Although we only used AIDR3D in this study, we believe that similar advantages of IR techniques for more stable measurements will be common among scanners from other manufacturers when considering that all IR methods greatly reduce image noise.12–16,19–21 Stabilizing CT-based emphysema measurements by using IR would be a breakthrough for using CT data in clinical care for COPD and for generating a larger study cohort using different CT scanners. CT-based COPD analysis invariably involves limitations with regard to standardizing measurements between different models and generations of CT scanners,22 which may be avoided or reduced by using IR methods. Longitudinal CT observations of COPD patients for longer periods or consolidating CT data from multiple institutions may also be achieved by using IR.
This study had several limitations. First, the number of patients was relatively small. However, because the entire process, including image generation and quantitative measurements, was completely automatic, we strongly believe that our results were not unduly skewed due to the small sample size. Second, because our aim was not to investigate quantitative airway analysis, the advantages of IR for airway analysis should be evaluated in additional studies.17 Third, we did not clearly demonstrate a direct association between image noise and changes in emphysema indices due to artifacts. However, several previous studies did demonstrate possible correlations between image noise and measurement errors during the quantitative analysis of emphysema.5–9 Fourth, we did not determine whether or not IR provided stronger correlations between emphysema measurements and spirometric values, as spirometry was not performed for this study. This should be investigated in future studies.
In conclusion, IR, represented by AIDR3D in this study, provides consistent measurements of LAA% and 15th percentile results at different tube current settings as compared with those using a conventional FBP method.