Model-based iterative reconstruction in ultra-low-dose pediatric chest CT: comparison with adaptive statistical iterative reconstruction.

PURPOSE: To evaluate image quality and dose reduction of ultra-low-dose pediatric chest CT reconstructed with model-based iterative reconstruction (MBIR), as compared with adaptive statistical iterative reconstruction (ASIR).

MATERIALS AND METHODS: Fifty-seven patients (mean age 14 years, M:F=31:26) who underwent ultra-low-dose chest CT reconstructed with both MBIR and ASIR were enrolled in the study. The subjective and objective image qualities of both reconstruction techniques were assessed by 3 radiologists, and compared using statistical analysis. We also evaluated radiation dose of ultra-low-dose chest CT as well as degree of dose reduction in comparison to the prior CT (either standard dose or reduced dose protocol) available in 36 patients.

RESULTS: The image quality of MBIR was superior to ASIR both subjectively and objectively. While MBIR showed preserved diagnostic acceptability in 100%, ASIR showed 92% at mean 0.31 mSv (range, 0.13-0.57 mSv) ultra-low-dose CT. In the 36 patients who underwent the prior CT, mean decrease in size-specific dose estimate (SSDE) and dose length product (DLP) at ultra-low-dose CT was 88% (range, 34% - 98%) and 86% (range,42% - 99%), respectively.

CONCLUSIONS: MBIR significantly improves image quality, as compared to ASIR. Furthermore, MBIR facilitates diagnostically acceptable ultra-low-dose chest CT with nearly 90% less radiation.