mixing parameters up (new segment + Dartel)

Submitted by Ham on Thu, 08/28/2014 - 14:17

Hello again!

 

Firs of all, congratulations! The new tool "dpabi"  looks amazing! Another step forward!

It could be a silly question, but if run an analysis, with New segmentation + Dartel and then I  Normalized to EPI template. Is this a bad processing criteria? 

How it can affect to my analysis? 

Or if I want to normalize to EPI templates  I need to run the old segmentation?

Forums:

YAN Chao-Gan

Thu, 08/28/2014 - 14:39

Hi Ham,

Why do you want to normalize to EPI template if you New Segment + DARTEL done?

Software wise, you can. Paper wise, you may need to rethink.

Best,

Chao-Gan

Ham

Thu, 08/28/2014 - 15:05

That's right, it has no sense. Just wondering to try different parameters. 

So, you recommend to try New segmentation + Dartel and and Normalize by Dartel? 

And old Segment with normalize to epi?


Thank you so much!

YAN Chao-Gan

Thu, 08/28/2014 - 15:10

Normalize to EPI, you don't need T1 image.
If you have T1 image, just go to New Segment + DARTEL. Old Segment only if you have 3 subjects.

Best,

Chao-Gan

Ham

Thu, 08/28/2014 - 15:35

The problem when I perform with New segment + Dartel and then I normalize with Dartel are the black spots. 

I know that I can avoid this doing smoothing after normalization. But, I'm not sure about this step.  So, in this case would you recommend to perform without segmentation and then normalize to the template?

What do you think?

 

best, 

 

YAN Chao-Gan

Thu, 08/28/2014 - 15:50

No, I would still use DARTEL.
I have asked, and Dr. John Ashburner has replied that issue for several times. So I choose to trust him. :)

Best,

Chao-Gan

Ham

Thu, 08/28/2014 - 16:05

     

            In Dr. John Ashburner We Trust

 

Thank you again,

 

Cheers!

hoptman

Thu, 08/28/2014 - 18:49

Hi Chao-Gan,

I had the same issues, if you recall. In my own experience, I had to use EPI normalizations to get good registrations.  But that might be because I had data from two magnets with different scan parameters (thickness, voxel size, etc).

Best,

Matt

YAN Chao-Gan

Thu, 08/28/2014 - 19:00

Hi Matt,
I think those mis-registration issues can be fixed by "bet" in DPARSF V3.0 Advanced Edition.

Best,

Chao-Gan

Ham

Fri, 08/29/2014 - 12:58

Dear Yan Chao-Gan, 

As you correctly pointed out, even with an apparently bad normalization using DARTEL, the results after smoothing looks good and quite similars (i.e; using EPI template).  
 

Could you please tell me if these parameters, which I used to perform with Dartel were properly used?  Please realize, that to obtain better results I need to filter before W and S, and put a FWHM of 8. 

http://s28.postimg.org/vu3zuxbjx/Captura_de_pantalla_2014_08_29_a_la_s_14_50_16.png

What do you think?

YAN Chao-Gan

Fri, 08/29/2014 - 13:30

Hi, I can't see the image, becaused my institute banned postimg.org.

Actually, you can post images here, just click

Best,

Chao-Gan

Ham

Fri, 08/29/2014 - 13:54

 

Thank you!

YAN Chao-Gan

Fri, 08/29/2014 - 13:57

How do you determine your normalization results are better if you put filter before W and S?

Ham

Fri, 08/29/2014 - 14:08

I know it sounds weird, but when I was trying to replicate different hypothesis,  just filtering before normalized & smoothed, I obtained similar aproximations. In the other case, If I put the filter after W/S I can't see nothing interesting, just a few significative clusters (from a list of 34 rois).

 

YAN Chao-Gan

Fri, 08/29/2014 - 14:13

This I don't know well. This is not a QC issue but a research issue, deserve a paper to study it.

For QC issue, I am thinking the normalization effects.

Best,

Chao-Gan

Ham

Fri, 08/29/2014 - 14:23

But do you think it could be a problem to submit a paper with data filtered before the normalization?

Best, 

Juan Gea. 

YAN Chao-Gan

Fri, 08/29/2014 - 14:30

If I am the reviewer, I don't think filter before spacial normalization is an issue.
Actually, DPARSFA, and my previous papers, have done some work "Calculate in original space". The spacial normalization only happens after the derivatives (e.g., FC maps, ALFF maps, ReHo maps) were calculated.

Best,

Chao-Gan

Ham

Fri, 08/29/2014 - 14:38

Pefect! Thank you so much!

 

Cheers!

hoptman

Fri, 08/29/2014 - 13:45

Hi Chao-Gan,

Alas, this did not help in my case (see attached picture).

Best,

Matt

YAN Chao-Gan

Fri, 08/29/2014 - 13:49

Hi Matt,

It's weird! You did "interactive reorienting" and "bet", right?

I don't know the cause, I suspect this is caused by that you still didn't make a contribution at http://rfmri.org/HelpUs.

;) I am teasing, Matt and I are good friends and know each other well.

Best,

Chao-Gan

hoptman

Fri, 08/29/2014 - 14:00




Hi Chao-Gan,

I forgot to do the interactive reorienting (DOH).  Will try again. And will contribute even if you are kidding.

-M

 

From: RFMRI.ORG [mailto:rfmri.org-bounces@rnet.co]
On Behalf Of The R-fMRI Network
Sent: Friday, August 29, 2014 9:49 AM

To: rfmri.org@rnet.co

Subject: Re: [RFMRI] mixing parameters up (new segment + Dartel)

 

[To post a comment, please reply to
rfmri.org@gmail.com ABOVE this line]

 

Commented by YAN Chao-Gan (YAN Chao-Gan)

Hi Matt,

It's weird! You did "interactive reorienting" and "bet", right?

I don't know the cause, I suspect this is caused by that you still didn't make a contribution at
http://rfmri.org/HelpUs.

;) I am teasing, Matt and I are good friends and know each other well.

Best,

Chao-Gan



Online version of this post:
http://rfmri.org/comment/3198#comment-3198

 

IMPORTANT NOTICE: This e-mail is meant only for the use of the intended recipient. It may contain confidential information which is legally privileged or otherwise protected by law. If you received this e-mail in error or from someone who is not authorized
to send it to you, you are strictly prohibited from reviewing, using, disseminating, distributing or copying the e-mail. PLEASE NOTIFY US IMMEDIATELY OF THE ERROR BY RETURN E-MAIL AND DELETE THIS MESSAGE FROM YOUR SYSTEM. Thank you for your cooperation.