Impact of reconstruction-based misalignment correction on image quality, reader agreement, and diagnostic confidence in single-source photon-counting CT
Related Experiment Video
Updated: Oct 30, 2026
Multimodal Cross-Device and Marker-Free Co-Registration of Preclinical Imaging Modalities
Published on: October 27, 2023
Kristóf Nagy1, Krisztián Kocsis1, Gergő Veréb1
1Department of Radiology, Medical Imaging Centre, Semmelweis University, 2 Koranyi Sandor Street, Budapest, 1083, Hungary.
Purpose:
Recently, a single-source photon-counting detector (PCD-CT) system with detector settings comparable to the dual-source system has been introduced. Although applicable to coronary CT angiography (CCTA), its lower temporal resolution may lead to more pronounced motion and misalignment artefacts. Image slab misalignment correction algorithm reduces these artefacts; therefore, we aimed to evaluate its effect on subjective image quality metrics on this PCD-CT system.
Methods:
CCTA images were reconstructed with and without using misalignment correction algorithm. Four readers with varying experience evaluated coronary vessels using three image quality metrics: motion, misalignment artefact, and diagnostic confidence. Linear mixed-effect models were used to assess the effects of reconstruction type, image quality metrics, vessels, and readers. To assess reader agreement, intraclass correlation and multiple agreement metrics were used.
Results:
102 patients (53.9% female) were included in the final analysis. The type of reconstruction significantly affected image quality metric scores (p = 0.031). Post-hoc analysis showed that misalignment corrected reconstruction significantly improved misalignment artefacts (Δ = 0.94, 95% CI, 0.88-0.99, p < 0.01) and diagnostic confidence (Δ = 0.296, 95% CI, 0.24-0.36, p < 0.01). Misalignment correction showed a more modest effect on motion artefacts (Δ = 0.065, 95% CI, 0.008-0.122, p = 0.026). Inter-rater reliability was good for both reconstruction types (ICC = 0.713 and ICC = 0.657 for original dataset and corrected dataset). Agreement with the most experienced reader improved using misalignment correction (Exact agreement with the least experienced reader: 43.3% and 54.1%).
Conclusion:
The correction algorithm reduced misalignment artefacts and enhanced subjective diagnostic confidence on single-source PCD-CT, with benefits observed across readers of varying experience.
