Eagle CT data set-how to improve CT resolution_ Preferred CT settings

I am segmenting the skeleton of an eagle that was electrocuted. He has since recovered and is doing well. The middle section of his lungs and spine are much darker and the resolution is not as clear as other areas of the skeleton. I have tried window leveling without much success.

  1. Are there Slicer features I can use to get better resolution of the central portion of his spine? The segmentation will be used for 3d printing.

  2. Are there preferred setting for CT’s? Again for 3D printing. In terms of resolution (voxel size, slice thickness, specific low or high pass filters) and field of view size?

Thank you.

Data sets were provided by Dr Scott Echols, www.avianstudios.com).

I’m not an eagle expert, but is it possible the bones are just less dense there, perhaps to be lighter? I say that because the adjacent anatomy doesn’t look like it’s lower resolution or intensity than muscles elsewhere. Unfortunately when bones are less dense simple thresholding doesn’t work well and you may be reduced to semi-automated approaches. The Grow from seeds tool in the Segment Editor is a good general purpose tool, but I don’t know how it will work for this data.

@muratmaga may have suggestions for working with animal data.

This looks like a medical CT scan, if so you probably not going to get a very good detail on the vertebra. Perhaps oversampling may help a bit? And also if the image is anisotropic, you may want to set the isotropic option in Crop Volume. They may help a bit during segmentation.

Steve, I’ll try grow from seeds again. I’m also questioning if there are optimal setting for a CT with contrast that is to be used for 3d printing. I get such a variety of quality with CT’s from vets. Some at a slice of 0.35-0.50mm can have terrible definition between organs and some are great. I understand typically you want a smaller slice for vascular models and better definition so 0.35-0.50mm is optimal. I’ve reached out to RSNA for preferred CT settings based on the type of model or anatomy. I’ll let you know what I find out. Thanks.

The image is isotropic at 0.35 with a spacing scale of 1.00x. I’ll oversample it at .50x and .25x and try it with grow from seeds to see if I get better definition. Thanks.

I would try using the CropVolume with something like 0.5 (oversampling) but crop out only the vertebral column (if that’s what you will be segmenting). That way you may be able to keep memory consumption low and do the computations faster. At minimum you wouldn’t have to wait too long while testing different approaches/parameters.