Author: Ishika Bhaumik, John M. Boone, Michael T Corwin, Eric S Diaz, Ahmadreza Ghasemiesfe, Andrew M. Hernandez, Sarah E. McKenney, Misagh Piran, Ali Uneri, Eric L White 👨🔬
Affiliation: UC Davis, UC Davis Health, University of California, Johns Hopkins Univ 🌍
Purpose: A new model CT scanner (Canon Aquilion One Insight) was recently installed at our institution, and it included a 3D Landmark (3DLM) scan for automatic scan planning, a new deep learning reconstruction (DLR) algorithm (PIQE) for improved resolution, and silver x-ray beam filtration for ultra-low-dose applications. The purpose of this work was to evaluate these new features in order to inform clinical protocol design with 3D measures of image quality.
Methods: A test phantom was scanned using the 3DLM acquisition (120kV/Ag, 0.2mGy CTDIvol), a conventional CT (120kV/Cu) without automatic exposure control “AEC” (4mGy CTDIvol), and CT scans with Cu and Ag filtration using AEC with an “AiCE-Standard” quality setting. Reconstruction was performed with Body and BodySharp kernels with AiCE (first-generation DLR), and PIQE with 0.42mm in-plane voxels and 0.5 and 5mm slice thicknesses. The scan with Cu filtration was defined as the “reference”. Automated software was used to measure HU accuracy, CNR, 3D-NPS, and 3D-MTF.
Results: The 3DLM scan (AiCE-BodySharp-5mm) provided satisfactory HU accuracy (3.0% mean absolute error) but lower CNR (32%-50%) compared to the reference, and a 56% decrease in the 3D MTF f10 was observed for the 3DLM scan (AiCE-Body-0.5mm). The reference scan reconstructed with PIQE-Body-0.5mm provided a 9% increase in the 3D MTF f10 and 28% lower noise compared to the reference scan with AiCE-Body-0.5mm, with clear differences in noise texture. With AEC ON, the Ag filter scan (120kV) resulted in a 48% decrease in CTDIvol compared to the reference (100kV/Cu), but a decrease in CNR (28%-58%) and increase in image noise (46%) was observed.
Conclusion: This study demonstrated that these new developments can provide specific improvements in image quality, specifically aligned with or extending beyond their intended purpose. This baseline evaluation may help guide the integration of these new scanner features into clinical protocol design.