Related Experiment Video
Updated: May 14, 2026

Identifying Coronary Artery Calcification on Non-gated Computed Tomography Scans
Published on: August 28, 2018
High scan-rescan repeatability of AI-enabled coronary plaque quantification from coronary CT angiography
Joel Lenell1, Caroline Park1, Jacek Kwiecinski1,2
1Biomedical Imaging Research Institute, Departments of Biomedical Sciences and Medicine, Cedars Sinai Medical Center, Los Angeles, CA, USA.
Serial cardiac computed tomography (CCTA) scans analyzed side-by-side using AI software demonstrate high reproducibility for coronary plaque quantification. Scan-specific thresholds offer greater consistency than fixed thresholds for assessing plaque burden over time.
Area of Science:
- Cardiovascular Imaging
- Artificial Intelligence in Medicine
- Radiology
Background:
- Coronary artery disease (CAD) assessment relies on accurate quantification of coronary plaque.
- Serial cardiac computed tomography angiography (CCTA) allows for monitoring plaque progression, but reproducibility is crucial.
- AI-enabled software offers potential for standardized and reproducible plaque analysis.
Purpose of the Study:
- To evaluate the scan-rescan reproducibility of coronary plaque quantification using a systematic side-by-side approach.
- To compare the consistency of AI-enabled plaque quantification with scan-specific versus fixed attenuation thresholds.
- To assess the reliability of quantifying total plaque (TP), calcified plaque (CP), non-calcified plaque (NCP), and low-density NCP (LD-NCP).
Main Methods:
- Thirty participants underwent serial CCTA scans within a median of 6 days.
- AI-enabled software quantified plaque volumes and burden (plaque volume indexed by vessel volume) side-by-side.
- Two assessors performed consensus readings, and inter-scan differences were analyzed using mean differences and repeatability coefficients (RC).
Main Results:
- Excellent intraclass correlation coefficients (≥0.96) were observed for all plaque metrics.
- Scan-specific thresholds demonstrated lower repeatability coefficients and narrower limits of agreement compared to fixed thresholds.
- High reproducibility was found for TP, CP, NCP, and LD-NCP volumes and burdens, with specific RC values reported for each.
Conclusions:
- Systematic side-by-side assessment with AI-enabled software provides highly reproducible coronary plaque quantification from serial CCTA scans.
- Scan-specific attenuation thresholds enhance consistency and reproducibility compared to fixed thresholds for plaque analysis.
- These findings support the reliable clinical use of serial CCTA for monitoring coronary plaque changes over time.
Related Concept Videos
Imaging Studies for Cardiovascular System VI: Calcium -Scoring CT
Imaging Studies for Cardiovascular System V: CT
Acute Coronary Syndrome III: Diagnostic Studies
