Starshot wobble too large

Google Groups
1

I am troubleshooting a starshot gantry image. X Jaws set to 0.5 cm on film. I have tried the following code in the Python shell:

from pylinac import Starshot
star_img = “C…tiff”

mystar = Starshot(star_img, sid=1000, dpi=300)
mystar.analyze()

The result is:

RuntimeError: The algorithm was unable to determine a reasonable wobble. Try setting recursive to False and manually adjusting algorithm parameters

Running:

mystar.plot_analyzed_image()

gives

Untitled2.png
Untitled2.png2016×1512 111 KB

mystar.image.filter(size=0.01, kind=‘gaussian’)

mystar.analyze()

gives:

Untitled2.png
Untitled2.png2016×1512 100 KB

print(mystar.results())

Result: FAIL

The minimum circle that touches all the star lines has a diameter of 922.186 mm.

The center of the minimum circle is at 1080.6, 1358.8

mystar.analyze(recursive=False)
mystar.plot_analyzed_image()

Untitled2.png
Untitled2.png2016×1512 115 KB

print(mystar.results())

Result: FAIL

The minimum circle that touches all the star lines has a diameter of 161.087 mm.

The center of the minimum circle is at 1155.8, 1338.0

Does anyone know why my diameter is so large? What parameter could I tweak to get this to work? I suppose I could use MLC to get the spokes smaller, but this is the smallest I can do with jaws.

Thanks,
Landon

2

Hi,

could you upload your tiff image please. Are you certain that your SID and DPI are correct?

Denis

Dne torek, 01. oktober 2019 00.43.16 UTC+2 je oseba Landon Clark napisala:

3

The first image appears inverted. The second I can’t really tell other than that the analysis circle profile is too small. Try the following for inversion:

star = Starshot(...) star.image.invert() star.analyze(...)

4

As for the large size, see if the dpmm and dpi are accurate. For tif images pylinac tries to find this tag automatically and only then looks for the passed in data.

`
print(star.image.dpmm)
print(star.image.dpi)

`

5

Thanks for everyone’s help on this. I ran

print(star.image.dpi)
1.0

Something went wrong on the image properties. I rescanned the film using 96 dpi and ran

print(star.image.dpi)
96.0

star.analyze() ran no problem and I have expected value of this machine for circle diameter. I did have to crop before it worked properly.

Here is a link to the tiff file. gantry_ix021.tif - Google Drive

Thanks,
Landon