Author: Alexander Alsalihi, Gary Y. Ge, Charles Mike Weaver, Jie Zhang 👨🔬
Affiliation: University of Kentucky 🌍
Purpose: The upcoming CMS regulation, titled “Excessive Radiation Dose or Inadequate Image Quality for Diagnostic Computed Tomography (CT) in Adults”, employs two measures, size-adjusted DLP (SAD) and global noise. The guideline proposes one SAD calculation method, but the references and supporting documents suggests multiple methods can be used interchangeably. This study investigates the variability between calculation methods for SAD.
Methods: The SAD equation requires the calculation of patient diameter (d) for each CT exam. Five methods to calculate d were evaluated: HU Thresholding, (LAT+AP)/2, SQRT(LATxAP), LAT Conversion, Water-Equivalent Diameter (WED). Four CT protocols were selected for this study, Renal Stone, Abdomen Pelvis W/O Contrast, Urogram, and Chest W/O Contrast. The first three protocols are low, routine, and high dose abdominal CT exams, respectively, as defined by CMS. Chest W/O Contrast is a routine chest CT exam. Patient images were retrospectively retrieved from our institution. For each of the protocols, 100 patients were collected. Only exams where the scan midslice displayed no anatomical cutoff were included. Kruskal-Wallis analysis was performed to determine differences in both d and SAD for each calculation method.
Results: Kruskal-Wallis analysis shows differences between methods for d and SAD for Renal Stone and Chest W/O Contrast protocols (p<0.05). Further pairwise comparisons show HU Thresholding generates lower SAD than both (AP+LAT)/2 and √(AP × LAT) methods for Renal Stone protocol (p<0.05). The (AP+LAT)/2 and √(AP × LAT) methods generates lower SAD than HU Thresholding and WED methods for the Chest protocol (p<0.05). This results in ~30% lower SAD values in the Chest protocol when using AP/LAT methods over other methods.
Conclusion: Our results highlight the inconsistency in calculation methods across different protocols and anatomical regions. This variability poses a concern for regulatory implementation and benchmarking.