Digital image verification for clinical quality assurance in radiation therapy

(1)为了对“安装误差”进行基础研究，我们通过观察考察了一次检查的6维(X、Y、Z、俯仰、侧滚、偏航)检测能力。我们使用了头部模型的模拟胶片和门户胶片。旋转比移动更难被检测到。在旋转时，偏航相对容易被检测出来，因为它的旋转轴是一条垂直于胶片平面的直线。(2)研究了对模拟胶片和射孔胶片进行合适的图像处理的方法，以检测骨骼结构的标志。通过傅立叶变换和高通滤波消除了软组织对基频偏差的影响。利用小型金属标记物进行扇束投影的放大率校正。(3)以骨骼结构的标志点为参照点，确定照射野。在侧视(左右)中，以耳道作为骨骼标志，检测摇摆和偏航的设置误差。如果两个耳道都能被识别，它们不仅是有用的，而且是金属标记物。(4)最后，在临床使用方面，我们通过模拟胶片和射野胶片测量了15例全脑照射的临床病例的设置误差。人类有时很难识别骨骼结构的标志。在没有金属标记的情况下，机器识别更难检测到设置错误。

(1) For basic research of 'setup error', we investigated the 6 dimensional (X, Y, Z, pitch, roll, yaw) detection ability of an inspection by watching. We use the simulation films and portal films of a head phantom. Rotations were more difficult to be detected than Shifts. And in rotations, yawing was comparatively easy to be detected than other to rotation, because it's axis of rotation was a perpendicular line of film plane.(2) We study the way of appropriate image-processing for simulation films and portal films to detect landmark of skeletal structures. The defferences of basical bias because of soft tissues were eliminated by Fourier transformation and high-pass filter. We use small metal marker for calibration of correction of magnification by fan-beam projection.(3) We discuss the landmark of skeletal structures as reference point to determine the fields of irradiation. In side-view ( left-right), auditory canals were usefull as skeletal landmark to detect the setup errors of rolling and yawing. If both auditory canals were recognized, they were usefull as well as metal markers.(4) Finally, for clinical use, we measured setup errors about fifteen clinical cases who were whole brain irradiated, through simulation films and portal films. It was sometimes difficult to recognize the landmarks of skeletal structures by human. And to detect setup errors were more difficult for machine recognize without some metal markers.