Additional Detector Rotation Controls

[hansungsangyul]

Dear WebCT Development Team,
I am currently conducting research on geometric calibration methods for cone-beam CT systems, specifically focusing on the effects of various geometric misalignments on reconstruction quality. While using your WebCT software, I've been able to successfully model rotation axis offset, mid-plane offset, and source-detector distance errors through the existing UI controls.
However, I'm encountering limitations when attempting to simulate detector rotation errors, particularly:

1. Skew (rotation around the z-axis)
2. Roll (rotation around the y-axis)

According to research literature, especially studies by Meng et al. (2013), these detector rotation errors significantly impact reconstruction quality and are critical parameters to include in a comprehensive calibration study.
In the current WebCT interface, I can only control:

1. Sample rotation (pitch, yaw, roll)
2. Beam source to sample distance (x, y, z)
3. Detector to sample distance (x, y, z)

Are there any hidden options, configuration file parameters, or API calls that would allow me to control detector rotation directly? Alternatively, would your team consider adding these controls to a future version of the software?

[WYVERN2742]

Thank you for the issue and I'm happy you're enjoying WebCT!

At the moment, there isn't a way in WebCT itself to implement detector skew or roll, however if you know python, there is a potential alternative.

WebCT uses the gVirtualXRay simulator software for its simulations, and this does support detector skew and roll by changing the detector's up and right vectors.

The easiest way to use it is with the WebCT's python export function:

1. Launch WebCT and setup the experiment how you want
2. Go to the Configuration Menu (The white cog on the bottom-left of the window)
3. Change export type to "Python File"
   
4. Save the file
5. Install CIL in a conda environment
6. Activate the environment and install gVXR with pip install gvxr
7. Running the previously downloaded python file should simulate and reconstruct your projections
   Detector positioning in gVXR is done by the vertical and right-hand vectors, add in this code after setup_gVXR(CONFIG) end of the file:

gvxr.setDetectorUpVector(x, y, z)
gvxr.setDetectorRightVector(x, y, z)

By WebCT's convention, we use gvxr.setDetectorUpVector(0, 0, -1) and gvxr.setDetectorRightVector(-1, 0, 0)
The source positioned at -y, and detector at +y, with the sample at (0,0,0)

The negative values are used to flip the detector for viewing due to differences in OpenGL's coordinate space compared to CIL's reconstruction space.

As for implementing this in WebCT, I'll add it to the list! It shouldn't be too complicated to add since we won't take the detector's skew and roll into account for the reconstruction
I can't guarantee a timeline, but it should be in the next update which will at minimum be a few weeks from now, sorry.

Hope this helps!

[WYVERN2742]

Note: There is currently an bug with the python export where you have to rename "Count" to "PhotonCount" in the settings if you're using a synchrotron source: #125

Before	After