normalisation and registeration?

Submitted by yamur on

Hi

I have a problem about coregisteration and normalization using DPARSF.
Since Coregisteration means overlaying  functional images on their structral image & Normalisation means warping both functional and structral images to fit to a standard template (am I wrong?), why the functional images are normalized by EPI or T1(instead of MNI ?) 
And  why I could not find an coregisteration step in DPARSF?
Any suggestions would be highly appreciated.

All the best,
Yamur

YAN Chao-Gan

Tue, 10/28/2014 - 21:57

Hi Yamur,

Please visit http://rfmri.org/Course to go over all the courses. Especially the Part II of Course V2.0 which will come soon. The basic concepts were covered there.

Best,

Chao-Gan 

Hi Chao-Gan

i watch them all. due to my low IQ that doesnt help!apparently!is it not possible for u to help???is it not a forum for beginners too?

Hi Yamur,
I sense frustration in your post. Be patient. This is a forum for beginners but you know, people can get busy.

You are right about the concept of normalization and coregistration. These two steps ensure that we can draw conclusion about function of brain regions and compare images between subjects.

The two steps are very common in fMRI data analysis. DPASF does these steps by default and does not offer an option to opt-out.

Also, could you please restate your question "why the functional images are normalized by EPI or T1(instead of MNI ?) ". it does not make any sense to me. functional image is often EPI, most of the time they are used interchangeably . T1 is used to capture anatomical images. MNI is a standard space which you can normalize your images (EPI or T1) to.  

Let me add a few words about image type and image space. Your raw functional image (EPI) is in its "original EPI space". If you do coregistration with your anatomical image (T1), you will transform your EPI from its original space to the space of T1, so that you can get EPI in "T1 space". If you do normalization, you will transform or warp your T1 and EPI in to a standard space (MNI or TLRC). if you use DPASF, you will get EPI and T1 in "MNI space".  

Hope these help.

Yang






------
Qingyang Li

Dear Yang

Thank you  so much.I'm so grateful. now all make sense!
 if you use DPASF, you will get EPI and T1 in "MNI space".  I have preprocessed my data with DPARSF but my resulted image doesn't seem to be in MNI space( as far as I know MNI space has a dimension of 91,109,91 and origin voxel of 46,64,37. However, my preprocessed image has a dimension of 61,73 ,61 and origin vocxel of 31,43,25).Thank you in advance.

Best,

Yamur

the dimension of your results depends on 2 things: voxels size and size of bounding box of the final images. If you are not familar with the two concepts, please look them up.

Images in MNI space does not have a definate dimension. The dimension of your results (61,73 ,61) does looks like dimension of a image in some sort of starndard space, and the voxel size of your results must be larger than the voxel size of the MNI space image your were comparing your results with.

For more information about MNI space, you can read:

http://www.nil.wustl.edu/labs/kevin/man/answers/mnispace.html

 

Hope this does not confuse you.

Dear Yang

Thank you so much for your answer. I  would like to ask one more question if you dont mind.
For a given image, how could I determine the image space? Beside my preprocessed image (61,73,61) I have an another image ( dimension:53,63,46  origin voxel: 27,38,18) and I would like to overlay the first image on the second. What should I do to be able to overlay them ?could I do it using DPARSF ?

MAny thanks in advance.

first of all, your second image does not seem to be in a standard space given the dimension of it. but you still can overlay the two images.

I don't know if DPASF can do it. I know AFNI can overlay two images with different dimension, but the two images must be in the same space. I am not aware of any other software that can overlay two images with different dimension. But you can always resample images to a certain dimension.

Yang

Hi,

DPABI Viewer can do that -- as long as the two images are in the same space. Different resolution, different rotation matrces in the header, are not issues with DPABI Viewer. Just make sure they are in the same space.

Best,

Chao-Gan

Dear Yang

What made you say "The dimension of your results (61,73 ,61) does looks like dimension of a image in some sort of starndard space,"
How could you know whether an image is in a standard space or not by only looking at the dimension?

Many thanks in advance.

It was just a guess.

People tend to make their slices in to square when normalize there images into a standard space. you have 61 by 61 slices, so I made the guess. I could be wrong.

Yang

Is there anyone on this email list that can remove me from it. I have nothing to do with this.
On Tue, Oct 28, 2014 at 7:27 PM, Qingyang Li <congel@gmail.com> wrote:
Hi Yamur,

I sense frustration in your post. Be patient. This is a forum for beginners but you know, people can get busy.

You are right about the concept of normalization and coregistration. These two steps ensure that we can draw conclusion about function of brain regions and compare images between subjects.

The two steps are very common in fMRI data analysis. DPASF does these steps by default and does not offer an option to opt-out.

Also, could you please restate your question "why the functional images are normalized by EPI or T1(instead of MNI ?) ". it does not make any sense to me. functional image is often EPI, most of the time they are used interchangeably . T1 is used to capture anatomical images. MNI is a standard space which you can normalize your images (EPI or T1) to.  

Let me add a few words about image type and image space. Your raw functional image (EPI) is in its "original EPI space". If you do coregistration with your anatomical image (T1), you will transform your EPI from its original space to the space of T1, so that you can get EPI in "T1 space". If you do normalization, you will transform or warp your T1 and EPI in to a standard space (MNI or TLRC). if you use DPASF, you will get EPI and T1 in "MNI space".  

Hope these help.

Yang






------
Qingyang Li