The comment section of the new website has Disqus, which allows you to post pictures and video. This gives me a better idea of whatever issue you're having with your data. Disqus also makes it easier for me to moderate the comments and to create threads, which I hope will make the whole question-and-answer process easier.
This is also part of my effort to transition everyone to the new website. I've been thinking about automatically redirecting any URL request for this site over to the new site, so be prepared if that happens sometime in the near future.
Hi Andy. I am a psychiatrist in training who has really become interested in neuroimaging and in particular studying connectivity. Where do you suggest I start? There is so much material out there.
ReplyDeleteHello,
DeleteThat's a good question; most of what I've learned was from reading the documentation from the different software packages and doing some experimenting on my own. If you're interested in functional connectivity, I have a few tutorials uploaded on my Youtube channel about how to do it in AFNI. For DTI analysis of tract statistics (which is related to connectivity) I would look into FSL's TBSS program to get started. If you want to go further, I recommend FreeSurfer's TRACULA program.
I also offer consulting services over Skype for a fee; if you are interested, please send an email to andrew.jahn@yale.edu.
Best,
-Andy
Hi Andy, I happened to came across your youtube video regarding fmri analysis. I am a student and have currently doing a research that involved fmri classification. There are few question that I really need helps on. There a fmri dataset in 4D (event related with different onset). What I am trying to do is to get the 3D slice volume that correspond to the events with activation signal colored on it and attempt to do the classification based on the images. Is it possible? Thank you!
ReplyDeleteE waste recycling centers bangalore, India
ReplyDeleteHi andy
ReplyDeleteI'm trying to motion correct some fmri data but I have a conceptual problem with fsl motion correction part.
it motion corrects each run individually to the middle image of that specific run, and the same operation for other runs.
suppose that I don't have T1 images and I don't want to register my functional images to mni template, my problem is why fsl doesn't have any option to register whole images of all runs to the middle image of the first run?
because spm has this option and I have better result of clustering using the data preprocessed by spm than fsl.
best regard
mahdiyar
Hi andy, I am start to analysis DTI on FSL 5.0 by GUI platform, that have one parameters are index.txt, i don't known how to figure out it, would you help to figure it out? thanks
ReplyDeleteHi Andy, thank you for your blog..
ReplyDeleteI am Francesco Neri, psychologist.
I am performing an fMRI analysis with a block design (task of memory VS control task).
I extracted the individual "swra" volumes.
I have executed a factorial model and for each subject I have separated my two conditions.. everything is OK, the model goes and give me out a decent output.
Now I would like to put my covariates (sex, age, scores in some tests), but I can not do it, I only have the option "regressors" in my model ..
could you help me please?
Thanks for your patience.
Francesco Neri
Hi Andy, thanks for all your posts. This has been really helpful for me.
ReplyDeleteI had a question about the second level or group level fMRI analysis. Do you think it is wrong to use individual subject's spmT** maps instead of con** maps to do group level analysis?
My understanding was that as t maps are just a summary statistics of the con maps, and doing a group level t-test on t-maps or con maps both should result in very similar group maps and is acceptable. Please let me know your thoughts on this.
Hi Tubz,
DeleteIt is recommended to use the con** maps, since they will be normally distributed, and the t-maps won't be. This will more likely make your second-level inferences valid, since the assumption of normality won't be violated.
Best,
-Andy
Hi Andy,
ReplyDeleteI'm trying to run a gPPI analysis for 2 groups of patients (drug and placebo group) who scanned 2 times (pre/post), each time point has 2 runs and 4 conditions. Briefly, it should be a 2(group)-by-2(time point)-by-4(conditions) design.
I preprocessed the pretreatment and posttreatment data with CONN separatly, within which i defined 2 sessions per subject for 2 runs and added the corresponding functional and anatomical images. Everything works well, i got the first level beta map for each patient during each time point and each condition.
Then i realized that i could not preform the second-level analysis (group-by-time interaction, main effect analysis) with CONN, because i processed the 2 time point data separately.
1) I used CONN second level analysis to get contrast maps between condition 2 and 3 (Emotion vs neutral) for each individual at each time point. Then i entered these contrast maps to SPM 12 to perform 2(group)-by-2(time point) factoral analysis to get treatment effect. With this method i did not get any significant results.
2)I then performed a 2(group)-by-2(time point) full factorial design using the beta maps from condition 2 (emotion, the most interested condition) derived from CONN to examine the treatment effect on functional connectivity. Finally, i got significant interaction effect and main effec of group after FWE correction.
I wanted to know if it was correct to preprocess the 2 time point data separately and then perform the second level analysis with SPM like waht i did above? If i was wrong, what should i do?
Best wishes,
Lu
Hi Andy,
ReplyDeleteI'm running a task-based gPPI analysis for 2 groups of patients (drug and placebo group) who scanned two-times (pre/post), each time point has 2 runs and 4 conditions.
Briefly, it should be a 2(group)-by-2(time point)-by-4(conditions) design study.
I preprocessed the pretreatment and posttreatment data with CONN separatly, within which i defined 2 sessions per subject for 2 runs and added the corresponding functional and anatomical images. Everything works well, i got the first level beta map for each patient during each time point and each condition.
Then i realized that i could not preform the second-level analysis (group-by-time interaction, main effect analysis) with CONN, since i processed the 2 time point data separately. I then used SPM 12 to do the second-level analysis:
1) first, i used CONN to get the contrast maps between condition 2 and 3 (emotion and neutral, the most interesed contrast) for each patient at each time point. then, i used those contrast maps to perform a 2(group)-by-2(time point) full factorial to get treatment effect. unfortunately, i did not get any significant interaction or main effect results.
2) i then performed a 2(group)-by-2(time point) full factorial design using the first-level beta maps from condition 2 (emotion, the most interested condition) derived from CONN to examine the treatment effect on functional connectivity. Finally, i got significant interaction effect and main effec of group after FWE correction.
I wanted to know:
if it was correct to preprocess the 2 time point data of gPPI separately and then perform the second level analysis with SPM like what i did above?
i noted that most task-based FC (gppi) was based on contrast of task and contral conditions, if method 2 (just use examin 1 condition) was correct?
If i was wrong, what should i do if i just interested in the treatment effect on functional connectivity of amygdala during emotion condition?
Best wishes,
Lu