Evaluation of a super‐resolution deep learning reconstruction algorithm in abdominal CT imaging—A qualitative and quantitative performance analysis
Journal of Applied Clinical Medical Physics · 2026
DOI: 10.1002/acm2.70633 ↗Kurz gefasst
Evaluationsstudie in der Zeitschrift Journal of Applied Clinical Medical Physics, 2026. Verfasst von Aria M Salyapongse und 8 weiteren Autor:innen. Beteiligte Einrichtung: Departments of Medical Physics and Radiology, University of Wisconsin-Madison, Madison, Wisconsin, USA. Hauptthemen laut MEDLINE: Deep Learning, Tomography, X-Ray Computed, Algorithms, Radiography, Abdominal. Die Arbeit nennt 28 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Background
A super‐resolution deep learning (DL) image reconstruction algorithm (Precise Image Quality Engine (PIQE)) was originally designed for cardiac CT, but is now available for abdominal CT.
Purpose
To examine objective and subjective image quality (IQ) improvements PIQE compared to Advanced Intelligent Clear‐IQ Engine (AiCE) in abdominal CT.
Methods
A retrospective analysis was conducted on 69 adult patient routine contrast enhanced abdominopelvic CT exams on a single Aquilion ONE/INSIGHT CT system (Canon Medical Systems, Otawara Japan). Images were reconstructed using PIQE (strength level L1 and L2) and AiCE (L1– the institutional standard). Four blinded radiologists assessed image noise, image contrast, small structure visibility, image sharpness, artifacts, and overall image preference with Likert scales. Reader agreement was assessed with Krippendorff's alpha. Circular regions of interest were placed on five slices on the left and right liver, portal vein, aorta, subcutaneous fat, and bilateral psoas muscles. CT number, noise, signal‐to‐noise ratios (SNRs), and contrast‐to‐noise ratios (CNRs) were determined. All significant differences between reconstructions were assessed via the Friedman test with post‐hoc Dunn‐Sidak corrections.
Results
Reader agreement was fair (). PIQE L2 was preferred for image contrast and image noise and PIQE L1 was preferred for image sharpness ( p < 0.05). CT numbers were significantly different between AiCE L1 and PIQE ( p < 0.05) and noise was statistically lowest in PIQE L2 compared to AiCE L1 ( p < 0.05). SNR and CNR differences were statistically significant ( p < 0.003), with PIQE L2 demonstrating the highest SNR and CNR.
Conclusion
The best subjective IQ metrics for image contrast, image noise, and image sharpness were obtained with PIQE. The best objective IQ metrics (SNR and CNR) were obtained with PIQE L2. This work supports improved image contrast and decreased noise when using PIQE as compared to AiCE.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Stichwörter der Autor:innen
Substanzen
Autor:innen und Institutionen
- Departments of Medical Physics and Radiology, University of Wisconsin-Madison, Madison, Wisconsin, USA.
- Department of Radiology, University of Wisconsin-Madison, Madison, Wisconsin, USA.
Zitieren
APA
Salyapongse, A. M., Wagner, M. G., Lubner, M. G., Schluter, K. L., Smith, M. R., Taya, M. D., Wentland, A. L., Szczykutowicz, T. P., & Toia, G. V. (2026). Evaluation of a super‐resolution deep learning reconstruction algorithm in abdominal CT imaging—A qualitative and quantitative performance analysis. Journal of Applied Clinical Medical Physics. https://doi.org/10.1002/acm2.70633
Vancouver
Salyapongse AM, Wagner MG, Lubner MG, Schluter KL, Smith MR, Taya MD, et al. Evaluation of a super‐resolution deep learning reconstruction algorithm in abdominal CT imaging—A qualitative and quantitative performance analysis. Journal of Applied Clinical Medical Physics. 2026. doi:10.1002/acm2.70633
BibTeX
@article{Salyapongse2026,
author = {Salyapongse, Aria M and Wagner, Martin G and Lubner, Meghan G and Schluter, Kelsey L and Smith, Matthew R and Taya, Michio D and Wentland, Andrew L and Szczykutowicz, Timothy P and Toia, Giuseppe V},
title = {Evaluation of a super‐resolution deep learning reconstruction algorithm in abdominal CT imaging—A qualitative and quantitative performance analysis},
journal = {Journal of Applied Clinical Medical Physics},
year = {2026},
doi = {10.1002/acm2.70633}
}
EUVIMED