Carotid Plaque-RADS: Inter- and Intrareader Agreement and Learning Curve Analysis in CTA

Area of Science:

• Radiology
• Vascular Imaging
• Stroke Prevention

Background:

• Carotid artery stenosis is a primary cause of ischemic stroke.
• Plaque composition and morphology are crucial for stroke risk assessment.
• Carotid Plaque-Reporting and Data System (RADS) standardizes plaque evaluation beyond stenosis degree.

Purpose of the Study:

• To evaluate the inter-reader and intra-reader agreement of Carotid Plaque-RADS.
• To assess the learning curve associated with Carotid Plaque-RADS implementation.
• To establish the reliability of Carotid Plaque-RADS in routine radiological practice.

Main Methods:

• Retrospective analysis of 500 computed tomography angiography (CTA) exams for carotid atherosclerosis.
• Three readers with varying experience levels independently assessed carotid arteries using Carotid Plaque-RADS.
• Inter-reader and intra-reader agreement calculated using Cohen's kappa (κ) across multiple assessment blocks.

Main Results:

• Substantial agreement (κ=0.78-0.88) between expert and intermediate readers.
• Moderate agreement (κ=0.50-0.74) between intermediate and beginner readers.
• Significant improvement in expert-beginner agreement across blocks, indicating a learning curve.
• Near-perfect intra-reader agreement (κ=0.88) for intermediate and expert readers.

Conclusions:

• Carotid Plaque-RADS demonstrates substantial to near-perfect inter-reader agreement among experienced and intermediate readers.
• A learning curve is evident for beginners, with agreement improving over time.
• High intra-reader agreement ensures grading consistency and data reproducibility for Carotid Plaque-RADS.