Estimate monoenergetic image from single polyenergetic image

[Dale-Black]

When creating a scan of a digital phantom, it's relatively straightforward to create a polyenergetic image via integration. Is it possible to estimate a monoenergetic image once you have the polyenergetic image, while remaining in the image domain?

[JeffFessler]

The answer is likely "no" in general, but the question needs clarification first.
Linear attenuation is a function f(x; E) that is an energy-indexed collection of images.
So your meaning of "monoenergetic image" presumably is just f(x; E_0) for some specific energy E_0.
But what is a "polyenergetic image"?

[Dale-Black]

By polyenergetic image, I mean a reconstructed image created from a polyenergetic xray spectrum

[JeffFessler]

Then probably you will need dual-energy or spectral CT for that unless your object happens to be just a single material of various densities. Look at works like https://doi.org/10.1002/mp.12096

[Dale-Black]

Got it, yes this makes sense. I have read that paper as well as a few of yours. Just wanted to check before building out some simple simulations. Thank you!

[JeffFessler]

Just this morning I saw a presentation by GE that mentioned their "True Enhance" method that uses AI (of course) for this problem. So maybe the answer is yes after all...

https://www.itnonline.com/content/ge-healthcare-highlights-commitment-delivering-better-patient-experiences-and-outcomes

[Dale-Black]

That is very interesting! I actually have a few follow up questions but they might be better suited over email. We (Imaging Physics Lab @ UCI) are beginning to dive into multi-material decomposition, which is one of your areas of expertise. Would you be willing to engage in an email thread with me and my advisor (Dr. Sabee Molloi, UCI), and if so, please do email me [email protected]

[Dale-Black]

Or if you'd rather, I would be happy to find your email online and email you directly first