To explore the feasibility of applying deep learning image reconstruction (DLIR) algorithm combined with low-dose contrast agent in abdominal enhanced CT Method 28 patients who needed upper abdominal enhanced CT scan due to their conditions were randomly divided into two groups, with 64 cases in each group. They underwent spectral CT abdominal enhanced examination with consistent scanning parameters. The conventional dose group was scanned with a contrast agent dosage of 1.5 ml/kg and reconstructed using the adaptive statistical iterative reconstruction (ASIR) algorithm, with an ASIR iteration rate of 60%. The low-dose group was scanned with a contrast agent dosage of 1.0 ml/kg and reconstructed using different levels of DLIR algorithms, including group A with medium-intensity deep learning image reconstruction (DLIR-M) and group B with high-intensity deep learning image reconstruction (DLIR-H).

evaluation indicators [noise, contrast-to-noise ratio (CNR), signal-to-noise ratio (SNR)] and subjective evaluation indicators (noise, overall image quality) were compared and analyzed among images reconstructed by different algorithms.

In objective evaluation, the noise of images in the conventional dose group, low-dose group A, and low-dose group B gradually decreased, and the differences between each pair of groups were statistically significant (P<0.05). The SNR of liver, SNR of pancreas, CNR of liver, and CNR of pancreas in the conventional dose group, low-dose group A, and low-dose group B gradually increased. There was no statistically significant difference between the conventional dose group and low-dose group A, but the differences among the other groups were statistically significant. In subjective evaluation, the subjective noise and overall image quality scores of the conventional dose group, low-dose group A, and low-dose group B were mostly above 4 points, which could meet the clinical diagnosis and treatment needs. The subjective noise scores gradually increased, and the differences between each pair of groups were statistically significant (P<0.05). In terms of subjective image quality scores, the score of low-dose group B was higher than that of low-dose group A and the conventional dose group, and the difference was statistically significant (P<0.05).

The image quality obtained by using DLIR algorithm combined with low-dose contrast agent is superior to that of ASIR algorithm combined with conventional dose contrast agent, and it can meet the needs of daily diagnosis. At the same time, it can reduce the required contrast agent dosage by about one-third.