一种锥束CT中平板探测器输出图像校正方法

摘要

为了消除平板探测器缺陷在锥束CT重建切片中引起的伪影,提出一种投影图像校正方法.通过对投影图像进行暗场和增益不一致校正,推导了像元增益系数公式;对增益系数以及像元输出的期望和方差进行直方图分析以识别坏像素,根据坏像素分布特征设计相应的插补方案;通过局部屏蔽区域数据统计暗场波动幅度并逐行修正.实验结果表明,该方法使投影图像灰度分布均匀,像元输出波动和切片伪影得到有效抑制.%To eliminate reconstructive slice artifacts of cone beam CT caused by defects in flat panel detector, a calibration method is proposed for projected images. On the basis of dark image and gain non-uniformity calibration, the gain coefficient is derived and bad pixels are identified based on a histogram analysis of gain coefficient, mathematical expectation together with variance of output. Then an appropriate interpolation method is designed according to the distribution of bad pixels. For calibrating the unstable dark images, the data under a local lead shield is accumulated, and the fluctuations on each line is computed and corrected line by line. Experimental results show that our proposed method can homogenize the gray level of projected images and reduce output fluctuations and reconstructive artifacts effectively.