Picket Fence Mean Picket Spacing

Dear all,

I’m new to pylinac, so please don’t be annoyed by my crude questions.

Currently, I’m trying to evaluate EPID Picket Fence images from the Unity MR-Linac using the pylinac PicketFence module. To keep it simple, I created an image of four pickets. Each is 1 cm wide and the picket to picket separation is 2 cm. Pickets extend in the LR direction, as this is the fixed orientation of the MLC in the system (90° on conventional machines).

The python script is running without errors and I can produce the attached PDF summary. The magnification factor (SID=2658/SAD=1430) is evaluated correctly for the Unity. Though, I’m having trouble with the calculated Mean Picket Spacing value of 13.9 mm. In my understanding it should be close to 20 mm, as this is the picket to picket separation.

This might be coincidentially, but the diffrence is a factor of 20 mm/13.9 mm=1.44, which is very close to the ratio of a conventional linac SAD of 1000 mm and the Unity SAD of 1430 mm. Could this be the source of the problem?

I’m attaching the DICOM EPID image, the pylinac results PDF and my humble python script. It would be great if somebody has a helpful comment on this topic.

Best regards,
David

pf211001.pdf (74 KB)

MRL_Picket_Fence.py (400 Bytes)

211001_RTI00004 (2 MB)

Hi David,

I think you are right with your assumption. Scaling is done via pylinac.core.image.DicomImage.dpmm(). The documentation has the following comment:
“The Dots-per-mm of the image, defined at isocenter. E.g. if an EPID image is taken at 150cm SID, the dpmm will scale back to 100cm.”

The corresponding code is:
if dpmm is not None and self.sid is not None: dpmm *= self.sid / 1000
which means the Isocenter Distance is hardcoded.

You could file a improvement request on github if you want. The relevant DICOM Tags are provided and it should be possible to use them.

with kind regards
Thomas

David,
Both you and Thomas are correct; the isocenter is assumed to be 1000mm. I will file a ticket to use the dicom tag if it exists. Thanks for reporting!