Catheter tip position estimated more accurately without shaft sensors
Shaft-Configuration-Adaptive Catheter Tip Position Estimation via Motor-History Conditioned Residual Learning
Robotics
Summary
Accurately knowing the tip position of flexible medical catheters is hard because the shape of the catheter shaft changes and friction causes errors. The authors designed a method that learns from the motor controls and their past motions to better predict the catheter’s tip location in 3D space without needing extra sensors on the shaft. They first calibrate a simple geometric model with a special motor sweep and then use a recurrent neural network to correct that model during use. This method reduced errors by over half compared to traditional models.
What this means in practice
- •For medical device engineers: Improve real-time position tracking of steerable catheters to enhance precision in minimally invasive heart imaging procedures without adding shaft sensors.
- •For robotic surgery system developers: Incorporate motor-history-conditioned learning to increase tip estimation accuracy during catheter navigation in complex anatomical paths, aiding robotic control algorithms.
Authors
Peihan Zhang, Michael C. Yip, Ankur Kapoor, Young-Ho Kim
Abstract
Tendon-driven continuum manipulators are widely used in medical applications, where accurate tip-position estimation is essential for precise navigation and instrument positioning. However, patient anatomy and procedural setup impose task-dependent unknown shaft configurations, while friction, slack, and compliance introduce hysteresis, making tip estimation challenging. This paper presents a motor-history-conditioned gated recurrent unit (GRU) residual estimator for three-dimensional catheter tip estimation without direct shaft-configuration sensing. First, an initial multidirectional sweep strategy is applied to calibrate a geometric catheter model backbone, and encode the motor-angle and drive-torque response into a shaft-configuration context vector. During subsequent motion, the context conditions a GRU that predicts a task-space residual correcting this backbone, relying on motor measurements alone. The context remains fixed for the current shaft configuration, while the recurrent state captures the evolving actuation history. Across four disposable intra-cardiac echocardiography catheters and 16 bent shaft configurations, the method achieves 3.3mm open-loop tip RMSE, a 59% reduction relative to the constant-curvature baseline.