Winston Lutz and Abnormally Large Couch


We are using Pylinac v2.2.6 and we got what I would think is a strange result for our Couch 2D isocenter diameter. I have attached both the zip file for the images and the report that was generated.

Looking at the results it doesn’t appear as if the couch has a substantially larger isocenter than the other axes of rotation. Am I missing something?


wl-report_2020-07-20_8f1c0b.pdf (616 KB) (5.03 MB)

Hi. I think your problem is that you don’t have many points. Pylinac assumes that the couch rotation will produce some kind of circle. So it is trying to fit a circle to your three points and ends with a large radius. These is what you get when you try to fit a circle (unknown center and radius) to three vectors (three images: one reference and two couch rotations).


Thinking about this again, maybe the circle fit is not the way to go. Because I have seen on certain linac that the BB points will not move along a circle, but sometimes along a jagged line when the couch is rotated. This will give a large radius as well.

Pylinac does expect more images than this. pylinac tries to be smart about the couch by fitting the data to a circle. This is to avoid offset errors if your BB isn’t perfectly set left/right. I.e. if your bb is 1mm to the right at couch=0 and your couch is perfect then your points will perfectly fit a circle of 1mm and pylinac sees the distance to the points to the fitted circle as 0 and your couch rotation will be reported as (and actually is, despite the 1mm shift) 0. A better implementation would be to find the minimum between a fitted circle and a simple centroid calculation.