梯度主动轮廓与曲波变换联合分割方法研究

Medical Image Segmentation Base on DCUT and Gradient Active Contour

随着医学技术不断发展,人们对肿瘤图像的分割要求日益提高,为了满足临床需要,提高医学图像分割的准确性,提出了一种基于梯度主动轮廓和第二代离散曲波变换(Discrete Curvelet Transform,以下简称:DCUT)的医学图像分割算法。该算法首先对医学数据进行离散曲波变换,获取增强后的医学数据,再利用Canny算子和形态学运算进行边缘检测,对处理后的数据利用梯度主动轮廓模型确定病灶区域的轮廓。本文选取了297组医学图像进行验证,实验结果表明:医学图像经过本算法处理后,边缘检测性能由传统算法的88.95%达到96.03%,分割位置的准确性得到进一步提高,目标边缘和轮廓提取更加清晰、稳定,有效提高了医学图像分割精确性。

With the continuous development of medical technology,segmentation of the tumor images requirements increasing. In order to meet the clinical needs,improve the accuracy of medical image segmentation,this paper presents a segmentation algorithm based on active contours and DCUT of medical image. The algorithm first of medical data are discrete curvelet transform, enhanced medical data acquisition,using Canny operator and morphological operations of edge detection,of processed data using the gradient vector flow active contour model to determine the focal areas of contour. This paper selected297 groups of medical images is verified. The experimental results show that: medical image after the algorithm segmentation, edge detection performance by 88. 95% of the traditional algorithm reach96. 03% and position accuracy of the segmentation can be further improved,the edge and the outline more clearly,effectively improve the medical image segmentation accuracy.