Author: Saree Alnaghy, Owen Thomas Dillon, Sean P Hood, Paul J. Keall, Ricky O'Brien, Tess Reynolds 👨🔬
Affiliation: Image X Institute, Faculty of Medicine and Health, The University of Sydney, University of Wollongong, Medical Radiations, School of Health and Biomedical Sciences, RMIT University 🌍
Purpose: Photon Counting Detectors (PCDs) have been demonstrated to improve resolution, contrast and lesion detectability in fan-beam CT. There is naturally a desire to bring these improvements to Cone-Beam CT (CBCT) however PCD flat panels have proven prohibitively difficult and expensive to manufacture. We present a novel combination of a narrow PCD with Carbon Nanotube (CNT) source arrays to perform inverse geometry CBCT, enabling PCD imaging of a usable field of view in a single gantry rotation.
Methods: We simulate CBCT acquisition using a conventional Flat Panel Detector (FPD) an equivalent sized Photon Counting Flat Panel Detector (PCFPD) and a novel system combining FPDs and currently available CNT arrays and thin PCDs that we name CNTPCD. The simulation incorporates quantum noise, electronic noise, scatter, source spectra and spectral attenuation at levels established experimentally with equivalent systems. We developed a novel RecOnstructiOn using Spatial and SPEctral Regularization (ROOSSPER) algorithm that incorporates PCD data across the entire spectrum when reconstructing each energy level. Images are compared in terms Root-Mean-Square-Error (RMSE).
Results: Relative to conventional FPD CBCT, PCFPD CBCT reduced RMSE by 67% and CNTPCD reduced RMSE 70% but requiring 30% higher imaging dose.
Conclusion: Photon counting CBCT with sufficiently large FOV is achievable with currently available thin PCDs when combined with CNT source arrays to bring improved CBCT imaging into the clinic. These results motivate our ongoing construction of a prototype system with the proposed layout.