Multi-Met WL Acceptable Image Acquisition Method


I am having trouble getting the multi-met winston lutz analysis to work but I couldn’t seem to find anything in documentation describing exactly what kind of planning method it will accept. Specifically, can it handle multi-met fields planned as individual fields (as in each BB is targeted by a separate field) or does it need to acquire all targets in one single field/image?

To give some background, we have an in-house multi-met phantom that has multiple targets that often don’t perfectly line up on an MLC interface. We’ve followed this paper’s recommendations to slightly adjust the collimator angle in order to align the desired BB to an MLC interface: However, using this method, only two BBs can be targeted at a time, isocenter and one off-axis target as the other ones don’t end up aligning. My hope was to create multiple fields to target one pair at a time before moving to the next gantry angle, but so far the code doesn’t seem to be liking that. I could be doing something entirely wrong in the code as well, but so far I keep getting errors stating that my off axis BBs can’t be found.

Any suggestions?

Hi Chase,

as far as I understand the documentation (see it should work with both single image multi BBs (like the ones provided with the demo) and single image for each BB. “For each BB and image combination, the image is searched at the nominal location for the BB and radiation field. If it’s not found it will be skipped for that combo. The BB must be detected in at least one image or an error will be raised.”

You did not write what your actual problem is. Did you get an error or just some wonky results?

If you have a dataset you could provide, I think James would be glad if you could upload it via his “image donation form”


Chase, Sorry for the late reply. I had a message written but didn’t hit send. As far as I understand it, this seems like a valid use case. If you send the images as Thomas mentioned I’ll take a look. I’m catching up on submitted bugs/phantoms but hopefully this one is straightforward.

Thanks for the response Thomas and James! My apologies for forgetting to include the right details. I managed to solve the issue but in case anyone else encounters the issue, this is the error message I received.

ValueError: Did not find any field/bb pairs for bb: {‘name’: ‘2’, ‘offset_left_mm’: 50.3, ‘offset_up_mm’: 47.7, ‘offset_in_mm’: 68.0, ‘bb_size_mm’: 5, ‘rad_size_mm’: 20}

I wasn’t sure if this meant that if it couldn’t find the field in any given image that it would throw an error, but in retrospect, I just had a rooky mistake and had entered in my BB offsets in cm instead of mm so it wasn’t searching in the right place.

Nice! Glad you got it working. I might add this use case to the docs for using Anton’s paper.