Blind deblurring of spiral CT images-comparative studies on edge-to-noise ratios.
Blind deblurring of spiral CT images-comparative studies on edge-to-noise ratios.
复制标题
DOI：
10.1118/1.1470500
复制
发表时间：
2002-05
期刊：
影响因子：
3.8
通讯作者：
M. Jiang;Ge Wang;M. Skinner;J. Rubinstein;M. Vannier
中科院分区：
文献类型：
--
作者：
M. Jiang;Ge Wang;M. Skinner;J. Rubinstein;M. Vannier
提出了一种基于边缘噪声比的螺旋CT图像盲去模糊算法。由于用于量化边缘和噪声效应的差异度量是不对称的，因此有几种方法来制定边缘噪声比。本文旨在通过体模和患者数据研究这些比率的性能。在体模研究中，研究表明所有比率都具有相似的属性，从而验证了盲去模糊算法。根据均方根误差（RMSE）标准，不同比率的图像保真度改善从29%到33%不等;为每个比率确定的最佳迭代次数从25到35不等。在数值模拟中，与最令人满意的性能相关联的那些比率被挑出来用于约33%的图像保真度改善。在自动盲去模糊与选定的比率，CT的空间分辨率基本上细化在所有的情况下测试。
A recently developed blind deblurring algorithm based on the edge-to-noise ratio has been applied to improve the quality of spiral CT images. Since the discrepancy measure used to quantify the edge and noise effects is not symmetric, there are several ways to formulate the edge-to-noise ratio. This article is to investigate the performance of those ratios with phantom and patient data. In the phantom study, it is shown that all the ratios share similar properties, validating the blind deblurring algorithm. The image fidelity improvement varies from 29% to 33% for different ratios, according to the root mean square error (RMSE) criterion; the optimal iteration number determined for each ratio varies from 25 to 35. Those ratios that are associated with most satisfactory performance are singled out for the image fidelity improvement of about 33% in the numerical simulation. After automatic blind deblurring with the selected ratios, the spatial resolution of CT is substantially refined in all the cases tested.