# Pilot clinical evaluation of artificial intelligence–driven guiding catheter simulation for optimizing percutaneous coronary intervention

## TL;DR

An AI system that helps choose the right guiding catheter before heart procedures was found to shorten operation times and reduce risks compared to traditional methods.

## Contribution

This study introduces an AI-driven simulation using CT scans to optimize guiding catheter selection in PCI, improving efficiency and safety.

## Key findings

- AI-assisted procedures had shorter total procedure times (68.5 vs. 91.8 minutes) and reduced radiation and contrast use.
- AI guidance led to fewer catheter exchanges and catheter-related events (3.6% vs. 16.4%).
- Procedural success was 100% in both groups with no major adverse events.

## Abstract

In percutaneous coronary intervention (PCI), a suboptimal choice of guiding catheter may compromise coaxial alignment and backup support, prolonging procedures and increasing radiation and contrast exposure. We assessed whether a computed tomography (CT)–driven, artificial intelligence (AI)–guided preprocedural simulation could improve procedural efficiency and safety.

In a single-centre prospective registry with historical controls, 55 consecutive elective procedures performed with CT-based AI-assisted guiding-catheter selection were compared with 55 procedures performed without assistance. The primary endpoint was total procedure time from arterial access to completion. Secondary endpoints included time to coronary engagement, radiation dose, contrast volume, and guiding-catheter-related events. Computed tomography–-based AI assistance was associated with shorter procedures (mean 68.5 vs. 91.8 min), shorter engagement time, lower radiation dose, and lower contrast use. Guiding-catheter exchanges were fewer, and catheter-related events were lower (3.6 vs. 16.4%; risk ratio 0.22; 95% confidence interval 0.05–0.98). Procedural success was 100% in both groups with no in-hospital major adverse cardiac or cerebrovascular events.

A CT-driven, CT-based AI-guided simulation for guiding-catheter selection was associated with greater procedural efficiency and a favourable profile in elective PCI. This approach, which standardizes catheter choice and is associated with fewer empirical catheter exchanges, warrants confirmation in multicentre randomized studies and may help optimize resource utilization in routine PCI.

Graphical AbstractAI-guided vs. conventional guiding catheter selection in PCI. Visual summary comparing conventional planning vs. an AI-guided, CT-driven preprocedural simulation for guiding-catheter selection. In conventional practice, catheter choice is based on operator judgement and may require multiple exchanges; the AI system integrates coronary CT to simulate engagement and recommend an initial guiding catheter. In this prospective registry, AI assistance was associated with shorter procedures and engagement times, lower radiation dose and contrast volume, fewer catheter exchanges, and fewer catheter-related events, while maintaining 100% procedural success. Numerical estimates are provided in the main text and tables. AI, artificial intelligence; CT, computed tomography; GC, guiding catheter.Study overview and key outcomes of CT-based AI guiding-catheter simulation. Diagram contrasting ‘No simulation’ (N = 55) vs. ‘AI simulation’ (N = 55). Top panels show conventional selection vs. CT-to-AI pipeline; bottom summary charts show that, in the AI-simulation group compared with the non-simulation group, total procedure time was 68.5 vs. 91.8 min (P = 0.02), engagement time 97.0 vs. 159.8 s (P < 0.01), GC exchange 1.8% vs. 10.9% (RR 0.17), device-delivery difficulty 16.4% vs. 34.6% (RR 0.47), and guide-extension use 10.9% vs. 27.3% (RR 0.40).

AI-guided vs. conventional guiding catheter selection in PCI. Visual summary comparing conventional planning vs. an AI-guided, CT-driven preprocedural simulation for guiding-catheter selection. In conventional practice, catheter choice is based on operator judgement and may require multiple exchanges; the AI system integrates coronary CT to simulate engagement and recommend an initial guiding catheter. In this prospective registry, AI assistance was associated with shorter procedures and engagement times, lower radiation dose and contrast volume, fewer catheter exchanges, and fewer catheter-related events, while maintaining 100% procedural success. Numerical estimates are provided in the main text and tables. AI, artificial intelligence; CT, computed tomography; GC, guiding catheter.