Low‐dose megavoltage cone‐beam computed tomography using a novel multi‐layer imager (MLI). Issue 4 (28th January 2020)

• Record Type: Journal Article
• Authors: Myronakis, Marios; Huber, Pascal; Lehmann, Mathias; Fueglistaller, Rony; Jacobson, Matthew; Hu, Yue‐Houng; Baturin, Paul; Wang, Adam; Shi, Mengying; Harris, Thomas; Morf, Daniel; Berbeco, Ross
• Is Part Of: Medical physics. Volume 47:Issue 4(2020)
• Page Start: 1827
• Page End: 1835
• Publication Date: 2020-01-28
• Subjects: EPID -- MLI -- MVCBCT -- MVCBCT dose; Medical physics -- Periodicals; Medical physics; Geneeskunde; Natuurkunde; Toepassingen; Biophysics; Periodicals; Periodicals; Electronic journals; 610.153
• DOI: 10.1002/mp.14017
• Languages: English
• ISSNs: 0094-2405

Abstract

Abstract : Purpose: The feasibility of low‐dose megavoltage cone‐beam acquisition (MVCBCT) using a novel, high detective quantum efficiency (DQE) multi‐layer imager (MLI) was investigated. The aim of this work was to reconstruct MVCBCT images using the MLI at different total dose levels, and assess Hounsfield Unit (HU) accuracy, noise and contrast‐to‐noise ratio (CNR) for low‐dose megavoltage cone‐beam acquisition. Methods: The MLI has four stacked layers; each layer contains a combination of copper filter/converter, gadolinium oxysulfide (GOS) scintillator and a‐Si detector array. In total, 720 projections of a CATPHAN ® phantom were acquired over 360° at 2.5, 6, and 6 MV flattening filter free (FFF) beam energies on a Varian TrueBeam LINAC. The dose per projection was 0.01, 0.0167, and 0.05 MU for 2.5, 6, and 6 MV FFF, respectively. MVCBCT images were reconstructed with varying numbers of projections to provide a range of doses for evaluation. Hounsfield Unit uniformity, accuracy, noise and CNR were estimated. Improvements were quantified relative to the standard AS1200 single‐layer imager. Results: Average HU uniformity for the MLI reconstructions was within a range of 95%‐99% for all of the energies studied. Relative electron density estimation from HU values was within 0.4% ± 1.8% from nominal values. The CNR for MVCBCT based on MLI projections was 2–4× greater than from AS1200 projections. The 2.5 MV beam acquisition with the MLI exhibited the lowest noise and the best balance between CNR and dose for low‐dose reconstructions. Conclusions: Megavoltage cone‐beam acquisition imaging with a novel MLI prototype mounted on a clinical linear accelerator demonstrated substantial improvement over the standard AS1200 EPID. Further optimization of MVCBCT reconstruction, particularly for 2.5 MV acquisitions, will improve image metrics. Overall, the MLI improves CNR at substantially lower doses than currently required by conventional detectors. This new high DQE detector could provide high‐quality MVCBCT at clinically acceptable doses.