磁共振电特性成像人体腹腔癌症仿真研究

发布时间：2018-06-10 17:57

  本文选题：磁共振 + 电特性　； 参考：《浙江大学学报(工学版)》2016年10期

【摘要】：将磁共振电特性成像(MREPT)方法用于人体腹腔部位,考察该成像方式用于人体腹腔肝癌及胰腺癌诊断的可行性.采用鸟笼线圈同时作为射频发射和接收,基于射频发射场测量值重构人体腹腔内组织的电导率和相对介电率.包含10种组织的人体腹腔模型上的仿真结果表明:在信噪比SNR=30的情况下,MREPT能够检测出位于组织内部半径为3cm的肝癌肿瘤和半径为1cm的胰腺癌肿瘤;由于逆问题重构公式中均质电特性的假设及拉普拉斯卷积的共同作用,可能无法检测出位于组织边界处的小体积癌症肿瘤.拉普拉斯卷积核的大小对重构结果的影响很大,在无噪声情况下,3×3×3的卷积核的重构效果最好,但当SNR=30~120时,5×5×5的卷积核是最佳选择.
[Abstract]:The magnetic resonance electrical characteristic imaging (MRECT) method was used in the diagnosis of hepatocarcinoma and pancreatic cancer in human abdominal cavity. The bird cage coils are used as radiofrequency transmitting and receiving simultaneously. The conductivity and relative dielectric rate of human abdominal tissue are reconstructed based on the measurements of RF emission site. The simulation results on a human abdominal model containing 10 kinds of tissues show that MREPT can detect hepatocellular carcinoma with 3cm radius and pancreatic cancer with 1cm radius under SNR 30. Due to the assumption of homogeneous electrical properties in the inverse problem reconstruction formula and the interaction of Laplacian convolution, it may not be possible to detect small volume cancer tumors located at the tissue boundary. The size of Laplace convolution kernel has great influence on the reconstruction result. The reconstruction effect of 3 脳 3 脳 3 convolution kernel without noise is the best, but 5 脳 5 脳 5 convolution kernel is the best choice when SNRN is 30 ~ 120.
【作者单位】： 厦门理工学院电气工程与自动化学院;休斯顿卫理公会派医院研究中心系统医药与生物工程系;厦门理工学院光电与通信工程学院;
【基金】：福建省自然科学基金资助项目(2013J05102;2016J01323) 福建省教育厅科研资助项目(JA15383)
【分类号】：R445.2;R735;TP391.9
，

本文编号：2004101