Dicom image appears squashed

dicom
measure
Tags: #<Tag:0x00007fb87e1ae940> #<Tag:0x00007fb87e1ae7d8>

#1

Hiya,

I’ve been try to get my DICOM images into imageJ correctly. I’ve managed to get them in via TudorDICOM (with help from this forum!), but now they appear squashed. You can see it when I open them in orthogonal view. I was expecting it to look round (the image is of a bucket of rocks). Does anyone know how to fix this?

When I open the DICOMs using medical imaging software they come out fine (not squashed).

Bonus points if anyone can help me with measuring things in orthogonal view! When I try to drop a line to measure it doesn’t work, it just moves the crosshairs around.

Thanks!

Ox


#2

Hi @oxlennon

Your image has a lateral (X-Y-Axis) resolution of 768x768px and an axial (Z-Axis) resolution of 227 pixel/slices. This means your voxels (volumetric pixel) are not isotropic (1x1x1) but more like (1x1x(768/227) = 1x1x3.38). In the menu “Image” -> “Properties…” (shortcut Crtl + Shift + P) you can change the voxel depth to 3.38 which should resolve your issue.

I don’t know how you can draw a line in the orthogonal view, but if you just open your stack everything works fine.


#3

Hi tibuch, thanks for the help!

I changed the voxels how you said, but although it now says it has a resolution of 768x768, the image is now stretched.

Here’s how it looks in imageJ:

Here’s how it looks in other software (3D slicer):

I don’t understand why it automatically comes up with the right size on other software but not in imageJ. Unfortunately I can’t work out a good workflow for making lots of length measurements in these other software packages, so I think it would be good to use ImageJ.

I want to be able to accurately measure the gaps between the rocks in three dimensions, so it’s important to me that 1) the sizing is accurate and 2) I can easily/quickly be sure I am measuring at the same point in the three directions (hence why I would like to use orthogonal view for measuring as I can place myself with the crosshairs).


#4

I think I might have fixed the problem (would appreciate confirmation if this is an okay approach!)

I used 3D Slicer to read the metadata for the DICOM and under the heading ‘0028,0030 - pixel spacing’ I got the measurement that indicates the spacing between the pixels in mm (in this case it was 0.3255208). I opened my DICOM in imageJ, went to analyse > set scale and set the settings as follows:
distance in pixels > 1
known distance > 0.3255208
Unit of length > mm

Then I went to image > properties and the pixel width, height, and voxel depth were all 0.3255208. I changed the voxel depth to 1 (I think this is because the metadata says ‘0018,0088 spacing between slices’ = 1 mm), resulting in this view when I went to orthogonal view:

That looks correct to me! Now I just need to figure out how to do the measuring…


#5

Which distance exactly? Maybe the shortest distance?

Without knowing what you are looking for, would it help to know the ratio of rock-volume to bucket volume?

Exactly what you had to do! :slight_smile: :+1:


#6

I’m actually looking to sample the gaps (placing a grid of points through the pile to chose where to sample) and get three measurements at each sample point (a length for each of x, y, and z gap size at each sample point). I’m interested in the size and shape of the individual gaps, so I don’t want to lose that kind of data by looking at more general/averaged out measurements (like overall volume). The purpose is for understanding the usefulness of these gaps as retreats for animals.


#7

In this case I would suggest to segment all the space and just compute all available 3D-geometric features, take a look at the results and then decide which are useful. If you could provide an example stack I can try to get you started :slight_smile: