基于CT图像的全心脏分割方法研究

Research on an Improved Localization Method of Whole Heart in CT Images

为了解决无法准确分割全心脏的难题,提出一种基于形态学操作和形态学梯度的阈值分割算法。传统的阈值分割由于图像本身灰度分布不均及噪声干扰等多种因素的影响,往往不能得到理想的分割结果。该算法结合形态学开闭操作重构图像,在形态学梯度的基础上,对图像进行自动阈值分割,不但可以消除噪声,还能较好地保留图像的边缘信息,得到全心脏组织。实验结果显示,基于形态学梯度的阈值分割算法对心脏的提取准确率较高,解决仅使用传统阈值分割或直接对普通梯度图像分割存在的问题。

To solve the problems of accurate location of whole heart in cardiac CT images, proposes an improved threshold segmentation algorithm based on morphological operation and morphological gradient. Because of the influence of image gray distribution is uneven and the noise interference factors, traditional threshold segmentation often cannot get ideal segmentation results. Combined with the morphological opening and closing operation, reconstructs the images, the cardiac images are automatically segmented, which can not only eliminate the image noise, but also better preserves the edge information of the CT images, and gets whole heart finally. The experimental results show that the method has improved segmentation accuracy of cardiac CT images. It can solve the problem of the threshold segmentation or threshold segmentation based on general gradient only used.