Volume Miscalculation

Hello,

I have been using the Segment Statistics Module to try and compute the volumes of segmentations. However, the values do not seem to be correct. For example, the volume of the right ventricle is shown to be greater than the volume of the left ventricle. Also the volume is much lower than what we found to be the stroke volume.

Is there any reason that my readings are coming out this way?

Thanks!

Volume computation is extensively tested and should be correct.

After you finished defining your seeds, you need to click Apply in “Grow from seeds” effect. Without that, you just measure volume of your seeds.

If this does not help then send the segmentation that you have problem with (or st least a few screenshots).

Hi,

Thank you for your response. I had used the Grow from Seeds Effect. We found the volume of blood pumped by the left ventricle per beat to be 9.7 ± 2.1 mm3 with some previous research…yet the left ventricle does not seem to have near that much volume itself.

image

image

Is this human data? Those values are too small. Can you confirm the spacing of your original dataset is correct?

These seem to be just the seeds that you used for region growing. Have you tried going to Segment Editor module and click Apply in Grow from seeds effect?

But @muratmaga is right, too. Non-clinical scanners often save their data with incorrect spacing that you need to fix manually. What is the spacing value displayed in Volumes module / Volume information section?

Hi,

Thank you for your response. This is not human data. We are using images of chicken embryo hearts. As for spacing I’m going by the information that the person I work under has provided me with.

Hi,

I did use the grow from seeds effect in the segment editor module. Here is the volume information:
image

image

Thanks

2633 voxels are about 50x50 voxels, which would be too few for a highly detailed surface that you show in the screenshot. So, it is a bit more likely that you compute volume for not the right segmentation node.

Can you share an example segmentation (upload somewhere and post the link here) so that we can have a look?

How do I upload the segmentation in order to share here?

You can upload the file to Dropbox, OneDrive, Google Drive, etc. and post the link here.

https://drive.google.com/file/d/15vlxpXtIvLiOe9_UWYHjOHLpQDRYqa5q/view?usp=sharing

I think one must download after clicking this link. Let me know if this link works.

You have only uploaded the .mrml file, which is an index to the actual data files. Please save the scene as a single-file bundle (.mrb file) by clicking the package icon in the save data window:

Thanks, I hope I uploaded it correctly this time.

@Angel as @lassoan indicated previously, you didn’t finalize the ‘grow from the seed’. You need to first initialize using the seeds and review the expected segmentation with those seeds. If you are satisfied with that you then need to hit apply. Here is a result from your share scene where I simply hit initialize followed by apply. I think these numbers make much more sense:

Hi,

Thank you so much for taking a look! These numbers are definitely closer to my predictions. Is there any reason as to why the left ventricle is still less in volume as compared to the right ventricle?

Thanks!

I think your segmentation can use a bit more work and you probably want to add some more detail, before actually looking at the numbers.

Yes, it looks like the segmentation does not cover the entire ventricle (it looks like up to about 5% of the volume may be missed in the segmentation). You can adjust the “Editable intensity range” so that includes a bit more of the dark regions (set the lower threshold value a bit lower) and use Margin effect to grow it a bit into these regions.

Thank you so much for your patience! I have one last question: why does the volume information change after I save to files?

Unsaved:
image

Saved:
image

Volume spacing should not change when you save the scene and reload it. What file format are you saving into?

I tried saving it as both .mrml file and .mrb file and it changed both times.