VISION-ICE: Video-based Interpretation and Spatial Identification of Arrhythmia Origins via Neural Networks in Intracardiac Echocardiography

A research team from Rice University and the Texas Heart Institute has introduced VISION-ICE, a novel AI framework designed to automate the localization of arrhythmia origins using routine intracardiac echocardiography (ICE) video data. The system addresses a critical bottleneck in electrophysiology, where current high-density mapping techniques and preoperative CT/MRI scans are notoriously time-consuming and resource-intensive. By leveraging ICE—a standard imaging modality already integrated into ablation procedures—the researchers propose a real-time clinical decision aid that could streamline workflows. The core innovation formulates arrhythmia source identification as a three-class video classification problem, distinguishing between normal sinus rhythm, left-sided arrhythmias, and right-sided arrhythmias.

The technical backbone of VISION-ICE is a 3D Convolutional Neural Network (3D CNN) trained to analyze spatiotemporal patterns in ICE video sequences. In a ten-fold cross-validation study evaluated on four previously unseen patients, the model achieved a mean accuracy of 66.2%, substantially outperforming the 33.3% random-chance baseline. This proof-of-concept demonstrates the feasibility of using deep learning for automated, video-based cardiac analysis directly in the operating suite. The immediate implication is the potential for faster, more targeted electrophysiological interventions, reducing procedural duration and burden. The authors note that future work will focus on expanding the dataset to enhance model robustness and generalizability across diverse patient populations, paving the way for eventual clinical integration.