中文摘要：

多模态医学图像的配准在医学诊断和治疗计划中起着重要的作用，特定区域轮廓提取是图像配准的重要基础，常规的轮廓提取算法不能满足多模态图像配准的要求。提出了一种用于多模态医学图像处理的特定区域轮廓提取算法。该方法首先对图像进行预处理，去除噪音，对图像进行平滑，采用Canny算子进行边缘提取之后，提出了延伸权的概念，并采用一种新的基于延伸权概念的边缘连接算法对边缘结果进行连接，最后采用填充算法获得特定的辅助区域并在该区域的基础上向外或者向内获取特定区域的轮廓。实验结果表明，该方法能够满足多模态图像配准的要求。

英文摘要：

Multirnodality medical image registration plays an important role in medical diagnoses. The contour extraction of special region is the base of image registration, however,general contour extraction algorithms can＇t achieve the satisfactory performances for multimodality image registration. This paper approves a novel contour extraction algorithm of special region to resolve the problems in multi - mode medical image registration. Firstly the algorithm processes some pretreatments which include eliminating noises and smoothing to the original image, and uses the arithmetic operators of Canny to detect the edge. Then it approves a new concept Of extend weight and uses a novel edge linking algorithm to deal with the edge information. Finally it uses seed filling algorithm to obtain the special assistant region,then based on that forth or entad to detect the contour of the special region. Experiments illustrate that this algorithm can achieve a satisfactory performance of multi - mode medical image registration.