Reducing radiographic retake rates with ultra-low dose pre-shots using deep learning-based noise reduction: A prospective clinical study.
Authors
Affiliations (4)
Affiliations (4)
- Department of Radiological Technology, Saiseikai Kawaguchi General Hospital, 5-11-5 Nishikawaguchi, Kawaguchi, Saitama, 332-8558, Japan. Electronic address: [email protected].
- Department of Radiological Technology, Saiseikai Kawaguchi General Hospital, 5-11-5 Nishikawaguchi, Kawaguchi, Saitama, 332-8558, Japan.
- Department of Radiological Technology, Faculty of Medical Technology, Teikyo University, 2-11-1 Kaga, Itabashi, Tokyo, 173-8605, Japan.
- Department of Radiological Sciences, Graduate School of Human Health Sciences, Tokyo Metropolitan University, 7-2-10 Higashi-Ogu, Arakawa, Tokyo, 116-8551, Japan.
Abstract
The pre-shot technique is an effective approach to lower retake radiography rates; however, it is limited by increased patient radiation exposure. This study aimed to evaluate the impact of an ultra-low-dose pre-shot enabled by deep learning-based noise reduction (Advanced Intelligent Clear-IQ Engine: AiCE) on reducing retake rates in wrist radiography. A CXDI-810C (Canon Medical Systems) was used as the X-ray detector. Phantom images simulating lateral wrist pre-shots with and without AiCE were acquired under fixed imaging conditions using an additional 1.0 mm copper filter. The tube current-time product was reduced from 2.8 mAs to 0.5 mAs. Fifteen radiological technologists determined the optimal imaging condition using a five-point confidence scale, with scores below three considered insufficient for retake decision-making. A prospective clinical evaluation was then conducted, comparing retake rates between a non-pre-shot group (n = 542) and a pre-shot group (n = 325). The entrance surface air kerma (K<sub>a,e</sub>) was estimated from phantom measurements. Without AiCE, no imaging condition met the threshold required for retake decision-making. Conversely, with AiCE, a tube current-time product of 1.6 mAs achieved acceptable image quality (3.29 ± 0.47), with all evaluators assigning scores ≥3; therefore, it was selected as the optimal condition. In the clinical evaluation, the retake rate markedly decreased from 14.38% to 0.61%. The pre-shot K<sub>a,e</sub> was 0.11 ± 0.03 μGy, corresponding to only 0.15% of the standard lateral wrist radiography dose. AiCE-enabled ultra-low-dose pre-shot imaging dramatically reduced retake rates while requiring minimal additional radiation dose. The ultra-low-dose pre-shot using AiCE represents a highly effective strategy to minimize retake imaging with negligible dose penalty, offering substantial potential for dose optimization and improved workflow efficiency in clinical radiography.