Application of deep learning-based precise image reconstruction algorithm in non-contrast abdominal CT scanning

Background: Reducing radiation dose without compromising diagnostic image quality remains a central challenge in non-contrast abdominal computed tomography (CT). This study aimed to evaluate the clinical value of a deep learning reconstruction algorithm, Philips-developed precise image (PI), in achieving this balance.

Methods: This retrospective study included 120 patients who underwent non-contrast abdominal CT scans and were randomly divided into three dose groups based on the DoseRight Index (DRI): Group A (DRI =18), Group B (DRI =20), and Group C (DRI =22). For each patient, raw data were reconstructed using filtered back projection (FBP), iDose4, PI standard, and PI smooth. Objective parameters—including CT value, standard deviation (SD), signal-to-noise ratio (SNR), and contrast-to-noise ratio (CNR)—were measured for the liver, pancreas, and abdominal aorta. Subjective image quality and radiation dose metrics were also analyzed.

Results: The PI smooth algorithm consistently yielded the best performance, with significantly lower noise and higher SNR and CNR than other reconstructions across all organs and dose groups (all P<0.05). In inter-group comparisons using PI smooth, Group B (DRI =20) demonstrated noise levels and SNR comparable to those of the higher- or lower-dose groups in key organs, and subjective image quality scores of PI smooth images were comparable between Group B and the other groups, with only a small but statistically significant difference between Group A and Group C (P=0.03).

Conclusions: The combination of a low-dose protocol (DRI =20) and the PI smooth reconstruction algorithm significantly reduces radiation exposure while maintaining both objective and subjective image quality comparable to conventional higher-dose scans. This approach offers a practical and effective strategy for implementing radiation dose optimization in routine abdominal CT imaging.