Rapid AI-Enabled Coronary Angiography Guidance Using Real-Time 3D Vessel Reconstruction in Angioplasty

Shanma E, Dhanalakshmi V, Shanthi V, Kaushik Prakash, Divya N, Durga B

doi:10.7492/c6q56n49

Authors

Shanma E, Dhanalakshmi V, Shanthi V, Kaushik Prakash, Divya N, Durga B Author

DOI:

https://doi.org/10.7492/c6q56n49

Abstract

Background:Percutaneous coronary intervention (PCI) mostly depends on 2D angiography which has restricted depth imaging and may degrade the appearance of the lesions. Recent advances in artificial intelligence (AI) methods can quickly and safely create a 3D reconstruction of coronary vessels, which could be helpful when performing angioplasty.

Objective:To assess the viability, precision, and clinical usefulness of real-time AI-assisted 3D vessel reconstruction during PCI to facilitate better decision making and procedures.

Method:Paired views Paired angiographic views, implemented with a deep-learning reconstruction engine optimized to note rapid frame-to-frame vessels, were applied to paired angiographic views in this prospective study. The system was observed to produce real-time bi-dimensional 3D modeling of target vessels, such as lesion length, curvature, lumen diameter, and geometry of stenosis. The procedure in which cardiologists used PCI and had access to both the conventional 2D angiography and the AI-produced 3D guidance. Reference standards used were quantitative coronary angiography (QCA) and intra-vascular imaging.

Results:The time of processing required by AI based reconstruction was 180ms per frame which allows real-time updates. Luminosity dimension and stenosis harshness appeared to have a great deal of precision with QCA (r = 0.92) and IVUS/OCT measurements (r = 0.88). Improvements in lesion assessment were found to occur in operators especially bifurcations and tortuous vessels. The duration of the procedure was cut by 12% and usage of contrast was cut by 15%.

Conclusion:The timely 3D reconstruction of excitement AI has a remarkable effect on angiography interpretation at the time of PCI, visualization of the lesion, device selection and minimization of use of resources associated with the procedure. The technology is a prospective development of precision-guided coronary intervention.