations on how to resolve this problem?
Thanks for your help.
Edith
↧
Viewing all 6884 articles
Browse latest View live
Search
<script src="https://jsc.adskeeper.com/r/s/rssing.com.1596347.js" async> </script>