作　　者： ; ; ; ;

机构地区： 郑州大学物理工程学院

出　　处： 《计算机应用与软件》 2010年第7期249-252,共4页

摘　　要： 肾小球滤过率GFR(glonerular filtration rate)是评价肾功能的重要指标,而肾动态图像中感兴趣区ROI(region of interest)的准确勾画又是影响GFR的重要因素。现有的肾动态定量分析软件的ROI仍需人工辅助勾画,主观性强。在综合了现有肾动态图像ROI勾画方法的基础上进行了ROI自动提取的相关研究。首先用自适应滤波器对数字图像进行滤波,减小图像的噪声;再用ostu法自动确定阈值形成二值图,接着用交替顺序滤波法对二值图像滤波,提取双肾轮廓;然后采用链码和与差的方法修补边界,使轮廓平滑;最后针对象素进行本底ROI设置,从而实现自动提取肾脏和本底ROI。 Glomerular Filtration Rate（GFR） is an important indicator in evaluating renal function,and exactly outlining the region of interest（ROI） in renal dynamic image is also an important factor affecting the GFR.Existing software for renal dynamic quantitative analysis still need manual assistant to outline when extracts the ROI,so is of strong subjective.On the basis of summarized existing ROI extraction methods of renal dynamic image,we had the relevant research on ROI automatic outlining.Firstly,adaptive filter is used to remove the noises in digital image.Secondly,ostu method is used to determine the threshold automatically and the binary image can be then obtained.Thirdly,after using alternating filter for smoothing the object in binary image,contour tracking method is used to extract the renal contours of both kidneys.Fourthly,the chain code sum and difference method is used to mend the image boundary to smooth the contour.Lastly,the background ROI is set in light of the pixels,thus,the automatic ROI extraction is realized.

关 键 词： 肾小球滤过率 图像分割 轮廓提取 形态学分析 链码差

领　　域： [自动化与计算机技术] [自动化与计算机技术]