A Field Free Line 3D Reconstruction Model for Magnetic Particle Imaging for Improved Sensitivity, Resolution, and High Dynamic Range Imaging

Abstract

Magnetic particle imaging (MPI) is a tracer-based imaging modality that detects superparamagnetic iron oxide nanoparticles in vivo, with applications in cancer cell tracking, lymph node mapping, and cell therapy monitoring. We introduce a new 3D image reconstruction framework for MPI data acquired using multi-angle field-free line (FFL) scans, demonstrating improvements in spatial resolution, quantitative accuracy, and high dynamic range performance over conventional sequential reconstruction pipelines. The framework is built by combining a physics-based FFL signal model with tomographic projection operators to form an efficient 3D forward operator, enabling the full dataset to be reconstructed jointly rather than as a series of independent 2D projections. A harmonic-domain compression step is incorporated naturally within this operator formulation, reducing memory overhead by over two orders of magnitude while preserving the structure and fidelity of the model, enabling volumetric reconstructions on standard desktop GPU hardware in only minutes. Phantom and in vivo results demonstrate substantially reduced background haze and improved visualization of low-intensity regions adjacent to bright structures, with an estimated $\sim$11$\times$ improvement in iron detection sensitivity relative to the conventional X-space CT approach. These advances enhance MPI image quality and quantitative reliability, supporting broader use of MPI in preclinical and future clinical imaging.