MingyeWu
commented
1 year ago Hi, dear user,
Airscan is intended to be noise-free, because, in real CT products, airscan is for the air calibration. It is repeated for many views, averaged along views, and normalized. It's nearly noise free. For example, if we make airscan with noise and repeat it to get the same mAs as the total phantom scans, the noise difference in the final image is less than 1%. And depending on the vendor, air calibration (airscan) can be with even much more mAs, which makes the difference even smaller.
However, the "airscan" in the simulation is actually vacuum scan. And the phantom scans are performed in vacuum. Because, otherwise, we need to add the surrounding air to all the phantoms. But the difference is negligible. So we ignored it.
In your results, figure C and D, they're only one view, that's why it looks very noisy.
-- Mingye
Hello, I am a student and for my scientific work I am trying to simulate a micro-CT experiment using XCIST. Here are my config files: (zip: AirExperiment.zip )
For the test, I conducted an experiment with a phantom consisting of air:
And I got the following results:
Where A is the row of the image saved with the extension ".offset"
Where B is the row of the image saved with the extension ".air"
Where C is a row from one of the images saved with the extension ".scan"
Where D is a row from one of the images saved with the extension ".prep"
It can be seen that the air-scan (B) is different from the projection at some angle from the phantom, which consists of air (= without phantom) (C)
Moreover, the image (A) is absolutely free of any noise. These moments greatly distinguish the obtained simulated CT projections from real data. Please tell me, are these the limitations of the simulator, or did I set the configuration files incorrectly?