Hi,
I wrote a program in Python which simulates a CT scan taking into account the Poisson distribution of recorded photons. Taking 128 projections over 180 degrees results in the following filtered backprojection image.
What could be the reason for these vertical lines? I tried removing the "duplicate projections" for 0 and 180 deg, and checked that the phantom does not attenuate too much at the edges.
Okay, it seems that if I take 128 projections over 360 degrees, the problem persists unless I remove the last one. Makes sense as now the object is sampled evenly. But why doesn't this work with 180 degree scan? I use parallel beam geometry and make the reconstruction with iradon of scikit-image.
The spikes appear to be an alignment issue in the reconstruction. This can be visualized as a shift in the image +/- Pi rotation. The 0-180 degre projections should align if they don't a cusp artifact will form in the direction of the mis-alignment. In this case you may want to check the counting math in your projections so when you select a 0-degree projection and an 180-degree projection they would be maximally aligned.
- Badges
- Science topic
- Similar topics
- Geoscience
- Asia
- India