Liu R R, Rudin S, Bednarek D R
Toshiba Stroke Research Center, Department of Physics, State University of New York at Buffalo, 14214, USA.
Med Phys. 1999 Sep;26(9):1802-10. doi: 10.1118/1.598684.
Image intensifier (II) distortion changes as a function of C-arm rotation angle because of changes in the orientation of the II with the earth's or other stray magnetic fields. For cone-beam computed tomography (CT), distortion correction for all angles is essential. The new super-global distortion correction consists of a model to continuously correct II distortion not only at each location in the image but for every rotational angle of the C arm. Calibration bead images were acquired with a standard C arm in 9 in. II mode. The super-global (SG) model is obtained from the single-plane global correction of the selected calibration images with given sampling angle interval. The fifth-order single-plane global corrections yielded a residual rms error of 0.20 pixels, while the SG model yielded a rms error of 0.21 pixels, a negligibly small difference. We evaluated the accuracy dependence of the SG model on various factors, such as the single-plane global fitting order, SG order, and angular sampling interval. We found that a good SG model can be obtained using a sixth-order SG polynomial fit based on the fifth-order single-plane global correction, and that a 10 degrees sampling interval was sufficient. Thus, the SG model saves processing resources and storage space. The residual errors from the mechanical errors of the x-ray system were also investigated, and found comparable with the SG residual error. Additionally, a single-plane global correction was done in the cylindrical coordinate system, and physical information about pincushion distortion and S distortion were observed and analyzed; however, this method is not recommended due to a lack of calculational efficiency. In conclusion, the SG model provides an accurate, fast, and simple correction for rotational C-arm images, which may be used for cone-beam CT.
由于影像增强器(II)相对于地球磁场或其他杂散磁场的方向发生变化,其失真会随C形臂旋转角度而改变。对于锥束计算机断层扫描(CT)而言,对所有角度进行失真校正至关重要。新的超全局失真校正由一个模型组成,该模型不仅能在图像的每个位置连续校正II失真,还能针对C形臂的每个旋转角度进行校正。使用标准C形臂在9英寸II模式下采集校准珠图像。超全局(SG)模型是通过对选定校准图像以给定采样角度间隔进行单平面全局校正而获得的。五阶单平面全局校正产生的均方根(rms)残余误差为0.20像素,而SG模型产生的rms误差为0.21像素,差异极小可忽略不计。我们评估了SG模型对各种因素的精度依赖性,如单平面全局拟合阶数、SG阶数和角度采样间隔。我们发现,基于五阶单平面全局校正使用六阶SG多项式拟合可获得良好的SG模型,且10度的采样间隔就足够了。因此,SG模型节省了处理资源和存储空间。还研究了X射线系统机械误差产生的残余误差,发现其与SG残余误差相当。此外,在圆柱坐标系中进行了单平面全局校正,并观察和分析了枕形失真和S形失真的物理信息;然而,由于计算效率不足,不推荐使用此方法。总之,SG模型为旋转C形臂图像提供了准确、快速且简单的校正,可用于锥束CT。