[Mristudio-users] Coregistration using AIR

[susumu]

Hi Jun,

 

Please see my response below.

 

1). The instruction stated that AIR in DTI Studio is for motion correction
and eddy current correction for multiple DTI aquisitions.  So if I have DTI
sequences already averaged by the scanner, can I use the AIR to correct for
motion and eddy current?

>I recommend not to average scans inside the scanner. Eddy current should be
correctable even if the images are already averaged. Motion correction is a
tricky question. It depends on when the second set is acquired and when the
motion happens. In the worst case, second scan take place after the first
complete DTI is finished so that there is ½ of scanning time lapsed between
the averaging (typically several minutes). If the motion happens between the
first and second scans, then motion correction may not work well. 

2). After the AIR process in DTI Studio, the saved matrix has corrected
gradient table.  If I proceed to calculate tensor, etc., is the modified
gradient table or the original one used in the calculation?  If the
corrected table is not used, should I calculate tensor by closing the DTI
Studio, loading all images, and use the modified gradient table for tensor
calculation?

>Good question. I need Hangyi to answer this. My guess is the corrected
table is applied by default if you choose this option. Please remember,
unless you uncheck the “create new image after registration”, AIR simply
creates corrected images. The subsequent calculation still uses the
uncorrected original data. In this case, you do have to save the corrected
images & table and reload them to use them. If you unckeck this button, then
the original images are overwritten by the AIRed images. In this case, the
subsequent tensor calculation will be based on the AIRed images. Then the
gradient table should be automatically corrected (although I have to confirm
this point with Hangyi).

3). I have been using FSL's eddy_correct program, and I can hardly find any
discription for the program.  Has anyone compared it with the AIR?

> I heard FLIRT in FSL works better than AIR. One of the AIR’s weak points
is its sensitivity to the threshold setting. The “create image 
” functions
in the “outlier rejection” and calculation of subtracted images when there
are more than two repeated measurements provide very good interface to
visually inspect the existence of the mis-registration. At this point, I
recommend you to confirm the results of AIR using these functions and the
the inspection window (“Original-ADC-Mean-STD” button).

Thank you for your wonderful tools and timely support!

Jun Yi Wang

 

 

----- Original Message ----
From: susumu <susumu at mri.jhu.edu>
To: "DTI Studio, ROI Editor, Landmarker Questions/Support"
<mristudio-users at mristudio.org>
Sent: Friday, July 25, 2008 11:09:48 AM
Subject: Re: [Mristudio-users] Coregistration using AIR

Dear Dr. Takaya,

 

Do you want to coregister, combine two DTI scans and calculated a tensor? If
that's the case, you can treat the two scans as two signal averaging. Load
both of them, coregister the first scan to the second scan using the AIR in
DtiStudio, and do tensor calculation.

 

If you want to get separate tensor fields from the two datasets and then
coregister for ROI drawing, I recommend you to use Landmarker. You first
have to calculate all DTI-based images you need (tensor, FA, ADC, b0, etc).
You can then bring b0 images from the two datasets into Landmarker and
perform AIR. You can save the transformation matrix and apply it to other
images. Alternatively, if you load all data at once, AIR is automatically
applied to all loaded data (of course you have to specify which image you
want to use to drive AIR).

 

One important function of Landmarker is that it can apply the transformation
matrix to tensor files too (it does tensor reorientations if rotation
components are involved).

 

Susumu

 

  _____  

From: mristudio-users-bounces at mristudio.org
[mailto:mristudio-users-bounces at mristudio.org] On Behalf Of Shigetoshi
TAKAYA
Sent: Monday, July 14, 2008 1:48 AM
To: Mristudio-users at mristudio.org
Subject: [Mristudio-users] Coregistration using AIR

 

Hello

 

I have two scans of DWI with time interval for each patient (1st scan and
2nd scan). I would like to coregister the images in the 2nd scan onto those
in the 1st scan in each subject using Automatic Image Registration (AIR)
algorithm implemented in DTIStudio, for the purpose of drawing ROIs in the
same brain regions in both scans.

 

First, I did eddy current correction and motion correction using affine
transformation in each scan (1st scan and 2nd scan).

Then, I coregistered the B0 image in the 2nd scan onto that in the 1st scan.
And I would like to adjust the same transformation matrix to the rest of the
images in the 2nd scan.

 

I know that the transformation matrix can be saved when conducting AIR. Can
we use this file to transform the rest of the images in the 2nd scan?

 

Many thanks,

 

Shigetoshi Takaya

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.mristudio.org/pipermail/mristudio-users/attachments/20080909/c15875a4/attachment-0001.html