# Visual coronary calcium scoring to support opportunistic CAD screening: comparative evaluation of three established systems and introduction of a novel scoring system

## TL;DR

This study compares different visual methods for scoring coronary artery calcium and introduces a new scoring system that balances accuracy and practicality for cardiovascular risk assessment.

## Contribution

The paper introduces a novel scoring system, the Weston Extent Score (WES), which improves diagnostic accuracy by combining two existing visual scoring methods.

## Key findings

- All visual scoring methods showed strong correlation with the Agatston score (Spearman ρ > 0.87).
- Visual scoring had the highest inter-observer agreement (Kappa = 0.94, ICC = 0.97).
- The new WES score achieved 80.9% accuracy and 93.8% macro-averaged specificity.

## Abstract

Coronary artery calcium (CAC) scoring is an established marker of atherosclerotic burden and cardiovascular risk. While the Agatston score is the clinical gold standard, alternative visual scoring methods—including the Visual Ordinal Score, Weston Score, and Vessel-specific extent-based score—are increasingly used, particularly in non-gated or opportunistic CT imaging. This study aimed to compare the diagnostic performance, inter-observer reliability, and correlation of different visual scoring methods against the Agatston score.

A total of 299 cases were evaluated using ECG-gated CT scans. Each case was independently scored in a blinded fashion by two observers using three visual methods: (1) Visual Ordinal Score (VS), (2) Weston Score (WS) and (3) Vessel-specific extent-based score (VSES). A novel visual CAC score was derived by combining Weston and Vessel-specific extent-based scoring (= Weston Extent Score, WES). Cohen’s Kappa and Intraclass Correlation Coefficients (ICC) were used for inter-observer agreement. Classification performance was assessed against Agatston-based categories (No CAC, Mild, Moderate, Severe), including accuracy, precision, sensitivity, and specificity. Correlation analyses were conducted using Pearson and Spearman coefficients.

All scoring methods showed high correlation with the Agatston score (Spearman ρ > 0.87; p < 0.001). Visual scoring demonstrated the highest inter-observer agreement (Kappa = 0.94, ICC = 0.97), followed by Weston (Kappa = 0.90) and Vessel-Specific scores (Kappa = 0.77). Visual scoring also yielded the highest accuracy (Observer 1: 91.3 %, Observer 2: 90.0 %) The newly derived WES score achieved 80.9 % accuracy, with macro-averaged specificity of 93.8 % and improving diagnostic accuracy compared to WS and VSES.

Different visual scoring offers excellent reproducibility and diagnostic accuracy for CAC classification, with strong correlation to the Agatston score. The newly-derived WES score could be useful in providing a practical balance regarding volumetric information (CAC densitiy) and anatomical distribution of CAC. These findings support the implementation of structured visual CAC scoring in clinical and opportunistic CT settings.