Issue rMTF Leeds TOR

[daniel maneval]

Hi everyone,

I am encountering an issue with the rMTF calculation in pylinac.

When running the analysis on my images, the rMTF curve is not normalized to 1 at zero frequency. More concerningly, the rMTF 10% value is negative, which is physically impossible. Even though the ROIs seem perfectly positioned over the high-contrast inserts, the resulting MTF calculation appears flawed.

I tried inverting the image:

1. With inversion: The rMTF curve looks much better (correctly normalized to 1. However, the ROI positioning becomes completely broken/misaligned.

2. Without inversion: The ROIs are perfectly placed, but the rMTF values are incorrect (non-normalized and negative) as shown below.

It seems I am caught between a correct ROI detection and a correct MTF calculation. Has anyone experienced this? Is there a recommended way to handle this polarity/normalization conflict without displacing the ROIs?

Thanks for your help ! 

[tbe...@gmail.com]

Hi,

is it just the window level or are you using a different version of the phantom than the one shown in the documentation? 

Did you try the analysis with the leedsTORBlue also?