Idea
AI model delivering accurate, multi-view coronary angiography analysis for improved diagnosis and prognosis in cardiovascular care.
Research Paper
Core Innovation
This paper introduces DeepCORO-CLIP, a multi-view foundation model trained on a large-scale angiography video-text dataset, integrating multiple projections with attention-based pooling for study-level coronary assessment. It surpasses prior single-frame or single-projection AI methods by enabling comprehensive detection of stenosis, thrombus, calcification, and disease progression with external validation and clinical deployment readiness.
Why It Matters
Coronary angiography interpretation is variable and time-consuming, limiting consistent diagnosis and treatment planning. DeepCORO-CLIP automates comprehensive coronary assessment with high accuracy and fast hospital deployment, reducing diagnostic variability and enabling scalable, data-driven cardiovascular care. This improves patient outcomes and streamlines clinical workflows across healthcare systems.
Market Size (TAM)
$2–10B TAM for AI cardiovascular imaging; $500M–$1B SAM from hospitals and imaging centers. Driven by rising cardiovascular disease burden and demand for automated diagnostic tools.
Potential Customers & Pain Points
- Hospitals – Need faster consistent coronary diagnosis
- Cardiologists – Require comprehensive multi-view analysis
- Medical imaging companies – Demand scalable AI integration
- Healthcare systems – Seek improved cardiovascular outcome prediction
Business Model
Subscription-based SaaS platform licensing AI coronary angiography analysis to hospitals and imaging centers, with options for on-premise deployment and integration support.
Competitive Landscape
- HeartFlow
- Caption Health
- Ultromics
- Cleerly
Implementation Challenges
- Regulatory approval for clinical AI tools
- Integration with diverse hospital imaging systems
- Clinician trust and adoption of AI interpretations
- Data privacy and security compliance
Validation Strategy
- Conduct multi-center clinical trials to confirm diagnostic accuracy and impact on patient outcomes
- Obtain regulatory clearances (FDA
- CE) for clinical use
- Pilot deployments in partner hospitals to refine workflow integration
- Collect real-world usage data to improve model performance and user experience
Research Paper Overview
DeepCORO-CLIP: A Multi-View Foundation Model for Comprehensive Coronary Angiography Video-Text Analysis and External Validation
Summary
DeepCORO-CLIP is a multi-view AI model trained on over 200,000 coronary angiography videos integrating multiple projections for study-level assessment. It detects significant stenosis and other coronary conditions with high accuracy, outperforms clinical reports, predicts major adverse cardiovascular events, and estimates cardiac function. The model is externally validated and deployable in hospitals with fast inference times, supporting automated coronary angiography interpretation at point of care.