CZT-based photon-counting-detector CT with deep-learning reconstruction: image quality and diagnostic confidence for lung tumor assessment
Open Access
- 7 March 2025
- journal article
- research article
- Published by Springer Nature in Japanese Journal of Radiology
- Vol. 43 (7) , 1132-1144
Abstract
Purpose: This is a preliminary analysis of one of the secondary endpoints in the prospective study cohort. The aim of this study is to assess the image quality and diagnostic confidence for lung cancer of CT images generated by using cadmium–zinc–telluride (CZT)-based photon-counting-detector-CT (PCD-CT) and comparing these super-high-resolution (SHR) images with conventional normal-resolution (NR) CT images. Materials and methods: Twenty-five patients (median age 75 years, interquartile range 66–78 years, 18 men and 7 women) with 29 lung nodules overall (including two patients with 4 and 2 nodules, respectively) were enrolled to undergo PCD-CT. Three types of images were reconstructed: a 512 × 512 matrix with adaptive iterative dose reduction 3D (AIDR 3D) as the NRAIDR3D image, a 1024 × 1024 matrix with AIDR 3D as the SHRAIDR3D image, and a 1024 × 1024 matrix with deep-learning reconstruction (DLR) as the SHRDLR image. For qualitative analysis, two radiologists evaluated the matched reconstructed series twice (NRAIDR3D vs. SHRAIDR3D and SHRAIDR3D vs. SHRDLR) and scored the presence of imaging findings, such as spiculation, lobulation, appearance of ground-glass opacity or air bronchiologram, image quality, and diagnostic confidence, using a 5-point Likert scale. For quantitative analysis, contrast-to-noise ratios (CNRs) of the three images were compared. Results: In the qualitative analysis, compared to NRAIDR3D, SHRAIDR3D yielded higher image quality and diagnostic confidence, except for image noise (all P < 0.01). In comparison with SHRAIDR3D, SHRDLR yielded higher image quality and diagnostic confidence (all P < 0.01). In the quantitative analysis, CNRs in the modified NRAIDR3D and SHRDLR groups were higher than those in the SHRAIDR3D group (P = 0.003, DLR images provided the highest image quality and diagnostic confidence for lung tumor evaluation, followed by SHRAIDR3D and NRAIDR3D images. DLR demonstrated superior noise reduction compared to other reconstruction methods.Keywords
Funding Information
- Canon Medical Systems Corporation (C2023-255)
- Japan Society for the Promotion of Science (22K07708)
This publication has 50 references indexed in Scilit:
- Lung Adenocarcinoma at CT with 0.25-mm Section Thickness and a 2048 Matrix: High-Spatial-Resolution Imaging for Predicting InvasivenessRadiology, 2020
- Noise and spatial resolution properties of a commercially available deep learning‐based CT reconstruction algorithmMedical Physics, 2020
- Prognostic Value of Pretreatment MRI in Patients With Prostate Cancer Treated With Radiation Therapy: A Systematic Review and Meta-AnalysisAmerican Journal of Roentgenology, 2019
- Ground-Glass Opacity Is a Strong Prognosticator for Pathologic Stage IA Lung AdenocarcinomaThe Annals of Thoracic Surgery, 2019
- Correction to: Deep learning reconstruction improves image quality of abdominal ultra-high-resolution CTEuropean Radiology, 2019
- Photon-counting Detector CT: System Design and Clinical Applications of an Emerging TechnologyRadioGraphics, 2019
- Deep learning–based image restoration algorithm for coronary CT angiographyEuropean Radiology, 2019
- Improvement of image quality at CT and MRI using deep learningJapanese Journal of Radiology, 2018
- Photon-counting CT: Technical Principles and Clinical ProspectsRadiology, 2018
- Evaluation of Cartilage Invasion by Laryngeal and Hypopharyngeal Squamous Cell Carcinoma with Dual-Energy CTRadiology, 2012