Room: Room 202
Purpose: In some cases, radiotherapy planning requires two scans: one for dose calculation, and the other for tumour delineation aided by contrast enhancement. Producing virtual non-contrast (VNC) electron density maps by removing the contribution of the iodine-based contrast agent could eliminate the need for an additional scan, thus decreasing the imaging dose to the patient. Photon counting CT is a promising technology to be used to produce contrast agent fraction and electron density maps simultaneously. The purpose of this work is to show the potential of a novel post-reconstruction approach to extract VNC physical parameters from contrast-enhanced photon counting CT.
Methods: The Bayesian eigentissue decomposition (BETD) method (Lalonde et al, 2017) is adapted to estimate the contrast agent fraction in addition to the VNC electron density. A simulation environment is developed to compare BETD with two reference methods that allows the extraction of iodine content and VNC electron density: 1) a three-material decomposition (3-MD) and 2) the photoelectric and Compton (PC) decomposition. CT numbers of human tissues mixed with the contrast agent (2%) are simulated with realistic noise and density variations for three energy bins.
Results: The 3-MD, PC decomposition, and BETD methods respectively show a mean error (bias) +/- root-mean-square error of 0.12 Â± 2.00, 0.11 Â± 2.00, 0.001 Â± 0.78 for the contrast agent fraction (%) and of 0.00 Â± 0.11, 0.00 Â± 0.07, 0.00 Â± 0.03 for the VNC electron density, demonstrating that BETD is more accurate and precise than other methods for contrast-enhanced photon counting CT.
Conclusion: In the context of contrast-enhanced scans for radiotherapy, the BETD method with photon counting CT is a promising approach for reducing the imaging dose without compromising on the quality of the treatment. In addition, the improvement in iodine fraction quantification with BETD is promising for future K-edge imaging applications.
Funding Support, Disclosures, and Conflict of Interest: MS acknowledges funding from the Natural Sciences and Engineering Research Council of Canada.