Using ifanbeam for a reconstruction, getting "false" negative values

Hello,

I'm fairly new to Matlab's ifanbeam function. I'm attempting to reconstruct data collected with a fan geometry, hence my need for ifanbeam rather than just iradon. My data to reconstruct is essentially integrated areas of curves measured at evenly spaced integrals around a source of interest. These curves are absorption profiles, so their value is inherently non-negative as we are not creating photons. Therefore, the integrated areas under these curves must also be non-negative. (I am using spectral fit functions, rather than the actual raw data for the integration.. so noise etc. cannot make the data negative.)

My problem seems to lie in the reconstruction process, when using ifanbeam I obtain a reconstructed matrix which contains negative values. Like I said, this is not physically possible for my experiment. Does anyone with some experience with ifanbeam know of how to restrict this? A filter? I've read through the Matlab help file, but would appreciate some more detailed info if anyone has it.

Thank you for reading! I'd appreciate any advice! Kristin

Answer 1

[Matt J]

No, there's no way to prevent negative values in a filtered back projection reconstruction. They are a bi-product of the discrete signal processing that real world computers must use, instead of the ideal continuous domain theoretical version of the algorithm. Usually, the negative values appear as sinc-ripple in the image background. People typically just ignore them, or clip them by thresholding the image.

There are more complicated, iterative reconstruction methods that do allow you to put non-negativity (and other) constraints on the solution, but they are more computationally demanding and you will probably have to code them yourself. You will have to think carefully about whether this level of complexity is worthwhile for what you are doing.