Harmonic-Aware Transformer for Real-Time Catheter Localization in Interventional Procedures of Magnetic Particle Imaging

2026-07-02

AI Summary

This study addresses the limitations of conventional magnetic particle imaging (MPI)–guided interventional procedures, which suffer from high latency, poor real-time performance, and insufficient robustness under complex motion due to reliance on image reconstruction. To overcome these challenges, the authors propose a harmonic-aware Transformer framework that, for the first time, applies Transformer architecture directly to MPI catheter tracking by end-to-end predicting the three-dimensional tip position from raw voltage signals, thereby eliminating the reconstruction step. The approach employs frequency-domain preprocessing to extract harmonics of orders 2–8, enhancing signal-to-noise ratio and motion-related features, and integrates a six-layer, eight-head attention encoder with explicit modeling of spatiotemporal dependencies across three axes. Ex vivo experiments demonstrate sub-millimeter localization accuracy (L2 error as low as 0.103 ± 0.092 mm; MAE ranging from 0.165 to 0.655 mm) and ultra-low inference latency of 0.55 ms per frame (equivalent to 1800 FPS), significantly outperforming existing methods.

Abstract

Magnetic particle imaging (MPI) enables real-time, radiation-free tracking of magnetic nanoparticle-coated instruments, making it highly suitable for interventional procedures. This study proposes a harmonic-aware transformer framework that directly predicts catheter tip positions from raw MPI voltage signals, eliminating the need for image reconstruction and reducing computational latency. The framework incorporates frequency-domain preprocessing to isolate the 2nd to 8th drive-field harmonics, enhancing the signal-to-noise ratio while preserving motion-relevant features. A transformer architecture with six encoder layers and eight attention heads is employed to learn spatio-temporal dependencies across the three receive axes (x, y, z) for accurate three-dimensional position estimation. The model is trained on simulated MPI signals and evaluated on real in vitro datasets under standard, bending, and heartbeat-like motion conditions. The proposed method achieves sub-millimeter localization accuracy, with a minimum L2 error of 0.103 +/- 0.092 mm and mean absolute errors (MAEs) of 0.039 +/- 0.046 mm, 0.054 +/- 0.049 mm, and 0.060 +/- 0.044 mm along the (x, y, z) axes, respectively, for the bending dataset. Across all datasets, the MAE ranges from 0.165 mm to 0.655 mm, demonstrating consistent performance. The optimized inference achieves a latency of 0.55 ms per frame and a throughput of approximately 1800 frames per second, confirming real-time capability. Compared with conventional MPI-guided approaches relying on image reconstruction, the proposed framework provides improved accuracy, reduced latency, and enhanced robustness under complex motion conditions. These results highlight the potential of harmonic-aware transformer models as efficient and scalable solutions for real-time catheter localization in interventional MPI.

Problem

• Magnetic Particle Imaging
• Catheter Localization
• Real-Time Tracking
• Interventional Procedures
• Harmonic Signals

Innovation

• Harmonic-aware Transformer
• Magnetic Particle Imaging (MPI)
• Real-time Catheter Localization
• Image-free Tracking
• Spatio-temporal Attention