Jump to content
Sign in to follow this  
MissingLink

CT scan resolution?

Recommended Posts

I have a CT scan that I would like to print.  While looking at the data, I found that the scan isn't uniformly sampled in all directions.
 

The DICOM image axes - call it X-Y - are high resolution - about 800x1000. The number of slices in Z for each is a little over 100, always giving a slice thickness of 3mm (compared to 0.5mm for the other axes).

 

Anyone know why the sampling isn't uniform?  The non-uniformity results in stair-stepping and missing detail (see pic).

 

I suspect the data was decimated for one of a few reasons: (1) to take less storage overall, (2) to make it easier for the Dr to review (100 images vs 1000),  (3) to make it fit on a CDROM or (4) to reduce the noise in any particular slice.  The full data probably exists somewhere, but I don't know what to ask for.  The radiology dept just gave me a blank stare when I asked for the high-resolution version.

 

FWIW, the machine is a Philips ICT-256.

 

 

stairstep.png

Share this post


Link to post
Share on other sites

You are diving into the deep topic of medical imaging scans for 3D printing. I wrote a blog article on how to get the most out of your scan here. Take a look as it is very helpful.

 

3mm is not bad for a CT scan for 3D printing.  In order to understand why the radiology department created 3 mm slices, you need to understand how CT scans work. Modern CT scanners do not acquire data in slices. Rather, the acquisition is helical. The patient moves in the scanner on the Z axis at a fixed speed, while the x-ray tube spins around that axis during acquisition. As a result, relative to the body, the x-ray tube moves around like a helix, i.e. corkscrew. The raw data from this acquisition is stored in memory in the CT scanner. It is then reconstructed into flat slices that can be of any thickness anywhere from 0.5 mm to 5 mm.

helical-scan.png

 

Thinner slices are not always better however. There is a fixed number of x-ray photons that were acquired during the scan. When the slices are created after the fact, the data that those photons created is spread among the slices. If you have a lot of very thin slices then there are few photons per slice. Just like with a handheld camera when you shoot in low light, having a low number of photons results in a grainy image. The exact same things happen with a CT scanner. Thin slices tend to be very grainy and it can be difficult to detect abnormalities when the image quality is poor. Thicker slices on the other hand are fewer in number and have more photons per slice and thus are less grainy. Think of a nice photograph from a handheld camera on a bright sunny day. There are so many photons to make the image, the image is crystal clear. Thicker slices, while giving a higher quality image, are also thicker, and very small structures are harder to see.

 

Therefore, if you had a CT scan and reconstructed 0.5 mm slices, those images would be much grainier and of lower quality than if you had reconstructed with 5 mm slices. When the scan is being taken the radiologist determines what slice thickness is the best for diagnosing the problem at hand. If you're getting a CT scan of the abdomen for appendicitis, you will get 5 mm slices. If you have a problem with your middle tier and are getting a CT scan of the mastoid, you'll probably get 0.5 mm slices.

 

It should be noted that the raw data from the scan acquisition takes up a lot of memory. While the reconstructed slices are saved in the hospitals radiology system, the raw data from the scan is stored on the physical scanner itself. Typically after a few days that data is purged to make room for new scan data. So, after a few days it is not possible to generate any new slices from the scan, as the raw data has been purged.

 

My guess is that 3 mm slices is as good as you are ever going to be able to get from your scan. To reduce the stairstep artifact, run a smoothing algorithm on your model. This should reduce that appearance.

 

I hope this helps

 

Dr. Mike

 

 

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
Sign in to follow this  

×
×
  • Create New...