Deep-learning image-reconstruction algorithm for dual-energy CT angiography with reduced iodine dose: preliminary results. Issue 2 (February 2022)

• Record Type: Journal Article
• Authors: Noda, Y.; Nakamura, F.; Kawamura, T.; Kawai, N.; Kaga, T.; Miyoshi, T.; Kato, H.; Hyodo, F.; Matsuo, M.
• Is Part Of: Clinical radiology. Volume 77:Issue 2(2022)
• Page Start: e138
• Page End: e146
• Publication Date: 2022-02
• Subjects: Medical radiology -- Periodicals; Radiotherapy -- Periodicals; Radiotherapy -- Periodicals; Radiology -- Periodicals; Societies, Medical -- Periodicals; Medical radiology; Radiotherapy; Electronic journals; Periodicals; 616.0757
• DOI: 10.1016/j.crad.2021.10.014
• Languages: English
• ISSNs: 0009-9260

Abstract

Abstract : AIM: To evaluate the computed tomography (CT) attenuation values, background noise, arterial depiction, and image quality in whole-body dual-energy CT angiography (DECTA) at 40 keV with a reduced iodine dose using deep-learning image reconstruction (DLIR) and compare them with hybrid iterative reconstruction (IR). MATERIAL AND METHODS: Whole-body DECTA with a reduced iodine dose (200 mg iodine/kg) was performed in 22 patients, and DECTA data at 1.25-mm section thickness with 50% overlap were reconstructed at 40 keV using 40% adaptive statistical iterative reconstruction with Veo (hybrid-IR group), and DLIR at medium and high levels (DLIR-M and DLIR-H groups). The CT attenuation values of the thoracic and abdominal aortas and iliac artery and background noise were measured. Arterial depiction and image quality on axial, multiplanar reformatted (MPR), and volume-rendered (VR) images were assessed by two readers. Quantitative and qualitative parameters were compared between the hybrid-IR, DLIR-M, and DLIR-H groups. RESULTS: The vascular CT attenuation values were almost comparable between the three groups ( p= 0.013–0.97), but the background noise was significantly lower in the DLIR-H group than in the hybrid-IR and DLIR-M groups ( p< 0.001). The arterial depictions on axial and MPR images and in almost all arteries on VR images were comparable ( p= 0.14–1). The image quality of axial, MPR, and VR images was significantly better in the DLIR-H group ( p< 0.001–0.015). CONCLUSION: DLIR significantly reduced background noise and improved image quality in DECTA at 40 keV compared with hybrid-IR, while maintaining the arterial depiction in almost all arteries. Highlights: DLIR significantly reduced background noise and improved SNR. Arterial depiction in DLIR was almost comparable with hybrid-iterative reconstruction. DLIR-H could improve image quality compared with hybrid-iterative reconstruction.