Picket fence analysis questions

Hello James:

I’d like to ask two questions about picket fence. The first one is about the result of the analysis of the picket fence module. The result with my image is this:

Picket Fence Results:
100.0% Passed
Median Error: 0.022mm
Mean picket spacing: 0.0mm
Picket offsets from CAX (mm): 35.8 35.8 35.8 35.8 35.8 35.8 35.8
Max Error: 0.093mm on Picket: 6, Leaf: 12

It seems that the mean picket spacing and the picket offsets from CAX are wrong. I send the image attached in case you can have a look. Maybe the problem is caused by the version of the software used to acquire the image (AM Maintenance VARIAN).

The second question is: Wouldn’t it be interesting to know, apart from the max and median error of the peaks that you already calculate, the thicknesses of the FWHM that you obtain? I mean, maybe the peak between two opposing leaves is centered but the gap between them is thicker than the average. In this case there would be a problem with the positioning of these two leaves. I don’t know if it does make sense??

Thank you very much

Roberto Berenguer

SII01257.dcm (385 KB)

Thanks for the notification; I’ve fixed the issue, and uploaded a new bugfix version to pypi (1.5.4). Conda update will come shortly. It wasn’t on your end, just me not testing as fully as I should =)

As for the picket thickness, this is an important question and one I’ve been trying to determine the value of. Anyone who’s created software quickly realizes that you should only add what’s important. Pylinac has evolved from a minimalist sort of standpoint and assumes there are only a few things you need to know about the analysis. A good example of the opposite is RIT, which gives you reams of data that, quite honestly, I don’t see the point of. As a clinical medical physicist what I’m most concerned about is: Are there any problems (yes/no)? If so, where are the problems (which leaf/angle/etc)? Further questions like Has this been going on for a while? is a good question but pylinac doesn’t really have a database aspect (yet!). QATrack+ is a good fit for that and I’m hoping that the two projects might be able to assimilate at some point. So, all that to say that adding something like a FWHM thickness test might not be a bad idea, but I’m trying hard to hold off until I’m able to 1) do it well and 2) not create mountains of useless data. I also believe such a situation where both leaves are off in opposite directions exceedingly rare. I’ve never seen it in the wild myself. More often, one leaf is lagging or leading, and all the pickets that leaf pair creates are off. This is a long answer to a simple question, but it hits at the heart of what questions this project tries to answer about analyses. So, to actually answer your question, I may end up doing a FWHM analysis, but it would be something that answers the question: “Are any leaf pairs wider or skinnier than they should be?” It wouldn’t give you all the FWHMs (pylinac would calculate them internally), just tell you if something was off. This is the stage I’m trying to take pylinac in the next few versions–interpretation of results over against lots of numbers.

I’d love to hear some other thoughts about this from anyone else. Thanks for bringing up the topic Roberto!


Hello James:

I agree with you completely. I like the way you present the results with just little data in order to summarize the analysis. It is much more clear. My previous message was just a suggestion. I agree with you that it is quite difficult that two leaves fail at the time and following a symmetric pattern in opposite directions, especially for a VARIAN machine.

Thank you.

Roberto Berenguer