Conformal Coronary Calcification Volume Estimation with Conditional Coverage via Histogram Clustering

2025-04-14

IEEE International Symposium on Biomedical Imaging

AI Summary

This study addresses the clinical problem of over-reporting coronary artery calcification (CAC) quantification in CT imaging, which leads to misdiagnosis and inefficient resource utilization. We propose the first histogram-clustering-based conditional conformal prediction framework for uncertainty calibration—applicable across diverse 3D U-Net architectures (deterministic, Monte Carlo Dropout, and deep ensembles) without model retraining or fine-tuning. The method generates clinically interpretable confidence intervals for CAC volume, precisely delineating risk-stratification thresholds (e.g., Agatston score ranges), significantly enhancing discriminability of confidence estimates across risk categories while rigorously guaranteeing nominal coverage. Our key innovation lies in integrating histogram clustering into conditional conformal prediction, enabling model-agnostic, calibration-free uncertainty quantification. This yields a reliable, generalizable, and clinically actionable tool for trustworthy CAC risk assessment.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Reasoning under Uncertainty: Uncertainty Representations
• Constraint Satisfaction and Optimization: Satisfiability Modulo Theories

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Economics, Online Markets and Human Computation: Trust and reliance of crowd workers and data experts on GenAI

Abstract

Incidental detection and quantification of coronary calcium in CT scans could lead to the early introduction of lifesaving clinical interventions. However, over-reporting could negatively affect patient wellbeing and unnecessarily burden the medical system. Therefore, careful considerations should be taken when automatically reporting coronary calcium scores. A cluster-based conditional conformal prediction framework is proposed to provide score intervals with calibrated coverage from trained segmentation networks without retraining. The proposed method was tuned and used to calibrate predictive intervals for 3D UNet models (deterministic, MCDropout and deep ensemble) reaching similar coverage with better triage metrics compared to conventional conformal prediction. Meaningful predictive intervals of calcium scores could help triage patients according to the confidence of their risk category prediction.

Problem

• Estimating coronary calcium volume with calibrated coverage
• Reducing over-reporting of calcium scores in CT scans
• Improving patient triage via confidence-based risk prediction

Innovation

• Cluster-based conditional conformal prediction framework
• Calibrated coverage without retraining segmentation networks
• Improved triage metrics with meaningful predictive intervals