Assessing variations in 3D image quality in chest CT across sites and scanners
Medical Physics · 2026-04 · 53 (4)
DOI: 10.1002/mp.70402 ↗Kurz gefasst
Fachartikel in der Zeitschrift Medical Physics, online erschienen am 01.04.2026. Verfasst von Andrew M Hernandez und 11 weiteren Autor:innen. Beteiligte Einrichtung: Department of Radiology, University of California Davis Health, Sacramento, California, USA. Hauptthemen laut MEDLINE: Radiography, Thoracic, Imaging, Three-Dimensional, Tomography, X-Ray Computed, Tomography Scanners, X-Ray Computed. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 47 Literaturangaben. Gefördert von: NHLBI NIH HHS, National Institute of Biomedical Imaging and Bioengineering (NIBIB).
Abstract (Originalsprache)
Background
Large CT image databases are critical to the development and validation of protocol harmonization strategies and AI models. However, image quality (IQ) can vary significantly due to differences in technology and site practices, even for scanners from the same manufacturer, which can have considerable effects on harmonization efforts and AI algorithm development. Quantifying the 3D IQ variability represented in these databases is essential to characterize protocol heterogeneity, the training data distribution, and to assess test data overlap.
Purpose
To quantify and characterize sources of 3D IQ variability in chest CT across scanner models and imaging sites.
Methods
A previously‐reported phantom and its automated analysis software was employed across 6 sites, 3 CT manufacturers, and 27 scanners (10 unique models) to rigorously quantify IQ metrics, including image noise, image contrast (low‐contrast objects: −26 HU, 41 HU; medium: −84 HU; high: 276 HU, 817 HU), the 3D MTF (axial plane and oblique direction), and the 3D NPS. The MTF was summarized using the frequency at 10% modulation “ f 10 ”, and NPS using average frequency “ f avg ”. Sites were instructed to use their non‐contrast chest CT protocols for fixed mAs phantom scanning at CTDI vol,32 levels of 2.1 (low dose “LD”), 4.2 (MD), and 6.3 mGy (HD), and provide four reconstructions, including standard ‘Std’ and thin ‘Thn’ slice thicknesses for sharp “Sh” and soft “So” kernels. Variance component analysis (VCA) was performed on the resulting phantom series to quantify sources of variability in IQ partitioned into inter‐scanner model, inter‐site, and residual (i.e., unexplained) components.
Results
IQ metrics dominated by inter‐scanner differences (> 50% of total variance) included image noise for Sh‐Std‐LD/MD, Sh‐Thn‐LD, and So‐Std/Thn for all doses (51%–84%), high‐contrast (276, 817 HU) image contrast for all protocols (68%–93%), medium‐contrast image contrast for all MD/HD protocols (64%–88%), low‐contrast (−26 HU) image contrast for So‐Std‐MD (65%) and Sh‐Std‐LD/MD (64%/51%), axial and oblique for nearly all protocols (58%–88% and 56%–58%, respectively), axial f avg for Sh‐Thn across all dose levels (51%–65%), and oblique f avg for So‐Thn‐HD (58%) and So‐Thn across all dose levels (56%–91%). Metrics dominated by inter‐site differences were image noise for Sh‐Std/Thn‐HD (57%/58%), low‐contrast (41 HU) for So‐Std‐MD/HD (77%/54%) and Sh‐Std‐MD (64%), and axial f avg for So‐Thn‐LD/MD (54%, 70%). Residual variance was the dominant contributor to variability in medium‐contrast image contrast for So/Sh‐Std‐LD (91%/56%), low‐contrast (41 HU, −26 HU) image contrast for So‐Std‐LD (58%–70%) and Sh‐Std‐LD/HD (76%/66%), and axial for So‐Thn‐HD (54%).
Conclusion
This multisite phantom study demonstrates that substantial, systematic variability in 3D IQ exists across scanners, sites, metrics, dose levels, and reconstruction settings despite use of a routine non‐contrast chest CT protocol. Variability was primarily driven by scanner model differences, with additional contributions from inter‐site differences and residual variance for select metrics and protocols. Regardless of controlled imaging conditions that limit variability from patient anatomy, size‐dependent dose effects, and tube current modulation, substantial variability in 3D IQ persists. The size‐specific, relative trends observed in this study emphasize the importance of accounting for scanner model, site, and protocol‐specific heterogeneity when using large CT image databases.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Department of Radiology, University of California Davis Health, Sacramento, California, USA.
- Department of Biomedical Engineering, Johns Hopkins University, Baltimore, Maryland, USA.
- Department of Radiology, MaineHealth Maine Medical Center, Portland, Maine, USA.
- Department of Radiology, University of Washington, Seattle, Washington, USA.
- Department of Radiology and Nuclear Medicine, Rush University Medical Center, Chicago, Illinois, USA.
- Department of Radiology, Henry Ford Health, Detroit, Michigan, USA.
Förderung
- NHLBI NIH HHS
- National Institute of Biomedical Imaging and Bioengineering (NIBIB)
Zitieren
APA
Hernandez, A. M., Uneri, A., Bevins, N., Kanal, K., Zamora, D., Supanich, M., Maloney, B., Mahesh, M., McNitt-Gray, M. F., Kinahan, P., Siewerdsen, J. H., & Boone, J. M. (2026). Assessing variations in 3D image quality in chest CT across sites and scanners. Medical Physics, 53(4). https://doi.org/10.1002/mp.70402
Vancouver
Hernandez AM, Uneri A, Bevins N, Kanal K, Zamora D, Supanich M, et al. Assessing variations in 3D image quality in chest CT across sites and scanners. Medical Physics. 2026;53(4). doi:10.1002/mp.70402
BibTeX
@article{Hernandez2026,
author = {Hernandez, Andrew M and Uneri, Ali and Bevins, Nicholas and Kanal, Kalpana and Zamora, David and Supanich, Mark and Maloney, Benjamin and Mahesh, Mahadevappa and McNitt-Gray, Michael F and Kinahan, Paul and Siewerdsen, Jeffrey H and Boone, John M},
title = {Assessing variations in 3D image quality in chest CT across sites and scanners},
journal = {Medical Physics},
year = {2026},
volume = {53},
number = {4},
doi = {10.1002/mp.70402}
}
EUVIMED