PyLinac Changes V3.1

[Robert Richardson]

Hi all,

We recently upgraded from 3.0.16 to 3.1 and noticed tests using PyLinac now give different results when re-running old data, has anyone else encountered this?  Is there a location I can look in to see the expected changes (other than the release notes for 3.1)?

The tests which have changed are:

Winston Lutz:  Now showing In-Plane Gantry Rotation compared to Relative Gantry Displacement

Dose Rate Gantry Speed:  Although the analysis looks the same, the magnitude of the average values for the corrected segments has changed (from ~15 to 120)

Fence: Analysis/plotted images look the same but slightly different results

CATPhan: Now fails in V3.1 using the same code we had in 3.0.16

[Randle Taylor]

Hi Robert,

The Pylinac documentation has a change log which you can see here:  https://pylinac.readthedocs.io/en/stable/changelog.html.   As you noted there have been some changes to the analysis and method signatures which may require you to update your calculation procedures.

Randy

--
You received this message because you are subscribed to the Google Groups "QATrack+" group.
To unsubscribe from this group and stop receiving emails from it, send an email to qatrack+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/qatrack/8c5242b3-fdd2-41e0-93d1-b4758d8be0d3n%40googlegroups.com.

[James Kerns]

Hi Robert, 

    Look over the changelog Randy pointed out for details on the pylinac changes. It depends on exactly what version of pylinac you were on before, but:

WL: It's just a name change in the title; same data.

VMAT: This is likely due to the change of normalizing images. This was introduced because some images were saturated and the range of data was too small, causing some localization problems (e.g. all data between 65,100-65,500).

PF: This is likely due to the MLC arrangement enhancement:  Picket Fence module documentation — pylinac 2.4.0 documentation The exact leaf lines are probably slightly different (+/-1 pixel row) since they are now a priori and not empirical.

Catphan: See if this is a QAT calculation error or a pylinac error. If it's a pylinac error then post on the pylinac forum and we'll see what we can do: Pylinac - Google Groups

[Robert Richardson]

Thanks for the responses.  

The WL is giving different results not just a name change, although minor which we can probably review and accept, same with the Fence & DRGS it was just unexpected, I'll take a look at the PyLinac Change log.

The CATPhan is an error "CTP404CP504: Object has no attribute _setup_hu_rois" used for extracting the HU pixel values of the various sensitometry inserts, I'll look over the  PyLinac documentation and post in the other group if needed as i believe it's a PyLinac issue.  

The CATPhan is the only test that's an issue as it no longer works, the others we can just accept their new analysis methods following a simple risk assessment.

Thanks again, and reassuring to know that for most these where expected changes (if I'd read the right documentation anyway!)

Rob