← Back ICRA 2026

From Patient-Specific Digital Twin to Real-World Phantom: Autonomous Right Heart Catheterization

Yaxi Wang, Mengzhe Xu, Wenlong Gaozhang, Helge Arne Wurdemann

PDF

AI summary

Key figure (auto-extracted from paper)

A CNN-trained digital twin successfully transfers to a real-world robot, enabling highly accurate and consistent autonomous right heart catheterization in patient-specific static and dynamic phantoms.

Right heart catheterization sim-to-real transfer digital twin autonomous robotics CNN policy interventional cardiology

Problem

Manual right heart catheterization suffers from navigation inaccuracies and high clinician workload due to operator-dependent variability. Existing robotic approaches struggle to transfer simulation-trained policies to real-world, patient-specific dynamic environments without extensive physical demonstration data.

Approach

The authors build a patient-specific cardiac digital twin in SOFA to simulate catheter navigation and collect behavioral data. This data trains a CNN policy model that is then deployed on a physical robot to autonomously guide a Swan-Ganz catheter in real-world phantoms.

Key results

Digital twin trajectory replicability validated with sub-5mm positional error
Auto-RHC achieved ≥96% success in static and ≥94% in dynamic heartbeat phantoms
Navigation consistency improved by ≥34.63% over expert manual operation
CNN policy model trained on virtual data achieved >93.5% validation accuracy

Why it matters

Provides a scalable sim-to-real framework for reducing clinician burden and improving precision in minimally invasive cardiovascular interventions.

Abstract

Right heart catheterization (RHC) is a critical procedure for diagnosing and managing cardiovascular dis- eases (CVDs) such as heart failure, congenital heart disease, pulmonary edema, and pulmonary hypertension. However, cur- rently prevalent manual RHC procedures requires continuous communication of clinicians between the main control room and the operating room, leading to navigation inaccuracies and increased physical workload for clinicians during prolonged procedure. To overcome these challenges, this paper introduces a robotic system that enables autonomous RHC (Auto-RHC) by transferring a catheter decision-making model from patient- specific digital twins to real-world robotic intervention using deep learning (DL) algorithms. By creating a patient-specific (PS) digital twin using the Simulation Open Framework Ar- chitecture (SOFA) and conducting virtual RHC interventions, images capturing the catheter balloon’s position and aligned behavioral datasets were collected and utilized as input for a convolutional neural network (CNN) architecture. The trained catheter decision-making model derived from the digital twin was then transferred to real-world implementations of robot- assisted Auto-RHC. Experimental results validated the perfor- mance of the digital twin and demonstrated that the real- world robotic Auto-RHC achieved a high success rate across both static (≥96%) and dynamic heartbeat (≥94%) patient- specific cardiac phantoms. Furthermore, Auto-RHC enhanced navigation consistency by ≥34.63% compared to expert manual operation.

Index terms

Medical Robots and Systems Deep Learning Methods Simulation and Animation