Ghani Muhammad U, Ren Liqiang, Wong Molly, Li Yuhua, Zheng Bin, Rong Xiujiang John, Yang Kai, Liu Hong
From the *Center for Biomedical Engineering and School of Electrical and Computer Engineering, University of Oklahoma, Norman, OK; †Department of Imaging Physics, The University of Texas MD Anderson Cancer Center, Houston, TX; and ‡Department of Radiation, Massachusetts General Hospital, Boston, MA.
J Comput Assist Tomogr. 2017 Jan;41(1):82-89. doi: 10.1097/RCT.0000000000000483.
The aim of this study was to investigate the noise power properties of a micro-computed tomography (micro-CT) system under different operating conditions.
A commercial micro-CT was used in the study that used a flat panel detector with a 127-μm-pixel pitch and a micro-focus x-ray tube. Conical tubes of various diameters were used under different acquisition conditions. Multidimensional noise power spectrums were used as a metric to investigate the noise properties of the system. Noise power spectrum was calculated from the difference data generated by subtraction of 2 identical scans. The noise properties with respect to various parameters that include the impact of number of projections, x-ray spectra, milliampere-second, slice location, object diameter, voxel size, geometric magnification (M), back-projection filters, and reconstruction magnification (Mrecon) were studied.
At a same isocentric exposure rate of 270 mR/s, the noise power was much lower for the image reconstructed with 3672 views (122 seconds) as compared with the 511 views (17 seconds), whereas at a fixed isocentric exposure of 4600 mR, the noise power levels were almost similar. Image noise with a 50-kV beam was higher as compared with the 90-kV beam at a same isocentric exposure. Image noise from a 16-mm-diameter conical tube was much lower as compared with the 28- and 56-mm tubes under identical isocentric exposures. The choice of back-projection filter influences noise power spectrum curves in terms of width and amplitudes. Reconstruction magnification applied during the reconstruction process increased the noise power at lower spatial frequencies but reduced the noise power at higher spatial frequencies. It can be established that, for small details corresponding to high spatial frequencies, reconstruction magnification can provide an improved signal-to-noise ratio. At all spatial frequencies, the in-plane images had lower noise power levels as compared with the z-plane images.
The noise power properties investigated in this study provide important image quality references for refined cone beam system development, optimization, and operations.
本研究旨在调查微型计算机断层扫描（micro-CT）系统在不同运行条件下的噪声功率特性。
本研究使用了一台商用微型计算机断层扫描设备，该设备采用了像素间距为127μm的平板探测器和微焦点X射线管。在不同的采集条件下使用了各种直径的锥形管。多维噪声功率谱被用作衡量系统噪声特性的指标。噪声功率谱由两次相同扫描相减生成的差值数据计算得出。研究了与各种参数相关的噪声特性，这些参数包括投影数量、X射线光谱、毫安秒、切片位置、物体直径、体素大小、几何放大倍数（M）、反投影滤波器和重建放大倍数（Mrecon）。
在相同的270 mR/s等中心曝光率下，与511次视图（17秒）重建的图像相比，3672次视图（122秒）重建的图像噪声功率要低得多，而在固定的4600 mR等中心曝光下，噪声功率水平几乎相似。在相同的等中心曝光下，50 kV光束的图像噪声比90 kV光束更高。在相同的等中心曝光下，直径16 mm的锥形管产生的图像噪声比28 mm和56 mm的管子低得多。反投影滤波器的选择会在宽度和幅度方面影响噪声功率谱曲线。重建过程中应用的重建放大倍数在较低空间频率下增加了噪声功率，但在较高空间频率下降低了噪声功率。可以确定的是，对于对应于高空间频率的小细节，重建放大倍数可以提高信噪比。在所有空间频率下，平面内图像的噪声功率水平比z平面图像低。
本研究中调查的噪声功率特性为改进锥束系统的开发、优化和操作提供了重要的图像质量参考。
