Cho Seungryong, Bian Junguo, Pelizzari Charles A, Chen Chin-Tu, He Tong-Chuan, Pan Xiaochuan
Department of Radiology, University of Chicago, Chicago, Illinois 60637, USA.
Med Phys. 2007 Dec;34(12):4923-33. doi: 10.1118/1.2804924.
Cone-beam microcomputed tomography (microCT) is one of the most popular choices for small animal imaging which is becoming an important tool for studying animal models with transplanted diseases. Region-of-interest (ROI) imaging techniques in CT, which can reconstruct an ROI image from the projection data set of the ROI, can be used not only for reducing imaging-radiation exposure to the subject and scatters to the detector but also for potentially increasing spatial resolution of the reconstructed images. Increasing spatial resolution in microCT images can facilitate improved accuracy in many assessment tasks. A method proposed previously for increasing CT image spatial resolution entails the exploitation of the geometric magnification in cone-beam CT. Due to finite detector size, however, this method can lead to data truncation for a large geometric magnification. The Feldkamp-Davis-Kress (FDK) algorithm yields images with artifacts when truncated data are used, whereas the recently developed backprojection filtration (BPF) algorithm is capable of reconstructing ROI images without truncation artifacts from truncated cone-beam data. We apply the BPF algorithm to reconstructing ROI images from truncated data of three different objects acquired by our circular cone-beam microCT system. Reconstructed images by use of the FDK and BPF algorithms from both truncated and nontruncated cone-beam data are compared. The results of the experimental studies demonstrate that, from certain truncated data, the BPF algorithm can reconstruct ROI images with quality comparable to that reconstructed from nontruncated data. In contrast, the FDK algorithm yields ROI images with truncation artifacts. Therefore, an implication of the studies is that, when truncated data are acquired with a configuration of a large geometric magnification, the BPF algorithm can be used for effective enhancement of the spatial resolution of a ROI image.
锥形束微型计算机断层扫描（microCT）是小动物成像最常用的选择之一，正成为研究移植疾病动物模型的重要工具。CT中的感兴趣区域（ROI）成像技术可以从ROI的投影数据集重建ROI图像，不仅可用于减少对受试者的成像辐射暴露以及对探测器的散射，还可能提高重建图像的空间分辨率。提高microCT图像的空间分辨率有助于在许多评估任务中提高准确性。先前提出的一种提高CT图像空间分辨率的方法是利用锥形束CT中的几何放大率。然而，由于探测器尺寸有限，对于较大的几何放大率，该方法可能导致数据截断。当使用截断数据时，Feldkamp-Davis-Kress（FDK）算法会产生带有伪影的图像，而最近开发的反投影滤波（BPF）算法能够从截断的锥形束数据重建无截断伪影的ROI图像。我们应用BPF算法从我们的圆形锥形束microCT系统采集的三种不同物体的截断数据中重建ROI图像。比较了使用FDK和BPF算法从截断和未截断的锥形束数据重建的图像。实验研究结果表明，从某些截断数据中，BPF算法可以重建质量与从未截断数据重建的图像相当的ROI图像。相比之下，FDK算法产生带有截断伪影的ROI图像。因此，这些研究的一个启示是，当以大几何放大率配置采集截断数据时，BPF算法可用于有效提高ROI图像的空间分辨率。
