Encoding of Demographic and Anatomical Information in Chest X-Ray-based Severe Left Ventricular Hypertrophy Classifiers

2025-05-31

AI Summary

Left ventricular hypertrophy (LVH) screening relies heavily on expensive imaging modalities (e.g., echocardiography, MRI), limiting accessibility in primary care settings. Method: We propose an end-to-end deep learning classifier that uses only routine chest X-ray (CXR) images—without requiring anatomical measurements or demographic variables—to detect severe LVH. Crucially, our framework implicitly disentangles and quantifies demographic and anatomical feature representations via mutual information neural estimation (MINE), enhancing model interpretability and revealing intrinsic discriminative mechanisms. Contribution/Results: Evaluated on an independent test set, the model achieves high diagnostic performance (AUROC = 0.92, AUPRC = 0.78), demonstrating that standard CXRs contain sufficient discriminative information for severe LVH identification. This work establishes a low-cost, deployable paradigm for early cardiac disease screening, bridging clinical utility with algorithmic transparency.

Technology Category

Computer Vision: Interpretability, Explainability, and TransparencyMachine Learning: Multi-class/Multi-label Learning & Extreme ClassificationKnowledge Representation and Reasoning: Diagnosis and Abductive Reasoning

Application Category

Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for rankingWeb Mining and Content Analysis: Large pretrained models with web dataEconomics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets

Abstract

While echocardiography and MRI are clinical standards for evaluating cardiac structure, their use is limited by cost and accessibility.We introduce a direct classification framework that predicts severe left ventricular hypertrophy from chest X-rays, without relying on anatomical measurements or demographic inputs. Our approach achieves high AUROC and AUPRC, and employs Mutual Information Neural Estimation to quantify feature expressivity. This reveals clinically meaningful attribute encoding and supports transparent model interpretation.

Problem

• Predicts severe left ventricular hypertrophy from chest X-rays
• Avoids reliance on costly echocardiography or MRI
• Uses Mutual Information to ensure transparent model interpretation

Innovation

• Direct classification from chest X-rays
• Uses Mutual Information Neural Estimation
• Achieves high AUROC and AUPRC