5-DOF Microcoil Positioning System Utilizing Single-Axis Electromagnetic Transmitter

Abstract

The lack of 3-D localization impedes the advancement of various intracorporeal medical devices. Developing an occlusion-free, small-sized, and high-precision positioning system remains a significant challenge in the practical application of interventional robotics. In this article, a novel 5-degree of freedom (DoF) positioning system utilizing single-axis electromagnetic field excitation coil is designed to locate a microcoil with a size of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$\phi$</tex-math></inline-formula> 1.45 × 5 mm. An XGBoost-based induced electromotive force (EMF) prediction model is proposed to correct the deviation of the magnetic dipole model near the excitation source. Employing the model, a positioning dataset under rotating magnetic field is synthesized. Subsequently, a CNN-LSTM based 5-DoF positioning model is designed and trained, and the transmitting coil speed characteristics of the model were experimentally verified. Evaluation experiments for the two models are performed separately. The results demonstrate the XGBoost-based EMF prediction model improved the prediction accuracy by 34.4% compared to the conventional magnetic dipole model. The static average localization error of the 5-DoF positioning model is 2.53 mm and the orientation error is 2.24<inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$^\circ$</tex-math></inline-formula> within the <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$\phi$</tex-math></inline-formula> 150 × 70 mm volume. The dynamic tracking experimental results indicate that the localization tracking error is 4.25 mm and the orientation error is 3.44<inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$^\circ$</tex-math></inline-formula>, which are 62% and 78% higher than the Levenberg–Marquardt algorithm. The navigation experiment conducted in a coronary artery phantom demonstrated the potential for use in narrow tracts.