Custom HU phantom

Google Groups
1

Insert-loading-pattern-of-the-Gammex-467-phantom-Indices-refer-to-the-insert-names-of.png
Insert-loading-pattern-of-the-Gammex-467-phantom-Indices-refer-to-the-insert-names-of.png649×651 20.8 KB
Hi,

I’m finding Pylinac very useful for analysing Catphan scans. We also have a gammex HU phantom similar to the one shown in the image which has 16 HU inserts. Do you have a recommendation of where to start with adding a custom CT phantom like this? Would treating it as a partial catphan scan work? or better to start from scratch using some of the Pylinac core modules?

Thanks,
Tim

2

Hi Tim,

you could start by upvoting my pylinac user suggestion here:
https://pylinac.featureupvote.com/suggestions/279202/analyze-hu-from-gammex-rmi-phantom

If you intend to write your own code you could probably reuse the catphan code quite easily:
https://pylinac.readthedocs.io/en/latest/cbct.html#customizing-modules

regards
Thomas

3

Funny you bring it up. This has been a request on the RadMachine side from several customers and a generic Gammex algorithm is in the queue =). Since this is an image analysis algorithm it will be open source as normal. Generally speaking, pylinac is moving away from the pass/fail evaluations to simply reporting the values acquired since QAT/RadMachine is meant for the reference and tolerance testing. All this to say the algorithm will probably take the form of a typical catphan class except for reference values. Especially w/ the gammex phantom where you can swap out inserts.

If you have datasets I’ll happily take them and any suggestions are welcome!

4

Hi,

Thank you both for your replies. Thats good news that it’s on the list and will be open source. That method would suit me as we’re using QAtrack+ for the tolerances.

I’ve had a quick go at using the current catphan class with just the 404 module but it was below the minimum number of slices allowed. Is there an easy was to override that? I guess I would need to adjust the section that identifies the catphan as well to avoid it failing. It might be worth me waiting for it to be added at a future version though.

I’ll find a couple of examples scans to send over.

Thanks
Tim

5

Hi James,

Here are a couple of example scans. One is the Gammex phantom and the other is the older RMI version with different insert locations.

Thanks
Tim

Gammex CT2.zip (4.43 MB)

6

RMI CT1.zip (4.55 MB)

7

Hi James,

I just uploaded two CT datasets via the image donation form ( Pylinac image donation form ). One with the official titanium insert and one with a custom tungsten. The titanium has a 10mm titanium rod, the tungsten one is only 5mm in diameter.

regards
Thomas

8

Thank you both! I have to make a Primus phantom algorithm first for a RadMachine client, but plan on working on this after that.