A novel magnetic localization algorithm for wireless endoscopic capsules in e-Health

Abstract

Wireless Capsule Endoscopy (WCE) enables non-invasive visualization of the entire gastrointestinal (GI) tract but faces a critical challenge: accurate real-time localization and orientation of the capsule. The present study demonstrates a high-performance magnetic localization system capable of estimating the capsule's 5D parameters (3D position and 2D orientation) with high precision and efficiency. Using the Levenberg–Marquardt Algorithm (LMA) to solve nonlinear magnetic field equations, the proposed method was validated in MATLAB and implemented on a Kintex-7 FPGA. The software algorithm was validated in a realistic simulation environment, achieving an average positioning error of 0.027 ± 0.012 mm and an orientation error of 0.8 ± 0.3° under simulated sensor noise. The corresponding FPGA implementation attained a throughput of 410.694 MHz with minimal resource usage, demonstrating the feasibility of real time, low power execution. Benchmarked against state-of-the-art techniques in simulation, such as the differential geomagnetic method (3 mm, 0.014 s) (EEE Trans Instrum Meas 71:1–10. 2022, 10.1109/TIM.2021.3129206) and active electromagnetic tracking (0.79 mm, 0.240 s) (IEEE Sens J 18(3):1178-1186. 2018, 10.1109/JSEN.2017.2779560), the proposed passive magnetic system shows significant improvements in accuracy and hardware efficiency, achieving superior accuracy (0.027 mm) with competitive real-time speed (0.166 s). This work presents a co-designed algorithm and architecture that, under simulation, achieves high-precision localization metrics. This represents a step toward technologies that could enable more precise diagnostics and paving the way for future therapeutic applications.