摘要
为提高电阻抗静态重构图像的质量,研究将离散变差函数引入到重构算法中,形成混合正则化重构算法,并对计及颅骨的颅内异物进行了重构成像。与常用的Tikhonov正则化算法相比,混合正则化重构算法对重构信息的分辨能力显著提高,能有效克服低电导率颅骨对颅内信息的屏蔽效应,显著改善所得重构图像质量,与设定病例的医学图像相符。这一成果提高了电阻抗静态成像技术在颅内异物定位方面的质量,为电阻抗静态成像技术的实用化打下了基础。
To improve the quality of the restored images, a variation function as part of regularization penalty term was introduced to the reconstruction algorithm of EIT. The images with intracranial foreign body were reconstructed based on the proposed mixed regularization algorithm. Compared with Tikbonov regularization algorithm, the mixed regularization algorithm displayed powerful resolution, which overcame the shielding effect resulted from low conductivity of skull. The restored image improved the quality in locating intracranial foreign body by static EIT, laying a basis for the application of EIT.
出处
《中国生物医学工程学报》
CAS
CSCD
北大核心
2007年第5期695-699,共5页
Chinese Journal of Biomedical Engineering
基金
国家自然科学基金重点项目(50337020)
重庆市科委自然科学基金(2006BB5212)。
关键词
电阻抗成像(EIT)
变差函数
混合正则化算法
颅内异物
electrical impedance tomography （EIT）
variation function
mixed regularization algorithm
intracranial foreign body
