[Mristudio-users] Applying field map to DTI ?

Wed May 18 18:39:36 EDT 2016

Hi Dr. Mori,

Thank you for this informative response. Indeed we are interested in measuring local FA within lesion volume. But those volumes need to be defined on FLAIR. I suppose in theory we could define the ROIs on FLAIR, register both FLAIR and DTI to the common T2 space, then use the LDDMM approach. As you rightly mention, the interpolation alone may introduce more error than we correct. And of course in the stroke setting TIME = BRAIN so we cannot add something like T2 acquisition without a strong rationale.

Thanks again—much appreciated!

Dave

From: mristudio-users [mailto:mristudio-users-bounces at mristudio.org] On Behalf Of Susumu Mori
Sent: Tuesday, May 17, 2016 10:11 AM
To: DTI Studio, ROI Editor, DiffeoMap Questions/Support
Subject: Re: [Mristudio-users] Applying field map to DTI ?

Hi Dave,

There are many tools that allow registration of two images with different contrasts.
However, the cross-contrast image registration has inherent limitations; there is no guarantee that the two different contrasts share exactly the same anatomical boundaries. For example, in an extreme case, the red nuclei can be clearly seen by T2 but very faint by T1. If it is not visible in one of the images, there is no way to register such structures between T1 and T2. The WM/cortex boundaries can be seen both in T1 and T2, but we can't assume that the boundaries are anatomical the same. Also, all MRI has a certain amount of spilling of bright signals, which mean, bright objects become bigger. So, as simple as the ventricle boundaries between T1 and T2 could be different; the ventricle volumes measure by T1 could be 500ml while that from T2 could be 520ml, even from the same subject scanned at the same time.

If two images of interest (in your case b0 and FLAIR) are from the same subject, we can limit the amount of image deformation such that the cross-contrast registration would be based on global alignment. This is a valid approach. However, if you need local (elastic) deformation, cross-contrast registration does not have a unique solution (there is no right answer because images do not contain enough anatomical information for accurate voxel-to-voxel registration).

So, I guess we can say, cross-contrast registration is inherently ill-posed.

Now, the problem is, the B0-susceptibility distortion seen in EPI is very local (elastic). So the distorted and undistorted images need to have the same contrast for accurate distortion correction.

Having said that, I want to ask you why you need to correct the distortion.

If you are interested in voxel intensities such as FA and MD, the FA of structure "A" won't change before and after the distortion correction because it just moves the location of voxels (except for the effect of voxel interpolation during the undistortion).

If you are interested in size measurements, distortion is important but you can't have much confidence if the volumes are correct for severely distorted areas, even after distortion correction anyway.

So, depending on what you want to measure in which location, you may not need distortion correction (or distortion correction simply move the voxels you measure without impact on the results).

Susumu

On Mon, May 16, 2016 at 11:52 AM, Scott, David <David.Scott at bioclinica.com<mailto:David.Scott at bioclinica.com>> wrote:
Hi Dr. Mori and MRIstudio group,

May I follow up on the question of using T2w image for distortion correction? We have an investigator whose protocol does not include a T2 acquisition, nor does he wish to add one. However, the protocol does include T2-FLAIR. Could the FLAIR image be used in the same way?

Many thanks,

Dave

From: mristudio-users-bounces at mristudio.org<mailto:mristudio-users-bounces at mristudio.org> [mailto:mristudio-users-bounces at mristudio.org<mailto:mristudio-users-bounces at mristudio.org>] On Behalf Of Susumu Mori
Sent: Sunday, December 08, 2013 3:57 PM
To: DTI Studio, ROI Editor, DiffeoMap Questions/Support
Subject: Re: [Mristudio-users] Applying field map to DTI ?

Unfortunately, MriStudio software doesn't offer the solution #1 (phase map based) and #3 (inverted distortion).

On Thu, Dec 5, 2013 at 10:11 AM, Dorian P. <alb.net at gmail.com<mailto:alb.net at gmail.com>> wrote:
Thank you Dr. Mori.

I am also looking at time concerns, both in acquisition and postprocessing. LDDMM is a bit long to run for clinical purposes. I used to do registration in Diffeomap alone (AIR linear -> AIR non-linear using white matter map segmented from T1 as template), but this did not resolve a recent problem I had. The subject had a tiny tumor (1cm) near the insula. I could see the arcuate passing near the tumor in FA maps, but there was no way to register this onto MP-RAGE (the arcuate seemed passing through the tumor instead). Thus I need some better solution to have millimetric precision.

How would I go for applying solutions 1 and 3 Dr Mori, that is applying the field map or correcting with inverted gradients (posterior-anterior rather than anterior-posterior).

Thank you for your quick response.

2013/12/5 Susumu Mori <smoriw at gmail.com<mailto:smoriw at gmail.com>>
Dorian,

DTI, which is based on single-shot echo planar imaging, is known to distort due to B0 field inhomogeneity.
We can map this field inhomogeneity using a separately measured field map (thus, a measurement of field is needed in addition to DTI). With this field map, we can theoretically calculate the amount of distortion and undo it.

However, this field map technique has some its own difficulties. From what I have heard, this method is not perfect.
When we judge if the distortion is well corrected, we usually compare the distortion-corrected DTI image, such as b0 image, with conventional anatomical scans, such as T2-weighted images. If the corrected b0 image matches very well with a T2-weighted image, we say, "it is well corrected".

So, an alternative approach is, to nonlinearly (such as LDDMM) register DTI images to a T2-weighted anatomical scan. This approach doesn't require a separate field map data, but do requires a T2-weighted image. This approach is explained in the #8 of Getting Started here (https://www.mristudio.org/wiki/user_manual/diffeomap).

Another method is to acquire two sets of b0 image with different phase-encoding gradients, which generate two b0 images with a completely opposite distortion directions. From these two images, a middle image can be calculated. This approach, however, requires such two b0 data to be acquired.

So, all three approaches require some extra data. Which of the three is the best? Well, all three are equally good and equally imperfect.

On Thu, Dec 5, 2013 at 9:30 AM, Dorian Pustina <albnet at gmail.com<mailto:albnet at gmail.com>> wrote:
Hello everybody at MRIstudio,

I remember reading some paper that said field maps should be applied to DWI data to correct for deformations and bring everything closer to true anatomical space. Anybody knows how to apply this field map and whether it will really correct for the spatial distortions of DTI (thus making it easier to register DTI on MP-RAGE)?

Thank you

Dorian
TJU

_______________________________________________
mristudio-users mailing list
mristudio-users at mristudio.org<mailto:mristudio-users at mristudio.org>
http://lists.mristudio.org/mailman/listinfo/
Unsubscribe, send a blank email to: mristudio-users-unsubscribe at mristudio.org<mailto:mristudio-users-unsubscribe at mristudio.org>

_______________________________________________
mristudio-users mailing list
mristudio-users at mristudio.org<mailto:mristudio-users at mristudio.org>
http://lists.mristudio.org/mailman/listinfo/
Unsubscribe, send a blank email to: mristudio-users-unsubscribe at mristudio.org<mailto:mristudio-users-unsubscribe at mristudio.org>

_______________________________________________
mristudio-users mailing list
mristudio-users at mristudio.org<mailto:mristudio-users at mristudio.org>
http://lists.mristudio.org/mailman/listinfo/
Unsubscribe, send a blank email to: mristudio-users-unsubscribe at mristudio.org<mailto:mristudio-users-unsubscribe at mristudio.org>

________________________________

Confidentiality Notice: This e-mail transmission may contain confidential or legally privileged information that is intended only for the individual or entity named in the e-mail address. If you are not the intended recipient, you are hereby notified that any disclosure, copying, distribution, or reliance upon the contents of this e-mail is strictly prohibited. If you have received this e-mail transmission in error, please reply to the sender and then delete the message from your computer. Thank you.

_______________________________________________
mristudio-users mailing list
mristudio-users at mristudio.org<mailto:mristudio-users at mristudio.org>
http://lists.mristudio.org/mailman/listinfo/
Unsubscribe, send a blank email to: mristudio-users-unsubscribe at mristudio.org<mailto:mristudio-users-unsubscribe at mristudio.org>

________________________________

Confidentiality Notice: This e-mail transmission may contain confidential or legally privileged information that is intended only for the individual or entity named in the e-mail address. If you are not the intended recipient, you are hereby notified that any disclosure, copying, distribution, or reliance upon the contents of this e-mail is strictly prohibited. If you have received this e-mail transmission in error, please reply to the sender and then delete the message from your computer. Thank you.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mristudio.org/pipermail/mristudio-users/attachments/20160518/ef90a653/attachment-0001.html>