Author: Gary Y. Ge, Charles Mike Weaver, Jie Zhang 👨🔬
Affiliation: University of Kentucky 🌍
Purpose: The recently introduced CMS regulation on CT dose and image quality mandates the use of global noise as the metric for image quality assessment. The regulation cites two methods for calculating global noise, the ‘Wisconsin method’ and ‘Duke method,’ without prescribing which method or metric to use in reporting. This lack of specificity raises concerns about the consistency of these methods in evaluating image quality. This study compares these methods to evaluate their consistency across protocols.
Methods: Global noise was calculated using five metrics derived from the two methods. The ‘Wisconsin method’ measures the mean and mode of ROIs in air and tissue, while the ‘Duke method’ calculates the mode of ROIs in tissue. Four CT procedure types were selected for analysis based on CMS dose categories: Renal Stone (low dose), Abdomen/Pelvis (routine dose), Urogram (high dose), and Chest (routine dose). CT images from 100 patients were retrospectively retrieved for each procedure. Correlation maps assessed relationships between metrics across protocols, Kruskal-Wallis tests determined differences between metrics, and permutation tests of slice-based calculations compared tissue metrics for each patient.
Results: Correlation analysis showed variability across CT protocols, with strong agreement (r>0.7) among all metrics for Renal Stone, Chest, and Abdomen/Pelvis protocols, but weaker correlations for Urogram protocols, with r≤0.45 in 4/10 comparisons. Correlations between mode metrics were the most robust, showing very strong agreement (r>0.9) across all protocols. Significant differences (p<0.05) were observed in all metrics, with wis_tissue_mean consistently showing the highest global noise values across 4 tested protocols. Permutation tests further indicated significant differences (p<0.05) in tissue metrics for all but 4/400 patients.
Conclusion: The inconsistencies among global noise metrics imply that the CMS goal of managing CT radiation dose could be undermined without specifying an image quality metric, particularly when comparing performance across medical centers reporting different metrics.