脊椎骨拐点智能识别方法研究  

Research of Intelligent Identification for Corners of Vertebra

作　　者：岑梓源[1] 李彬[1] 田联房[1] 

机构地区：[1]华南理工大学自动化科学与工程学院,广州510640

出　　处：《医疗卫生装备》2011年第11期14-16,19,共4页Chinese Medical Equipment Journal

基　　金：教育部博士点基金资助项目(200805610018);粤港关键领域重点突破项目(佛山专项)(2010Z11);广东省教育部产学研结合项目(2009B090300057);广东省自然科学基金资助项目(S2011010005811);国家质检总局科研计划项目(2011IK078)

摘　　要：目的:提出与实现一种脊椎骨拐点的智能识别算法,以取代医生手工操作,提高处理效率和识别的准确度与精度。方法:针对脊椎图像存在模糊、灰度变化大和倾斜现象的特点,提出一种基于霍夫(Hough)变换和先验知识的脊椎骨拐点的智能识别方法。该方法首先采用基于Hough变换的倾斜校正算法解决部分图像存在倾斜的现象,然后采用边缘算子和形态学分割脊椎图像,最后对脊椎图像进行感兴趣区域提取和目标点的定位与识别。结果:通过实验,脊椎拐点边缘信息得到有效保留,并且在倾斜情况下,脊椎图像得到有效校正,脊椎骨拐点得到有效识别。结论:采用该文提出的算法,能有效识别出脊椎骨拐点,并具有较强的适应性。Objective To detect the positions of the corners of the vertebra automatically by using intelligent recognition technique,which assists doctors in the interpretation of X-ray photography and improves processing efficiency.Methods A method of intelligent identification for corners of vertebra based on Hough transform and priori knowledge was presented.Firstly,to solve some of the images with tilt problem,a tilt correction method based on Hough transform algorithm was presented;then edge operator and morphology method were used to segment the image;finally,regions of interest（ROIs） of the vertebra in the binary image was extracted and target points were detected.Results The experiment showed that the information of the corners of the vertebra could be kept,and the images with tilt problem could be corrected,and the positions of the corners of the vertebra could be also detected successfully.Conclusion The corners of the vertebra can be detected effectively and adaptively using the proposed method.

关 键 词：智能识别 脊椎骨拐点 霍夫变换 图像分割 先验知识 

分 类 号：R318.0[医药卫生—生物医学工程] TP391.41[医药卫生—基础医学]