Feasibility of deep learning reconstruction algorithm combined with the "ultra triple-low" protocol at 60 kVp in CTPA: An image quality and radiation dose assessment.

Purpose

To evaluate the feasibility of a deep learning-based reconstruction algorithm (Clear Infinity, CI) combined with an "ultra triple-low" protocol (60 kVp, 10 ml iodine contrast, 2.5 ml/s) for pulmonary CT angiography (CTPA).

Method

Seventy patients (body mass index (BMI) < 28 kg/m2) were randomised to Group A (100 kVp, 30 ml, 4 ml/s, n = 35) and Group B (60 kVp, 10 ml, 2.5 ml/s, n = 35). Group B was subdivided into B1 (filtered back-projection (FBP)), B2 (hybrid iterative reconstruction (HIR), 50 % strength), and B3 (CI algorithm, 50 % strength). CT values, standard deviation (SD), signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR), artifact index, and subjective scores (5-point Likert) were compared. Inter-observer agreement was assessed using Kappa analysis.

Results

Group B reduced effective dose by 84.93 % (0.71 ± 0.09 vs. 4.71 ± 1.13 mSv, P < 0.001) and contrast volume by 66.7 % (10 ml vs. 30 ml). In Group B3, all pulmonary artery CT values exceeded 250 Hounsfield units (HU), and SNR, CNR, and subjective scores were superior (P < 0.001). Subjective artifact scores were also higher in Group B3 (P < 0.001). Inter-observer agreement was strong (κ = 0.87-0.92).

Conclusion

In patients with BMI < 28 kg/m2, the CI algorithm (50 % strength) with the "ultra triple-low" protocol provides diagnostic image quality while reducing radiation and contrast doses, showing potential as a safer approach for high-risk populations, warranting further validation.