基于Top-hat变换和角点检测的MRI脊椎分割算法  

SEGMENTATION OF SPINAL MR IMAGES BASED ON TOP-HAT TRANSFORMATION AND CORNER DETECTION

作　　者：钱鹰[1] 樊凡[1] 韦庆杰[1] 

机构地区：[1]重庆邮电大学图形图像与多媒体实验室,重庆400065

出　　处：《计算机应用与软件》2014年第7期201-205,共5页Computer Applications and Software

基　　金：国家自然科学基金项目(61171060);教育部留学回国人员科研启动基金项目(教外司留(2009)1509)

摘　　要：脊椎骨骼提取的精度直接影响到脊椎三维重建的效果。针对MR(Magnetic Resonance)脊椎图像成像不均匀的特点,提出一种利用Top-hat变换和角点检测结合的脊椎骨骼提取算法。首先利用Top-hat变换增强脊椎块与周围组织间的对比度,进而用大津法(OTSU)分割图像,获得脊柱的粗分割结果;然后使用改进的CSS(Curvature Scale Space)角点检测算法作用于粗分割图像,获取图像的有效角点;最后通过拟合角点,得到脊柱两侧的曲线,从而分割出脊柱区域。实验表明,该方法可以精确分割出脊椎部分,而且分割的结果可以应用于同一病人同序列的其他图像中,缩短了分割时间,具有较好的应用前景。Precision of spinal bones extraction directly affects the effect of three-dimensional spine reconstruction. Aiming at the characteristics of magnetic resonance （MR） spine imaging uneven, we present a spinal bones extraction algorithm which uses the combination of Top-hat transformation and comer detection. First, the Top-hat transformation is used to enhance the contrast between the spinal piece and the surrounding tissue. Then the OTSU algorithm is used to segment the enhanced image, and to obtain rough segmentation image. Thirdly, the improved curvature scale space （CSS） corner detection method is applied to rough segmentation image to get effective comer point of the spine. Lastly, the curves on both sides of the spine are obtained by fitting the corner, so that the spine part is segmented. Experiments show that this method can accurately segment the spinal part, and the segmentation results can be used to other images of the same patient with same sequence. This method shortens the segmentation time and has a good applied prospect.

关 键 词：脊椎分割 TOP-HAT变换 CSS角点检测 大津法 模板分割 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]