Yuan Li, Ping He, Junli Ma, Na Wu, Xiaoguang Wang, Jian Zhang
Radiation dose reduction in pediatric chest computed tomography (CT) is crucial because of increased radiosensitivity and potential long-term cancer risk. Low tube voltage combined with iterative reconstruction can reduce radiation exposure; however, the optimal reconstruction strength that balances image quality and diagnostic acceptability remains unclear. In addition, evidence remains limited regarding the optimal ASiR-V blending level in pediatric low-dose chest CT, particularly when subjective image quality, objective metrics, and radiation dose are evaluated together. This retrospective study included 200 children (0-6 years) undergoing chest CT. One hundred patients were scanned using an 80 kVp protocol reconstructed with 6 adaptive statistical iterative reconstruction-V (ASiR-V) blending levels (0-100%), and 100 patients underwent conventional 100 kVp CT. Subjective image quality was independently assessed by 2 radiologists using a five-point scale under lung and mediastinal window settings. Objective image quality was evaluated using image noise, signal-to-noise ratio (SNR), and contrast-to-noise ratio (CNR). Radiation dose parameters, including CTDIvol, dose-length product, and effective dose, were recorded and compared. Radiation dose metrics were significantly lower with the 80 kVp protocol than with the 100 kVp protocol (all P < .001), resulting in an approximately 48% reduction in effective dose. Image noise progressively decreased with increasing ASiR-V blending levels, accompanied by corresponding increases in SNR and CNR, with optimal objective values at 100% ASiR-V. Subjective image quality peaked at intermediate blending levels, with ASiR-V 80% achieving the highest diagnostic acceptability and scores comparable to those of conventional 100 kVp CT. Low-dose pediatric chest CT using 80 kVp acquisition achieves substantial radiation dose reduction while maintaining diagnostic image quality when optimized iterative reconstruction is applied. Among the 6 ASiR-V blending levels evaluated, ASiR-V 80% provided the most favorable balance between subjective diagnostic acceptability, noise suppression, and preservation of image texture, whereas ASiR-V 100% yielded the best objective noise-related metrics.