文章摘要

文章信息/Info

Title:: Quantitative analysis of magnetic particle imaging based on frequency selection

作者:: 王璐1; 2; 黄妍1; 2; 赵诣深1; 2; 杜洋3; 张璐1; 2△; （1.首都医科大学生物医学工程学院，北京 100069；2.首都医科大学临床生物力学应用基础研究北京市重点实验室，北京 100069；3.中国科学院自动化研究所中国科学院分子影像重点实验室，北京 100190）

Author(s):: WANG Lu1; 2; HUANG Yan1; 2; ZHAO Yishen1; 2; DU Yang3; ZHANG Lu1; 2; （1. School of Biomedical Engineering, Capital Medical University, Beijing 100069,China；2.Beijing Key Laboratory of Fundamental Research on Biomechanics in Clinical Application, Capital Medical University, Beijing 100069；3.Key Laboratory of Molecular Imaging, Institute of Automation, Chinese Academy of Sciences, Beijing 100190）

Keywords:: Magnetic particle imaging; Superparamagnetic oxide iron nanoparticles; System matrix; Reconstruction; K-means clustering; Image segmentation

摘要:: 为实现磁粒子成像（magnetic particle imaging, MPI）中示踪剂铁量化，本研究基于仿真程序，采集二维扫描图像，筛选信号频率并重建图像,以聚类结果作为先验信息，约束水平集函数的演化，对示踪剂图像进行分割，并借助Dice系数、IoU等参数定量评估分割效果。通过计算分割区域的信号强度总和，建立MPI信号与已知示踪剂铁含量的校正曲线。结果显示，频率筛选后显著缩短了信号重建时间；基于先验的水平集方法Dice系数和IoU均大于0.90，实现了肿瘤区域较精准的分割；通过本研究建立的MPI信号强度与铁含量的校正曲线，实现了示踪剂铁量化，平均误差为3.11%，最小误差0.03%。结果表明，基于先验的水平集方法可实现较精确的图像分割和铁量化，为MPI临床前定量研究提供参考。

Abstract:: To realize iron quantification of tracer in magnetic particle imaging(MPI), we acquired 2D scanned images based on a simulation program and reconstructed images after signal frequencies selection. The clustering results were used as a priori information to constrain the evolution of the level set function, and the segmentation effect was quantitatively evaluated by parameters such as dice coefficient and intersection over union(IoU).By calculating the total signal intensity of the segmented region, the calibration curve between MPI signal and known tracer iron content was established. It was found that the reconstruction time was shortened after frequency selection. The Dice coefficient and IoU of the segmentation results based on the prior level set method were both greater than 0.90, which realized more accurate segmentation of tumor region. Through the calibration curve between MPI signal intensity and iron content established in this study, the tracer iron quantification was realized, with an average error of 3.11% and a minimum error of 0.03%. The results show that the level set method based on the prior can achieve more accurate image segmentation and iron quantification, which provide reference for the preclinical quantitative study of magnetic particle imaging.