Author: Em Harkness, Francesco Ria, Ehsan Samei 👨🔬
Affiliation: Duke University, Clinical Imaging Physics Group, Department of Radiology, Duke University Health System 🌍
Purpose:
A recently introduced mathematical method quantifies performance-based clinical risk to create a risk-to-risk assessment with radiation risk, rendering the so-called total risk. However, the uncertainties associated with this framework were not defined. In this study, we developed a method to calculate such uncertainties to gauge the total risk potential to inform optimization and justification of CT procedures. The calculated uncertainties were associated with patient age, race, and sex, providing insight into different demographics.
Methods:
The study considered 100,000 digital twins based on census demographic data, (age: 0–99-year-old; male, female; Asian, Black, Hispanic, and white). The clinical risk (Rc) and radiation risk (Rr) calculations were done based on disease prevalence, organ dose-based radiation risk index, 5-year survival rate, false-positive to true-positive ratio, expected life expectancy loss associated with an incorrect diagnosis, and radiologist interpretive performance. The organ dose conversion coefficients and the NIH-SEER 5-year survival rates were used to calculate the risk uncertainties using the multi-variable error propagation formula, assuming each variable to be independent of the others. The results were sorted based on demographic groups to compare relative uncertainties.
Results:
Across all demographics, the relative uncertainty associated with Rr was 7% (0.0057±4×10-4 mortality per 100 patients) and Rc relative uncertainty was 8% (0.052±0.004 mortality per 100 patients). The demographics with the lowest Rr relative error were Asian females, Asian males, and Hispanic males at 6%; whereas the highest Rr was 8% in white males. The lowest Rc was associated with white males at 4%; whereas the highest Rc was reported for Black females at 14%.
Conclusion:
We established a robust method to evaluate the uncertainties related to clinical and radiation risk calculations in CT. The results showed the total risk potential to inform optimization and justification of imaging procedures with different stratification for patient demographics.