Real-Time AI Safety Monitoring for Cardiac Catheterization

PI-Technion, Prof. Ron Kimmel,
PI-Rambam, Prof. Yair Feld, MD

The Clinical Problem

During coronary angiography and intervention, procedures performed on hundreds of thousands of patients annually, rare but catastrophic events such as deep catheter intubation, vessel roofing, and distal wire perforation can occur in seconds. These hazards depend entirely on operator vigilance for detection, with no automated monitoring systems available. Even experienced cardiologists can miss critical warning signs during complex, high-stress procedures, potentially leading to vessel dissection, perforation, or other life-threatening complications.

AI-Driven Solution

We developed Cath-Alert, a real-time artificial intelligence system that continuously analyzes fluoroscopic video during cardiac catheterization procedures.

Cathalert

Using deep learning-based computer vision, the system automatically segments catheters, guidewires, and blood vessels in real-time, then identifies hazardous configurations before they result in complications. The AI acts as an intelligent “co-pilot” in the catheterization laboratory, augmenting human expertise with automated vigilance.

Research Goals

  • Create the first comprehensive annotated dataset of interventional fluoroscopy (>10,000 procedures)
  • Develop real-time AI models capable of processing 30+ frames per second
  • Achieve clinically meaningful detection accuracy across diverse procedural scenarios
  • Validate system performance across multiple medical centers
  • Demonstrate the feasibility of seamless integration into existing cath lab workflows

Research Achievements

  • Large-scale dataset: Annotated fluoroscopy data from approximately 10,000 coronary procedures, representing one of the largest such datasets globally
  • High detection accuracy: 78-86% accuracy in identifying procedural hazards
  • Real-time performance: 20-30ms processing time per frame, enabling immediate alerts
  • Multi-center validation: Successful prospective deployment at two independent medical centers

Clinical translation

Co-founded CATHALERT Ltd. (Technion-Rambam joint venture) to bring the technology to market.
Cathalert site