脑部MRI图像头皮三维提取及重建  

3D scalp extraction and reconstruction of MRI brain images

作　　者：马阿敏[1] 杨荣骞[1] 宁海[1] 白红民[2] 黎丽华[1] 吴效明[1] 

机构地区：[1]华南理工大学生物科学与工程学院,广州510006 [2]广州军区广州总医院神经外科,广州510010

出　　处：《计算机应用》2013年第5期1439-1442,共4页journal of Computer Applications

基　　金：国家自然科学基金青年基金资助项目(81101130);广东省科技计划项目(2012B031800026);华南理工大学中央高校基本业务费重点项目(2012ZZ0095)

摘　　要：开颅手术中,为了满足手术导航系统进行快速、准确定位的要求,提出一种脑部核磁共振成像(MRI)图像三维头皮轮廓提取方法。首先采用各向异性扩散滤波方法对图像滤波,借助BrainSuite3医学软件获取脑标记图像;然后将图像中脑实质部分剔除,根据非脑组织图像计算头皮组织的分割阈值;再利用数学形态学处理二值图像获取头皮轮廓;最后结合目标灰度信息,用移动立方体(MC)算法进行三维重建。实验结果表明,该方法结合阈值、数学形态学和MC算法,能连续、光滑地提取出头皮外轮廓,并使精度达到亚像素级别。A new scalp extraction method for Magnetic Resonance Imaging （MRI） brain images in 3D was proposed for the requirements of image guided surgery on fast and accurate positioning in craniotomy. Firstly, anisotropic diffusion fihering was used to reduce noises, and the brain parenchyma image was got by the medical software of BrainSuite3. Afterwards, the brain parenchyma was removed and the threshold of scalp from non-brain-image was calculated. Thirdly, outside surface was extracted by mathematical morphology. Finally, combined with the gray value of scalp, the scalp was reconstructed by Marching Cube （MC） algorithm. The experiments show that this method can extract scalp effectively and the precision reaches sub-pixel＇s level by connecting threshold, mathematical morphology and MC algorithm.

关 键 词：核磁共振成像图像分割 各向异性扩散滤波 数学形态学 三维头皮提取 亚像素 移动立方体算法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]