Estimating the differences between inter-operator contrast enhancement in cerebral CT angiography.

Area of Science:

• Radiology and Medical Imaging
• Neuroimaging
• Statistical Modeling

Background:

• Computed tomography angiography (CTA) is crucial for detecting arteries and diagnosing brain diseases.
• Reproducibility in vessel delineation is essential for follow-up and postoperative CTA evaluations.
• While factors influencing arterial contrast enhancement are studied, operator-specific effects remain unexplored.

Purpose of the Study:

• To evaluate inter-operator variability in arterial contrast enhancement during cerebral CTA.
• To utilize Bayesian statistical modeling for assessing these differences.

Main Methods:

• Bayesian statistical models were developed using cerebral CTA data from 2015-2018.
• Mean CT numbers of internal carotid arteries were the objective variable, with operator, sex, age, and fractional dose as explanatory variables.
• Markov chain Monte Carlo (MCMC) methods, including Hamiltonian Monte Carlo, were used for Bayesian inference.

Main Results:

• Bayesian analysis revealed that differences between operators fell within 95% credible intervals, including zero.
• The maximum mean difference in CT numbers between operators was a modest 12.59 Hounsfield units (HUs).

Conclusions:

• Operator-to-operator differences in postcontrast CT numbers for cerebral CTA are statistically insignificant.
• Within-operator variability, influenced by unmodeled factors, likely exceeds inter-operator differences.