Scanner-Specific Variability In Neck CT Image Quality and Dose Efficiency: A Multicenter Quantitative Analysis
Abstract
Purpose
To evaluate neck CT image quality (IQ) and dose efficiency for lesion assessment across multiple scanners, identifying scanner-level differences and protocol optimization opportunities.
Methods
This retrospective study included 957 contrast-enhanced CT neck soft tissue exams from 12 scanners across 8 hospitals over 6 months. IQ was objectively assessed using signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR), then subjectively via a blinded reader study performed by a radiologist who rated IQ-perceptibility on a 5-point scale. The ACR’s Achievable Dose (AD) (15 mGy) to Diagnostic Reference Levels (DRL) (19 mGy) coupled with the Rose Criterion was used as a standardized reference. Dose and size metrics included CTDIvol (3.94-34.61 mGy), and water-equivalent diameter (WED) (18-59 cm). Regression and ANCOVA assessed relationships between IQ, dose, and scanner type.
Results
CTDIvol showed no significant correlation with SNR or CNR (p>0.05), indicating that increasing radiation dose did not yield meaningful improvements in objective noise or contrast performance. CTDIvol also demonstrated no meaningful association with radiologist perceived IQ rating (R²≈0.03, p>0.05), and subjective perceptibility did not significantly correlate with SNR or CNR (p>0.05). However, regression analysis indicates that scanner type was the primary determinant of IQ. CTDIvol varied 5-fold across scanners (4.6-22.5 mGy), but radiologist ratings varied only 1.5-fold (3.0-4.6), with high-dose scanners (>18 mGy) achieving 4.30±0.14 versus 3.40±0.32 for low-dose scanners (<10 mGy). Correlations between ratings and individual metrics were weak to moderate (CTDI: r=0.744, SNR: r=0.438, CNR: r=0.218).
Conclusion
Our findings support re-evaluation of institutional protocols to reduce unnecessary radiation and promote scanner-specific dose standardization to achieve consistent IQ across scanner types.