Variation in aorta attenuation in contrast-enhanced CT and its implications for calcification thresholds

Area of Science:

• Radiology
• Medical Imaging
• Cardiovascular Imaging

Background:

• Computed tomography (CT) contrast media enhances vessel visualization but can interfere with accurate calcification measurements.
• Aorta attenuation (AoHU) varies with contrast enhancement, potentially leading to errors in plaque detection algorithms.

Purpose of the Study:

• To evaluate the variance in aorta attenuation across different contrast-enhanced CT scan phases.
• To quantify the expected plaque detection errors using uniform and dynamic thresholding methods.
• To assess the impact of demographic and scan variables on aorta attenuation.

Main Methods:

• Analyzed aorta attenuation (AoHU) in 10,000 abdominal CT scans across non-contrast, arterial, venous, and delayed phases.
• Correlated AoHU with demographic variables (age, sex, weight) and scan parameters (body location, slice thickness).
• Quantified plaque segmentation errors (false-negatives and false-positives) using uniform and dynamic thresholding approaches.

Main Results:

• Females exhibited higher AoHU than males in contrast-enhanced scans (p < 0.001).
• Weight showed a negative correlation with AoHU, but other factors explained minimal variance (R2 < 0.1).
• Uniform thresholds led to substantial false-negative and false-positive errors, while dynamic segmentation (3SD above mean AoHU) significantly reduced these errors.

Conclusions:

• CT contrast causes heterogeneous aortic enhancement, not easily predicted by patient or scan factors.
• Uniform thresholds for calcified plaque detection in contrast-enhanced CTs result in high classification errors.
• Dynamic thresholding based on local aorta attenuation minimizes errors and addresses sex-based biases, crucial for developing robust automated detection algorithms.