Feasibility of deep learning-based image reconstruction using TrueFidelity for coronary calcium scoring
Ji Won Kim, Tae Hoon Kim, Seung Hyun Lee, Ji Eun Nam, Chul Hwan Park
Abstract
We evaluated the feasibility of deep learning-based image reconstruction (DLIR) using GE TrueFidelity at high strength for coronary calcium scoring, compared with filtered back projection (FBP) and iterative reconstruction (IR; adaptive statistical iterative reconstruction-V with 80% blending), using a cardiac phantom and clinical data.
Introduction
Ischemic heart disease (IHD) remains a leading cause of morbidity and mortality worldwide [1–3]. Given its substantial global burden, considerable efforts have been made to improve risk prediction. Quantifying calcium in the coronary arteries has emerged as a feasible method for predicting IHD [4].
Materials and methods
This retrospective study was approved by the ethics committee and institutional review board of Gangnam Severance Hospital (3-2022-0291). The requirement for written informed consent was waived due to the study’s retrospective design. All methods were conducted in accordance with the relevant guidelines and regulations.
Results
Attenuation within the chamber was 2.4 ± 1.1 with FBP, 2.7 ± 0.6 with IR, and 2.3 ± 0.9 with DLIR. Image noise within the chamber was 10.8 ± 0.3 with FBP, 6.8 ± 0.2 with IR, and 6.0 ± 0.1 with DLIR. Calcium scores were 59.3 ± 2.2, 59.9 ± 2.3, and 59.1 ± 1.9 with FBP, IR, and DLIR, respectively.
Discussion
This study showed that DLIR achieved high overall concordance with FBP and IR for coronary calcium scoring while reducing image noise. However, DLIR showed systematic underestimation and non-negligible individual-level disagreement from FBP, with false-negative CACS 0 reclassification in a small subset of patients.
Conclusion
CACS with DLIR showed high overall concordance with CACS with FBP and IR. However, a degree of systematic underestimation was observed, and caution is warranted when interpreting these results, particularly in patients with low calcium burden or scores near CAC risk-category thresholds.
Citation: Kim JW, Kim TH, Lee SH, Nam JE, Park CH (2026) Feasibility of deep learning-based image reconstruction using TrueFidelity for coronary calcium scoring. PLoS One 21(9): e0358342. https://doi.org/10.1371/journal.pone.0358342
Editor: Suzanne Rose, Stamford Health System: Stamford Hospital, UNITED STATES OF AMERICA
Received: January 19, 2026; Accepted: August 31, 2026; Published: September 18, 2026
Copyright: © 2026 Kim et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: Individual-level clinical data cannot be shared publicly due to ethical and patient privacy restrictions. Data access requires approval from the Institutional Ethics Committee of Gangnam Severance Hospital and is available to qualified researchers through the Gangnam Severance Hospital Clinical Trials Center. Contact information: Gangnam Severance Hospital Clinical Trials Center Email: gnocr@yuhs.ac Telephone: +82-2-2019-4858 Website: https://ocr.yuhs.ac/CTC/Gangnam/ Dataset: De-identified clinical CT dataset from Gangnam Severance Hospital used in the study “Feasibility of deep learning-based image reconstruction for coronary calcium scoring” (IRB No. 3-2022-0291).
Funding: The author(s) received no specific funding for this work.
Competing interests: The authors have declared that no competing interests exist.
