Background correction issue before image reconstruction

[guo404-all]

Hello,

I encountered an issue regarding background measurements while performing reconstruction using the OpenMPI dataset.

In your provided reconstruction script reco.m, the system matrix and measurement data are loaded from systemMatrix_v2.mdf and measurement_V2.mdf, respectively.
Your reconstruction code includes the following lines:

isBG = h5read(filenameSM, '/measurement/isBackgroundFrame');
S = S(isBG == 0,:,:,:);

This indicates that you directly discard the background measurement frames.
As for the measurement data in measurement_V2.mdf, since it contains no background frames, you apply the Fourier transform directly.

However, I noticed that in both systemMatrix_v2.mdf and measurement_V2.mdf, the attribute /measurement/isBackgroundCorrected is set to 0. Does this mean that you performed image reconstruction without applying background correction to either the system matrix or the measurement data? If so, is this approach correct?

My understanding is that background correction should be applied to both the system matrix and the measurement data before proceeding with image reconstruction. Is this understanding correct?

I’d greatly appreciate your clarification on this matter. Thank you!

[tknopp]

Yes, it is basically all correct what you said. Background correction is beneficial in many cases but not always. In particular, the choice of the regularization parameters get a little more sensitive. In the case of OpenMPI data, the concentration is very high, so that background subtraction is not really necessary.