P244] Accuracy of ct numbers obtained by dira and monoenergetic plus algorithms in quantitative dual-energy computed tomography

Abstract

Purpose The aim of this work is to compare the accuracy of the DIRA [1] and Monoenergetic Plus [2] image reconstruction algorithms in dual-energy CT. Accurate CT numbers are needed for instance in radiotherapy planning. Methods A Siemens SOMATOM Definition Force scanner was used to scan a cylindrical PMMA phantom of diameter of 160 mm with four rod inserts of diameter 20 mm made of aluminum, teflon and low-density polyethylene. For the Monoenergetic Plus algorithm, a helical, dual-energy protocol with 80 and Sn150 kV was used. Images reconstructed with the ADMIRE algorithm were processed in the syngo.via Frontier software (Siemens) and virtual monoenergetic images at energies from 40 to 180 keV were obtained. For DIRA, axial, dual-energy scan was performed, and two sets of projection data were recorded. The projection data were then processed by DIRA, which used a two-material decomposition to the bone and bone marrow doublet and a three-material decomposition to the lipid, protein and water triplet. Virtual monoenergetic images were obtained from the computed material mass fractions. The reconstruction with DIRA was also performed for computer simulated projection data. CT numbers provided by both algorithms were compared to tabulated values. Results The Monoenergetic Plus algorithm performed best for PMMA with the maximum difference of 9 HU at 40 keV. For Aluminum, the difference of about 43 HU was observed at 60 and 180 keV. In case of DIRA, computer simulations predicted good agreement in CT numbers, differences of 13 HU and 2 HU at 50 and 88keV, respectively, were observed for aluminum. Preliminary results indicate that accuracy of CT values obtained from measured projections approaches those from computer simulations. Conclusions The Monoenergetic Plus algorithm performs two-material decomposition to iodine and water and then uses the data on concentrations for the prediction of CT numbers. This approach leads to errors especially for aluminum, nevertheless the correspondence between reconstructed and tabulated values was good in general. DIRA better suppressed beam hardening artifacts and provided more accurate CT numbers.