Yin Zhye, Yao Yangyang, Montillo Albert, Wu Mingye, Edic Peter M, Kalra Mannudeep, De Man Bruno
Image Reconstruction Laboratory, GE Global Research, Niskayuna, New York 12309.
X-ray and CT Laboratory, GE Global Research, Shanghai 201203, China.
Med Phys. 2015 May;42(5):2730-9. doi: 10.1118/1.4921065.
Traditionally, 2D radiographic preparatory scan images (scout scans) are used to plan diagnostic CT scans. However, a 3D CT volume with a full 3D organ segmentation map could provide superior information for customized scan planning and other purposes. A practical challenge is to design the volumetric scout acquisition and processing steps to provide good image quality (at least good enough to enable 3D organ segmentation) while delivering a radiation dose similar to that of the conventional 2D scout.
The authors explored various acquisition methods, scan parameters, postprocessing methods, and reconstruction methods through simulation and cadaver data studies to achieve an ultralow dose 3D scout while simultaneously reducing the noise and maintaining the edge strength around the target organ.
In a simulation study, the 3D scout with the proposed acquisition, preprocessing, and reconstruction strategy provided a similar level of organ segmentation capability as a traditional 240 mAs diagnostic scan, based on noise and normalized edge strength metrics. At the same time, the proposed approach delivers only 1.25% of the dose of a traditional scan. In a cadaver study, the authors' pictorial-structures based organ localization algorithm successfully located the major abdominal-thoracic organs from the ultralow dose 3D scout obtained with the proposed strategy.
The authors demonstrated that images with a similar degree of segmentation capability (interpretability) as conventional dose CT scans can be achieved with an ultralow dose 3D scout acquisition and suitable postprocessing. Furthermore, the authors applied these techniques to real cadaver CT scans with a CTDI dose level of less than 0.1 mGy and successfully generated a 3D organ localization map.
传统上，二维放射摄影预备扫描图像（定位扫描）用于规划诊断性CT扫描。然而，具有完整三维器官分割图的三维CT容积数据可为定制扫描规划及其他目的提供更优质的信息。一个实际的挑战是设计容积式定位扫描的采集和处理步骤，以提供良好的图像质量（至少足以实现三维器官分割），同时给予与传统二维定位扫描相似的辐射剂量。
作者通过模拟和尸体数据研究，探索了各种采集方法、扫描参数、后处理方法和重建方法，以实现超低剂量三维定位扫描，同时降低噪声并保持目标器官周围的边缘强度。
在模拟研究中，基于噪声和归一化边缘强度指标，采用所提出的采集、预处理和重建策略的三维定位扫描提供了与传统240 mAs诊断扫描相似的器官分割能力。同时，所提出的方法仅给予传统扫描剂量的1.25%。在尸体研究中，作者基于图像结构的器官定位算法成功地从采用所提出策略获得的超低剂量三维定位扫描中定位了主要的胸腹器官。
作者证明，通过超低剂量三维定位扫描采集和适当的后处理，可以获得与传统剂量CT扫描具有相似分割能力（可解释性）的图像。此外，作者将这些技术应用于CTDI剂量水平小于0.1 mGy的真实尸体CT扫描，并成功生成了三维器官定位图。
