In recent years,the segmentation of MRI brain images is becoming a new hotspot of medical image processing. The segmenta-tion of MRI brain images is a necessary preparation for image analysis. How well the image analysis,to some extent,depends on the accu-racy of image segmentation. So,accurate image segmentation is important for clinical diagnosis. Scalp,skull,muscle,blood vessels and other non-brain tissues contained in the brain MRI image will seriously affect the accuracy of segmentation of brain tissues. In this paper, strip skull from the MRI brain images with Otsu and morphological operators. Firstly,a method using the anisotropic diffusion equation was applied to suppress the noise of the MRI image. After that,the image was converted to binary image and then a series of morphologi-cal processing were applied to the binary image. Finally,the obtained binary brain tissue was mapped back to the original image. The test result has proved that the algorithm has a good performance in image segmentation.%近年来,脑部MRI图像分割问题渐渐成为医学图像处理领域一个新的研究热点. 脑部磁共振图像分割是对图像分析的必要准备,能否很好地对图像进行分析,一定程度上取决于图像分割的准确性. 由此可知,图像的准确分割对医学临床诊断有着重要意义. 然而,脑部MRI图像中的头皮、颅骨、肌肉、血管等非脑组织会严重影响脑组织的准确分割. 文中采用最大类间方差法和形态学算子剥离脑部MRI图像的颅骨部分. 首先,利用一种基于各向异性扩散方程的方法抑制MRI图像的噪声. 然后,运用最大类间方差法将图像转化为二值图,接着对二值图进行一系列形态学处理,最后,将得到的二值化脑组织模板映射到原图像. 实验结果表明该算法的分割效果良好.
展开▼
