Localization of a Magnetic Capsule Endoscope Using Bilinear Interpolation and Coordinate Transformation of an External Magnetic Field and the SNR-based Weighting Factor in Optimization

Abstract

A magnetic capsule endoscope (MCE) embedded with a permanent magnet (PM) can be actuated and controlled by an external magnetic field generated by a magnetic navigation system (MNS), enabling wireless examination of lesions in the gastrointestinal (GI) tract. For precise control of the MCE, accurate localization of the MCE is required, and prior researchers localized an MCE by estimating its generated magnetic field. However, it is difficult to separate the magnetic field of the MCE from that of the MNS due to modeling errors, sensor magnetization, and electronic noise. In this study, we propose a robust localization algorithm that operates even in the presence of a strong external magnetic field generated by a cylindrical PM of an MNS to actuate an MCE. We constructed a bilinear interpolation model of the external magnetic flux density measured at the central magnetic sensor when the cylindrical PM of the MNS moves along the axisymmetric plane. Then we applied coordinate transformation to predict the magnetic flux density at other sensors. Using this model, the magnetic flux density generated by the MNS can be subtracted from the measured magnetic flux density to determine that of the MCE. In addition, we incorporate a weighting factor considering the signal-to-noise ratio of the magnetic sensor into the proposed algorithm to increase the estimation accuracy of the position and orientation of the MCE. The proposed algorithm is validated through experiments with a 3×3 magnetic sensor array and a commercial MCE. We experimentally demonstrate that the proposed algorithm can estimate the position and orientation of the MCE within a clinically applicable range even in the presence of an external magnetic field to actuate an MCE.