# Different size image registration

**URL:** <https://discourse.slicer.org/t/different-size-image-registration/9304>\
**Category:** Support\
**Tags:** registration, segmentation\
**Created:** [November 26, 2019, 4:41pm UTC](https://discourse.slicer.org/t/different-size-image-registration/9304 "2019-11-26T16:41:45Z")\
**Posts on this page:** 5\
**Page:** 1

<div class="post-metadata">

**Author:** ![teosava](https://avatars.discourse-cdn.com/v4/letter/t/bbce88/32.png) [@teosava](https://discourse.slicer.org/u/teosava)\
**Post date:** [November 26, 2019, 4:41pm UTC](https://discourse.slicer.org/t/different-size-image-registration/9304/1 "2019-11-26T16:41:45Z")

</div>

Hi all,

I have to register two MRIs that represent the same patient (who is always in the same position). Note that one image is larger than the other as it represents a larger portion of the body.

I roto-translated the larger image in order to obtain an “initial manual registration” and everything is fine until this point.

Then, I applied the General Registration (BRAINS) to refine the initial registration. The results is a cropped image: the larger image is cropped to the same size as the smaller one. This is undesired since I would need the extra portion of the body originally provided by my larger image.

Suggestions?  
Thank you!

---

<div class="post-metadata">

**Author:** ![pieper](https://sea2.discourse-cdn.com/flex002/user_avatar/discourse.slicer.org/pieper/32/8_2.png) [@pieper](https://discourse.slicer.org/u/pieper)\
**Post date:** [November 26, 2019, 4:56pm UTC](https://discourse.slicer.org/t/different-size-image-registration/9304/2 "2019-11-26T16:56:42Z")

</div>

Can you just swap which one is fixed and moving?

Or you can use brains to calculate the transform and use it instead of the resampled image.

Hope that helps,  
-Steve

---

<div class="post-metadata">

**Author:** ![lassoan](https://sea2.discourse-cdn.com/flex002/user_avatar/discourse.slicer.org/lassoan/32/13_2.png) [@lassoan](https://discourse.slicer.org/u/lassoan)\
**Post date:** [November 26, 2019, 6:05pm UTC](https://discourse.slicer.org/t/different-size-image-registration/9304/3 "2019-11-26T18:05:57Z")

</div>

I’ve experienced many times that otherwise very stable registration methods (such as Elastix with default bspline parameter set) failed to converge if one image covered a significantly larger portion of the patient than the other. I would recommend to crop the images to cover approximately the same anatomical region. You can still apply the computed transformation to the entire original image.

---

<div class="post-metadata">

**Author:** ![teosava](https://avatars.discourse-cdn.com/v4/letter/t/bbce88/32.png) [@teosava](https://discourse.slicer.org/u/teosava)\
**Post date:** [November 27, 2019, 9:17am UTC](https://discourse.slicer.org/t/different-size-image-registration/9304/4 "2019-11-27T09:17:30Z")

</div>

Thank you very much for your replies.

How can I visualize the computed transform? I cannot find it after applying the registration (BRAIN). I expect to find a 4x4 matrix.

Moreover, will this transformation matrix refer also to the initial roto-translation (“initial manual alignment” mentioned above)? In other words, shall I apply the transformation matrix to the original volume or to the volume which I obtained after the “initial manual alignment”? If it is the second case, how does the matrix change?

---

<div class="post-metadata">

**Author:** ![lassoan](https://sea2.discourse-cdn.com/flex002/user_avatar/discourse.slicer.org/lassoan/32/13_2.png) [@lassoan](https://discourse.slicer.org/u/lassoan)\
**Post date:** [November 27, 2019, 2:00pm UTC](https://discourse.slicer.org/t/different-size-image-registration/9304/5 "2019-11-27T14:00:19Z")

</div>

> [@teosava](#):
>
> How can I visualize the computed transform?

You can see all transforms in Transforms module. If you computed a composite transform (e.g., linear+bspline) then you need to open “Information” section. To edit linear component of a composite transform, you need to split it first then choose the linear transform node.

 ![image](https://us1.discourse-cdn.com/flex002/uploads/slicer/original/3X/c/0/c0ddf01fcf9758b9a83cf1133f102376b3b0762d.png)