基于梯度下降法的双能CT双物质分解算法

发布时间：2018-02-02 17:50

本文关键词： 双能CT 双物质分解梯度下降图像重建计算步长　出处：《东北大学学报(自然科学版)》2017年07期 　论文类型：期刊论文

【摘要】：基物质分解是双能CT重建的重要步骤,其中双物质分解是常用的分解模型之一,该模型的核心关键是计算分解系数投影.为了更快计算它,提出了基于误差反馈梯度下降的双能CT双物质分解算法和基于Armijo-Goldstein梯度下降的双能CT双物质分解算法.由于计算了梯度下降步长,这两种方法能快速迭代求解基物质分解系数投影.同时他们有效地解决了双能CT重建的非线性问题.仿真实验结果显示,与传统查表匹配法相比,这两种算法稳定收敛,计算速度快,重建精度高,对临床应用有重要的意义.在重建结果精度近似的情况下,基于Armijo-Goldstein梯度下降的算法采用不精确线性搜索步长,因此它的运行速度更快.
[Abstract]:The decomposition of base matter is an important step in the reconstruction of dual energy CT, in which the decomposition of double matter is one of the commonly used decomposition models. The key of the model is to calculate the projection of decomposition coefficient. A double energy CT double matter decomposition algorithm based on error feedback gradient descent and a dual energy CT double matter decomposition algorithm based on Armijo-Goldstein gradient descent are proposed. Step length. These two methods can quickly iterate to solve the projection of decomposition coefficient of base matter. At the same time, they can effectively solve the nonlinear problem of dual-energy CT reconstruction. The simulation results show that compared with the traditional lookup table matching method. These two algorithms converge stably, have high calculation speed and high reconstruction accuracy, and are of great significance to clinical application. The algorithm based on Armijo-Goldstein gradient descent uses an imprecise linear search step, so it runs faster.
【作者单位】：东北大学中荷生物医学与信息工程学院;
【基金】：国家自然科学基金资助项目(61372014,61302013) 辽宁省科技厅工业攻关及成果产业化项目(2014305001)
【分类号】：R814.42;TP391.41
【正文快照】： 计算机断层成像技术(computed tomography,CT)在医学诊断中起着重要的作用,双能CT是新一代的CT系统.传统CT只利用X射线的衰减信息而忽略了能谱信息,只能得到扫描组织的形态学结构图像[1-2].而双能CT充分考虑了X射线的衰减信息和能谱信息,能同时得到扫描组织的形态学结构图像和

【相似文献】