Hi James,
First of all, thank you very much for creating and developing Pylinac. Open source projects is, in my humble opinion, a huge contribution to medical physics. (Same goes for PyDICOM… makes my everyday life so much easier.)
I am currently testing pylinac’s Winston-Lutz analysis. I observed some discrepancies in the gantry isocenter position that I would like to discuss on this forum.
For everyone to know, pylinac works great with Elekta EPID (iview) as long as you edit the DICOM’s metadata to add the collimator angle and couch angle. I used the standard Elekta Ballbearing to perform various WL tests.
I ran a Gantry WL where the ballbearing was shifted by a known distance (1 cm towards B). I took 8 images at different gantry angles. Colli=0 and Couch=0. I attached the images for the 4 cardinal angles to this post
Pylinac obtained the following result :
In [292]: wl.gantry_iso2bb_vector
Out[292]: Vector(x=-0.87, y=0.00, z=-0.06)
This result is very far from the expected 1 cm shift in X.
I did some debugging… All WLImages have consistent x_offset and y_offset, therefore I think that I didn’t make any mistakes in my DICOM metadata edition… but I cannot be sure.
There may be a bug in the cax_line_projection function… the Vector G=0 is ~20 cm away from the Vector at G=180.
What do you think ?
Michel
Shift1cmB_4ch.zip (5.28 MB)