血管内超声图像初始轮廓提取算法的研究

Research of Intravascular Ultrasound Image's Initial Contour Extraction Algorithm[JZ)]

摘要:

Snake模型在医学图像分割中的应用已经越来越广泛,但在应用该模型时,如何选取合适的初始轮廓是一个难题；在对血管内超声医学图像的研究基础上,提出了一种基于灰度信息与ROI区域的初始轮廓获取方法,根据IVUS图像的灰度特征对其进行自适应阈值分割以及面积滤波,然后获得分割轮廓点集进而得到snake初始轮廓点集；在matlab7.0环境分别对不同类别的2种IVUS图像的中外膜边缘提取进行仿真实验,实验证明该方法获得的初始snake轮廓较为逼近目标真实轮廓且适合于snake模型进行迭代收敛,由于其初始轮廓已较为接近目标真实轮廓,因此节省了snake模型的迭代次数,算法运行效率也优于手工提取初始轮廓的snake方法,可以较为方便的应用于实际领域。

Abstract:

Snake model in medical image segmentation has become increasingly widespread, but in the application of the model, how to select the appropriate initial contour is a problem. In this paper, we propose a method to extract the ROI’s initial contour based on the gray information of IVUS images .According to IVUS image’s grayscale features, firstly, adaptive thresholding and area filtering are used to reduce the complexity of the IVUS original images, then get splited contour points set and finally get the snake’s initial contour points set.Two kinds of IVUS images’ outer membrane extraction simulation experiments in Matlab7.0 environment prove that the initial snake contour obtained by this method is suitable for snake model’s iteration, as its initial contour is close to the true outline of the target, it saves more time and improve the efficiency of the algorithm compared with manual extraction method, it can be more easily applied to practical fields.