Application of a pulmonary nodule detection program using AI technology to ultra-low-dose CT: differences in detection ability among various image reconstruction methods.
Authors
Affiliations (3)
Affiliations (3)
- Department of Radiology, Graduate School of Medical Science, University of the Ryukyus, 1076 Kiyuna, Ginowan-shi, Okinawa, 901-2720, Japan. [email protected].
- Department of Radiology, Graduate School of Medical Science, University of the Ryukyus, 1076 Kiyuna, Ginowan-shi, Okinawa, 901-2720, Japan.
- Division of Radiology, Department of Medical Technology, University of the Ryukyus Hospital, Okinawa, Japan.
Abstract
This study aimed to investigate the performance of an artificial intelligence (AI)-based lung nodule detection program in ultra-low-dose CT (ULDCT) imaging, with a focus on the influence of various image reconstruction methods on detection accuracy. A chest phantom embedded with artificial lung nodules (solid and ground-glass nodules [GGNs]; diameters: 12 mm, 8 mm, 5 mm, and 3 mm) was scanned using six combinations of tube currents (160 mA, 80 mA, and 10 mA) and voltages (120 kV and 80 kV) on a Canon Aquilion One CT scanner. Images were reconstructed using filtered back projection (FBP), hybrid iterative reconstruction (HIR), model-based iterative reconstruction (MBIR), and deep learning reconstruction (DLR). Nodule detection was performed using an AI-based lung nodule detection program, and performance metrics were analyzed across different reconstruction methods and radiation dose protocols. At the lowest dose protocol (80 kV, 10 mA), FBP showed a 0% detection rate for all nodule sizes. HIR and DLR consistently achieved 100% detection rates for solid nodules ≥ 5 mm and GGNs ≥ 8 mm. No method detected 3 mm GGNs under any protocol. DLR demonstrated the highest detection rates, even under ultra-low-dose settings, while maintaining high image quality. AI-based lung nodule detection in ULDCT is strongly dependent on the choice of image reconstruction method.