Choi E J, Oh Y-W, Ham S Y, Lee K Y, Kang E-Y
Department of Radiology, Anam Hospital, Korea University, Seoul, South Korea.
Br J Radiol. 2008 Jun;81(966):463-7. doi: 10.1259/bjr/63217190. Epub 2008 Feb 18.
The aims of this study were to compare the image quality of coronal multiplanar reconstruction (MPR) images from axial spiral images with that of direct coronal spiral and sequential images, and to estimate and analyse the effect of an incremental change on the image quality using 64-detector row CT. 12 swine lungs were used. Five kinds of images from each lung specimen were obtained using 64-detector row CT. All images were analysed by categories and grades, and the direct coronal sequential images were used as the reference standard for the image quality. Statistical analysis was performed for the following categories: (i) inter-observer reliability, (ii) interaction between the observers and images, (iii) image analysis, (iv) anatomical structural analysis of each observer, (v) stair-step artefact and (vi) background noise. The overall image quality and the image quality of all anatomical structures of coronal MPR images with 0.67 mm slice increments were inferior to the image quality of the other images; this difference was statistically significant (p<0.05). Stair-step artefact was detected on coronal MPR images, and was more prominent on coronal MPR images with 0.67 mm slice increments than on coronal MPR images with 0.34 mm slice increments. The most severe background noise was detected on the direct coronal sequential images, but there was no significant difference between the direct coronal sequential images and the direct coronal spiral images. Background noise was least prominent on coronal MPR images with 0.67 mm slice increments. The increment process is important for improving the image quality of MPR images even when using 64-detector row CT. Coronal MPR images with 0.34 mm slice increments using 64-detector row CT showed a similar image quality to that obtained from the direct coronal images, and can be used instead. This means that the coronal MPR images obtained with 64-detector row CT could be as useful for evaluating the lung parenchyma as the axial high-resolution CT images.
本研究的目的是比较轴向螺旋图像的冠状多平面重建（MPR）图像与直接冠状螺旋图像及序列图像的图像质量，并使用64排CT估计和分析增量变化对图像质量的影响。使用了12个猪肺。每个肺标本使用64排CT获得五种图像。所有图像按类别和等级进行分析，直接冠状序列图像用作图像质量的参考标准。对以下类别进行了统计分析：（i）观察者间可靠性；（ii）观察者与图像之间的相互作用；（iii）图像分析；（iv）每个观察者的解剖结构分析；（v）阶梯状伪影；（vi）背景噪声。层厚增量为0.67mm的冠状MPR图像的整体图像质量及所有解剖结构的图像质量均低于其他图像；这种差异具有统计学意义（p<0.05）。在冠状MPR图像上检测到了阶梯状伪影，层厚增量为0.67mm的冠状MPR图像上的阶梯状伪影比层厚增量为0.34mm的冠状MPR图像上更明显。在直接冠状序列图像上检测到最严重的背景噪声，但直接冠状序列图像与直接冠状螺旋图像之间无显著差异。背景噪声在层厚增量为0.67mm的冠状MPR图像上最不明显。即使使用64排CT，增量过程对于提高MPR图像的图像质量也很重要。使用64排CT、层厚增量为0.34mm的冠状MPR图像显示出与直接冠状图像相似的图像质量，可替代使用。这意味着64排CT获得的冠状MPR图像在评估肺实质方面可能与轴向高分辨率CT图像一样有用。
