中值滤波!先验图像法去除CBCT图像噪声

CBCT Image Noise Reduction Based on Combination of Median Filtering and Prior Image

为解决CBCT图像的噪声去除问题,提出了一种基于中值滤波与先验图像结合的CBCT图像去噪算法。该方法首先通过中值滤波对原始图像进行去噪得到中间图像,再根据加权模型计算权重因子,将中间图像与原始图像进行加权融合成最终图像。Shepp-logan仿真实验及真实CBCT图像去噪实验的结果均表明:本算法能明显降低图像噪声,并充分保护图像复杂组织结构的信息,在计算效率和去噪效果之间取得了较好平衡,与中值滤波相比图像SNR提高了5%～23%,能快速有效去除锥形束图像噪声。

To solve the problem of noise reduction of CBCT imagea method based on combination of median filtering and prior image is proposed in this paper. This method is performed firstly by filtering the noisy CBCT image using median filtering algorithm,followed by combining filtered image and prior CBCT image with weighting factor according to the designed weighting model. Experiments for Shepp-logan phantom cases and clinical CBCT image case have shown that the proposed method apparently reduces the image noise and maintain the complex structure of the image,leading to a good balance between calculation efficiency and de-noising performance. Compared to individual median filtering,signal-to-noise ratio of proposed method increases 5% ~ 23%,which indicates the high reduction efficiency of our method.