摘要:
锥束图象重建算法正在快速发展,并用于重要的生物医学和工业应用中。在本文中,主要讨论有效的精确的、可能用于动态研究的算法,特别是近年发展起来的Grangeat类和Katsevich类的算法。这一选择是基于CT和显微CT的定量的和功能的应用需求。2002年,Lee和 Wang提出了圆周和螺旋情形Grangeat类的半扫描锥束算法,解决了短物体重建问题。 原理是利用在Grangeat类重建公式中的Radon 空间信息,在阴影区域进行适当的数据插值,从而抑制Feldkamp类重建算法的亮度减低伪影。在圆形和螺旋半扫描情形,首先我们确定位于不同采样冗余区域之间的边界。然后我们导出相应的加权函数以用于特征点的计算。就亮度减低伪影而论,Grangeat类半扫描重建算法优于Feldkamp类半扫描重建算法。2001年, Katsevich推导了第一个理论上精确的螺旋锥束滤波反投影型重建公式。其局限性是探测器窗口较大和待重建物体的半径较小。2002年 Katsevich改进了他的第一个公式。新的公式对病人的尺寸没有多少限制,而且相对旧公式假设了较小的探测器阵列。最近,Katsevich将他的方法推广到一般的扫描轨迹,证明了早期的两个公式是他的一般结果的特例。针对长物体的动态体成像,我们极其需要改进现有的Grangeat类和Katsevich类的算法。
关键词:
-
CT
/
-
显微CT
/
-
锥束
/
-
螺旋扫描
/
-
半扫描
/
-
图像重建
Abstract:
Cone-beam image reconstruction algorithms are in rapid development for major biomedical and industrial applications. In this report, we primarily focus on those algorithms that allow exact and efficient reconstruction and have potential for dynamic studies, which are recently developed Grangeat-type and Katsevich-type algorithms. This preference is due to the needs for quantitative and functional CT/micro-CT applications. Last year, Lee and Wang developed Grangeat-type half-scan cone-beam algorithms in the circular and helical scanning cases to solve the short object problem. In both the circular and helical half-scan cases, the boundaries between regions of different sample redundancies are first determined. Then, corresponding weighting functions are formulated for evaluation at various characteristic point. It was demonstrated that the Grangeat-type half-scan reconstruction clearly outperformed the Feldkamp-type half-scan reconstruction in terms of intensity dropping artifacts. In 2001, Katsevich derived the first theoretically exact reconstruction formula for the spiral cone-beam geometry in the filtered backprojection format. The limitations of this formula include a large detector window and a small radius of the object to be reconstructed. Last year, Katsevich improved his first formula remarkably. The new formula imposes little restriction on the size of the patient, and assumes a smaller detector array than the old formula. Recently, Katsevich generalized his method from the spiral scanning case to other trajectories, and proved that the earlier two formulas are special cases of his general formula. There are important needs to improve current Grangeat-type and Katsevich-type algorithms for dynamic volumetric imaging in the long object cone-beam geometry.
下载:
