To explore the application value of low radiation dose combined with AI deep learning reconstruction algorithm in postpartum pulmonary artery CTA

Objective Postpartum patients with pulmonary embolism are radiation-sensitive groups with breastfeeding needs and radiation anxiety. This study explores the feasibility and value of combining artificial intelligence deep learning reconstruction algorithm (ClearInfinity) with low radiation dose in pulmonary artery CT angiography (CTA) for postpartum patients, providing a basis for radiation protection and diagnostic optimization in postpartum population.Methods This study is a prospective study involving 60 patients with suspected or follow-up postpartum pulmonary embolism at Hubei Maternal and Child Health Hospital from September 2024 to April 2025. Patients were randomly divided into two groups: the conventional dose group (Group A, 30 cases) and the low-dose group (Group B, 30 cases). Group B was further randomly divided into two subgroups: Group B1 (ClearInfinity algorithm, 30 cases) and Group B2 (ClearView algorithm, 30 cases).Group A operated at a tube voltage of 120kV with 50% ClearView iterative reconstruction; Group B operated at a tube voltage of 100kV, using either 50% ClearInfinity deep learning reconstruction (Group B1) or 50% ClearView iterative reconstruction (Group B2).Compare the radiation dose parameters between the two groups, including Volume CT Dose Index (CTDIvol), Dose Length Product (DLP), and Effective Dose (ED); measure the CT values of the regions of interest (ROI) for the three groups; perform one-way analysis of variance (ANOVA) on the CT values, image noise (SD), signal-to-noise ratio (SNR), and contrast-to-noise ratio (CNR) among the three groups；Two radiologists with 8-10 years of experience in diagnostic imaging were invited to subjectively score the images using a 5-point scale, and Kappa test was applied to analyze the consistency of the results.Results The three groups showed no significant statistical difference (P>0.05) in general data such as age, postpartum days, and BMI. The average BMI was 26.55±2.883kg/㎡ in group A and 26.54±2.886 kg/㎡ in group B.Radiation dose: In group B, CTDIvol was 5.84mGy (A 43.3% decrease compared to 10.3mGy in group A), DLP was 176.92±11.97mGy · cm (a 36.87% decrease), and ED was 2.48±0.17mSv (a 36.73% decrease), all P<0.01.Objective indicators: The pulmonary artery CT values of all three groups were >300HU, meeting the diagnostic criteria. The ROI SD values of group B1 (8.32±1.08, 8.73±0.88, 8.50±0.91) were lower than those of group B2 (11.91±1.81, 13.73±2.01, 12.50±1.58) and group A. The SNR and CNR were more than 75% higher than those of group B2 and more than 43% higher than those of group A (all P<0.01).The subjective scoring showed a Kappa value of 0.75 to 0.88 (high consistency), with Group B1 (4.8±0.2) being significantly higher than Group B2 (4.33±0.3) and Group A (4.47±0.2) (P<0.01).Conclusions 100kV low-dose scanning combined with the ClearInfinity deep learning reconstruction algorithm can reduce the radiation dose by 36.73% in postpartum pulmonary artery CTA with a BMI of 24-28kg/㎡, while improving the image quality. This conforms to the principle of radiation protection and can alleviate the radiation anxiety and breastfeeding concerns of postpartum patients. It has definite clinical value; Its applicability to patients with a BMI of ≥28kg/㎡ or prenatal patients requires further research.