Guidance for Intra-cardiac Echocardiography Manipulation to Maintain Continuous Therapy Device Tip Visibility

TL;DR

This work tackles the challenge of maintaining continuous therapy device tip visibility during intra-cardiac echocardiography by proposing an AI-driven model that estimates both the device tip entry location and its incident angle in the 2D ICE imaging plane. A pretrained ultrasound foundation model extracts frame features, which are integrated with historical passing points and angles in a transformer network to predict the current passing point and incident angle in real time, achieving 25 Hz operation. The authors employ a hybrid data strategy, combining clinical ICE sequences with synthetic overlays and ground-truth via EM sensors in a water chamber, yielding 5,698 ICE-tip image pairs for training and evaluation. Quantitative results show moderate angular errors ($$\hat{A}$$) and a bounding-box IoU around 0.66, indicating effective tip tracking under varied motions, with potential to enable robot-assisted ICE catheter control and continuous device visibility in real procedures. Future work emphasizes expanding clinical data to improve generalization and robustness across complex anatomies.

Abstract

Intra-cardiac Echocardiography (ICE) plays a critical role in Electrophysiology (EP) and Structural Heart Disease (SHD) interventions by providing real-time visualization of intracardiac structures. However, maintaining continuous visibility of the therapy device tip remains a challenge due to frequent adjustments required during manual ICE catheter manipulation. To address this, we propose an AI-driven tracking model that estimates the device tip incident angle and passing point within the ICE imaging plane, ensuring continuous visibility and facilitating robotic ICE catheter control. A key innovation of our approach is the hybrid dataset generation strategy, which combines clinical ICE sequences with synthetic data augmentation to enhance model robustness. We collected ICE images in a water chamber setup, equipping both the ICE catheter and device tip with electromagnetic (EM) sensors to establish precise ground-truth locations. Synthetic sequences were created by overlaying catheter tips onto real ICE images, preserving motion continuity while simulating diverse anatomical scenarios. The final dataset consists of 5,698 ICE-tip image pairs, ensuring comprehensive training coverage. Our model architecture integrates a pretrained ultrasound (US) foundation model, trained on 37.4M echocardiography images, for feature extraction. A transformer-based network processes sequential ICE frames, leveraging historical passing points and incident angles to improve prediction accuracy. Experimental results demonstrate that our method achieves 3.32 degree entry angle error, 12.76 degree rotation angle error. This AI-driven framework lays the foundation for real-time robotic ICE catheter adjustments, minimizing operator workload while ensuring consistent therapy device visibility. Future work will focus on expanding clinical datasets to further enhance model generalization.

Figures (3)

• Figure 1: A series of ICE images are processed by the US foundation model, while prior passing points and incident angles are handled separately. Extracted features are combined and fed into a transformer network, which predicts the final passing point and incident angle from distinct output layers.
• Figure 2: The dataset composition. (a) Entry angle $a_{entry}$: The angle at which the tip enters the US fan area. (b) Rotation angle $a_{rot}$: The rotational angle between the tip and the center-line of the US image in 2D. (C) Tip passing location: The position of the device within the 2D US image.
• Figure 3: Representative results of the proposed method. Each column presents a different test sample. The first row shows the angular prediction results, while the second row visualizes the tip location prediction. The blue color represents the target tip, whereas the orange color denotes the predicted tip.