Development of an automated, reliable, and clinically meaningful artificial intelligence (AI) tool for diagnosing cardiac disease from conventional cardiovascular magnetic resonance (CMR) images

2026-07-23 · Source: cs.CV updates on arXiv.org · Field: Health & Wellbeing — Artificial Intelligence & Machine Learning, Health & Medical Research, Medical Devices & Health Technology · Depth: Advanced, quick

Summary

An automated artificial intelligence (AI) tool has been developed for diagnosing cardiac disease from conventional cardiovascular magnetic resonance (CMR) images. This tool addresses challenges like labor-intensive data curation and suboptimal model performance in existing AI models for CMR. It integrates an automated data curation pipeline using open-source large language models (LLMs) to extract diagnostic labels from narrative CMR reports and preprocesses multimodal imaging data, including cine and late-gadolinium-enhancement (LGE) CMR sequences. Three vision foundation models (DINO, VST, UMedPT) were fine-tuned on a dataset of 988 curated cases covering hypertrophic cardiomyopathy (HCM), dilated cardiomyopathy (DCM), ischemic cardiomyopathy (ICM), cardiac amyloidosis (CA), and normal controls (NOR). On an independent test set of 1067 patients, ensemble strategies achieved high diagnostic performance, with AUCs of 0.959 for HCM, 0.966 for CA, 0.872 for NOR, 0.848 for DCM, and 0.840 for ICM. All training and inference code, along with trained model weights, are publicly available.

Key takeaway

For clinical cardiologists and AI developers aiming to improve cardiac disease diagnosis, this automated AI tool offers a robust solution. Its integration of LLMs for data curation and ensemble vision models significantly enhances diagnostic precision and efficiency. You should consider exploring such AI-driven pipelines to streamline clinical workflows, reduce manual interpretation burden, and potentially improve patient outcomes in cardiovascular care.

Key insights

Automated AI, leveraging LLMs and vision models, significantly enhances cardiac disease diagnosis from CMR images.

Principles

• Automated data curation is critical for medical AI development.
• Ensemble strategies improve AI diagnostic accuracy and robustness.
• Multimodal imaging data enhances diagnostic performance.

Method

An automated pipeline uses LLMs for diagnostic label extraction from CMR reports and preprocesses cine/LGE CMR. Vision foundation models (DINO, VST, UMedPT) are fine-tuned in a two-stage approach, followed by ensemble strategies.

In practice

• Integrate LLMs for automated medical report data extraction.
• Combine cine and LGE CMR sequences for diagnosis.
• Utilize ensemble methods for robust AI diagnostics.

Topics

• Cardiovascular Magnetic Resonance
• Cardiac Disease Diagnosis
• Artificial Intelligence
• Large Language Models
• Medical Imaging
• Vision Foundation Models

Code references

• sinaamirrajab/CMR_CVD

Best for: Computer Vision Engineer, AI Scientist, Research Scientist, Domain Expert

Related on AIssential

• v333: Proceedings of CHIL 2026 — Machine learning, particularly LLMs, is rapidly expanding its utility across diverse clinical and administrative healthcare applications.
• Scientists develop AI tool to spot heart failure risk five years before it strikes — An AI tool predicts heart failure risk five years early by analyzing cardiac CT scans for invisible fat inflammation.
• Trusted Multi-View Deep Learning Classification of Fetal Congenital Heart Disease with Feature-level and Decision-level Fusion — The framework integrates multi-view echocardiographic data with uncertainty handling for robust fetal CHD classification.
• How AI Ultrasound Systems Are Improving Real-Time Diagnostic Accuracy in Clinical Settings — AI ultrasound systems enhance diagnostic accuracy and expand access by automating measurements, guiding image acquisition, and accelerating interpretation.
• Domain Knowledge Based Temporal-Spatial Graph Convolution Network for ECG Recognition — Integrating domain knowledge via graph networks significantly improves ECG recognition, especially for rare conditions.
• Learning neuroimaging models from health system-scale data — Health system-scale data enables AI foundation models to achieve superior, fair, and explainable neuroimaging diagnostics.

See Counsel's argued verdicts on the open AI decisions leaders are weighing →

Open in AIssential →

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