How to adjust the distorted image?

Support
82

Thank you, that makes sense. The reason I am having to resample is I need to make anatomical measurements on the mid-sagittal plane. And with oblique sagittal images, the true MSP (perfectly aligned to the patient coordinates) spans over multiple slices. This makes my annotation task hard. That’s why I figured resampling to [[0,0,-1],[1,0,0],[0,-1,0]] might work and I can make my annotations on one slice. Just wondering - are there any methods I could use in Slicer to help with this?

83

Maybe the ACPC module would work for your purposes.

84

Thank you, that will definitely help. The only issue is that I am working with CT images and as the soft resolution is not very great, marking the AC and PC is going to be a challenge. I will try it out nevertheless.

Best,
Sharada

85

The AC-PC line is optional. You can just place points on the mid-sagittal plane and the ACPC module will compute the transform that you need.

86

Hi Andras,

Thank you for that tip. I did my best at locating AC and PC with the axial volume, and also tried only with the midline points like you suggested. The results match pretty closely, but there are a few differences (Gray is With AC-PC, Cyan is without it). The computed transforms are also slightly different, but I believe it is not a matter of concern for the accuracy of further image processing. What do you think?

image
image1093×837 65.9 KB

I also have a follow-up:
With a sagittal volume, the AC PC is better identifiable. I computed the transform using the AC-PC module for alignment and applied it to the Axial volume instead. Both these volumes were acquired for the same study. With this, the alignment seems off, especially in the coronal view.

image
image955×784 146 KB

I suspect it is because Slicer is not taking into account the Sagittal acquisition transform which is:
image

This volume is obviously an oblique acquisition. I might be wrong with my theory but please let me know if you have any insight here.

Thanks again for all your help and advice with this!

-Sharada

87

If you only specify AC-PC line then the volume is rotated so that the line will be in an axial plane. However, you still need to drop points on the mid-sagittal plane if you want to fix rotation around the AP axis.

88

I see, that makes perfect sense. Sorry if I worded my questions unclearly before, but I wanted to know:

  1. If there’s a difference between just specifying the midline (5 points) versus midline + AC PC line. Visually, there is not much difference - as seen in my earlier screenshot. The computed transforms are different, but very close.
  2. When I load an oblique sagittal image, I don’t see an acquisition transform created. Does Slicer automatically account for this like it handles gantry tilt?

Thanks,
Sharada

89

AC-PC line is only used for rotating the volume so that the AC-PC line becomes “horizontal” (in an axial plane).

An oblique image can be represented in a volume node. Acquisition transform is only created if the image geometry is not a simple rectilinear grid with orthogonal axes (such as a volume acquired with tilted gantry or with non-uniform space between slices).