Quantitative coronary calcification analysis for prediction of myocardial ischemia using non-contrast CT calcium scoring

2026-05-23 · Source: cs.LG updates on arXiv.org · Field: Health & Wellbeing — Health & Medical Research, Medical Devices & Health Technology · Depth: Expert, quick

Summary

A novel machine learning framework predicts myocardial ischemia using quantitative coronary calcium assessment from routine non-contrast CT calcium scoring (CTCS) scans. The study analyzed 1,375 patients from University Hospitals Cleveland Medical Center who underwent both CTCS and regadenoson stress cardiac PET myocardial perfusion imaging. Researchers evaluated 74 variables, including clinical data, Agatston score, and calcium-omics features, identifying relevant features using XGBoost with Shapley Additive exPlanations (SHAP). Among 987 patients, 89 (9%) were positive for myocardial ischemia. The final model, incorporating the Agatston score, eight calcium-omics features, and age, achieved a precision of 98.9+/-3.0%, sensitivity of 79.2+/-8.4%, and an F1 score of 87.7+/-5.3%. The addition of calcium-omics features significantly improved predictive performance (p<0.05) compared to models using only clinical variables or clinical variables with the Agatston score.

Key takeaway

For cardiologists and radiologists assessing cardiovascular risk, integrating quantitative calcium-omics features from non-contrast CTCS scans can significantly enhance myocardial ischemia prediction. You should consider adopting machine learning models that incorporate these advanced features, as they offer improved precision and sensitivity over traditional Agatston scoring alone. This approach supports more accessible and accurate cardiovascular risk stratification, potentially guiding earlier intervention decisions.

Key insights

Quantitative calcium-omics features from CTCS scans enhance myocardial ischemia prediction beyond traditional risk factors.

Principles

• Calcium-omics features improve ischemia prediction.
• XGBoost and SHAP identify key predictive features.
• Non-contrast CTCS offers accessible risk stratification.

Method

A machine learning framework uses XGBoost with SHAP for feature selection from 74 variables, then trains predictive models with 5-fold cross-validation on CTCS and clinical data.

In practice

• Integrate calcium-omics into existing CTCS workflows.
• Use XGBoost/SHAP for feature engineering in medical ML.
• Consider non-contrast CTCS for broader risk assessment.

Topics

• Myocardial Ischemia Prediction
• Coronary Artery Calcification
• CT Calcium Scoring
• Calcium-omics
• Machine Learning
• XGBoost
• Cardiovascular Risk Stratification

Best for: AI Scientist, Research Scientist, Data Scientist

Related on AIssential

• Beyond the Blood Draw: Explainable Machine Learning for Non-Invasive Dysglycemia Risk Screening — Explainable ML models can non-invasively screen for dysglycemia risk using readily available data, outperforming existing clinical scores.
• Promise and challenges of heart chamber segmentation from non-contrast CT scans using contrastive unpaired image translation: a feasibility study — Unpaired image translation can enable deep learning segmentation of heart chambers from non-contrast CT without direct non-contrast annotations.
• Time-Conditioned and Multi-Time Survival Prediction from 2D PET/CT Projections in Lung Cancer — Temporal modeling significantly enhances PET/CT-based overall survival prediction in NSCLC patients.
• A Patient-Specific Pulmonary Arterial Tree Digital Twin to Extract Pulmonary Embolism Biomarkers — Automated digital twins of pulmonary arteries can rapidly quantify PE severity and clot distribution.
• A Multimodal and Explainable Machine Learning Approach to Diagnosing Multi-Class Ejection Fraction from Electrocardiograms — Multimodal machine learning can accurately stratify LVEF from ECG and EHR data, enhancing accessibility.
• Label-free multiphoton microscopy and machine learning for recognition of hepatocellular carcinoma — Label-free multiphoton microscopy combined with machine learning accurately identifies hepatocellular carcinoma.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by cs.LG updates on arXiv.org.