Room 207
Purpose: To optimize detector count, position, and beam flux in a novel x-ray fluorescence emission tomography (XFET) system to minimize dose and maximize image accuracy when jointly reconstructing metal and attenuation maps for use in theranostic applications.
Methods: A 20x20x20 numerical phantom comprised of 4 mm voxels contained gold concentrations of 0.0, 0.005, and 0.022 g/cm³ in a soft tissue background. XFET simulations were completed for varying orientations of 2 and 4 detectors placed around this object, and for incident beam fluxes ranging from 10¹⁰ to 10¹⁴ photons/cm²s in order of magnitude increments. A previously developed image reconstruction algorithm was used to jointly reconstruct gold density and attenuation images for each beam intensity and detector combination. Dose deposited from each incident x-ray beam was approximated, and image reconstruction accuracy for each beam flux and detector combination was evaluated with a normalized RMSE metric.
Results: The use of two parallel detectors generally outperformed the use of two orthogonal detectors in metal and attenuation reconstruction accuracy, evident in improved NRMSE values. Unsurprisingly, use of four detectors gave further improvement and offered qualitative reduction of blurring in reconstructed attenuation maps. Image reconstruction accuracy plateaued at a beam flux of 10¹¹; greater beam fluxes provided little benefit while administering prohibitive doses. Using an optimal four detectors, a beam flux of 10¹¹ provided accurate images (NRMSE(metal)= 0.472, NRMSE(attenuation) = 0.257) while administering acceptable dose to the object (1.76 mGy).
Conclusion: Using four detectors and a beam flux of 10¹¹ photons/cm²s in XFET maximizes metal and attenuation image reconstruction accuracy and minimizes dose. Two parallel detectors offer similar performance, with increased attenuation map blurring. At a beam flux of 10¹¹ photons/cm²s, XFET administers less dose than CT (1.76 mGy) and accurately images metal at concentrations nearing the limits of CT capabilities.
Funding Support, Disclosures, and Conflict of Interest: This work was supported by the NIH under grant number R01EB026300, and by the AAPM/RSNA Graduate Fellowship. Partial funding was provided by the NIH S10-OD025081 shared instrument grant.