基于系统矩阵的磁粒子成像重构研究进展

Research progress on reconstruction for magneticparticle imaging based on system matrix

磁粒子成像(magnetic particle imaging,MPI)是一种新型、前沿的层析成像方式,对具有生物相容性的超顺磁纳米粒子进行浓度分布成像,具有高灵敏度、高分辨率、无辐射等特征。基于系统矩阵的MPI重构是MPI重构研究的重要分支,较X-空间(X-Space)法具有更精确的重建效果。本文简要介绍了MPI的成像及重构原理;重点阐述了基于系统矩阵的MPI重构研究现状,对比分析了系统矩阵的测量与模型构建方法,归纳总结了吉洪诺夫正则化、LASSO算法、非负的融合套索模型与联合重建法的优劣;最后针对公开的数据库提出合理的频率筛选、系统矩阵稀疏模型构建、正则化方法探索改进的几点研究建议,以期为MPI的重构研究提供一定的方法参考,更好地促进未来磁粒子成像在临床医学中的良好应用。

Magnetic particle imaging is an emerging and frontier tomographic imaging modality that determines the concentration of biocompatible superparamagnetic nanoparticles with high sensitivity,high spatial resolution and no radiation.MPI reconstruction based on system matrix is an important branch of MPI reconstruction research,and has more accurate reconstruction effect than X-Space method.This paper briefly introduces the imaging principle of MPI and its reconstruction theory,and focuses on the MPI reconstruction progress based on system matrix,analyzes and compares the different methods for system matrix construction of measurement and model,summarizes the advantages and disadvantages of Tikhonov regularization method,LASSO algorithm,non-negative fused lasso model and joint reconstruction method.Finally,the paper proposes certain research suggestions for the open database,such as reasonable frequency screening,system matrix sparse model and regularization improvement method,expecting that this review may provide a convenience for researchers in MPI reconstruction and promote the future application of MPI in clinical medicine.