1、 多分类支持向量机论文：支持向量机在脑部 MRI 图像微小多目标分割的应用【中文摘要】在大脑磁共振成像(Magnetic Resonance Imaging, MRI)图像中,脑组织的轮廓非常复杂和不规则,且样本数目有限,不适合使用传统的基于经验风险最小化的分割方法。而支持向量机(Support Vector Machines, SVM)是基于统计学习理论发展起来的一种有监督的分类方法,它根据结构风险最小化原则,在解决小样本、非线性及高维模式识别问题中表现出许多特有的优势。因此,本文开展利用支持向量机对磁共振图像脑部基底节区内的尾状核(caudatum)、壳核(putamen)和苍白球(pal
2、lidum)进行分割研究。支持向量机最初用于二分类问题,在此采用有向非循环图的方法将若干个两类分类器组合成多类分类器。这样所得到的多分类支持向量机的方法可以很好地实现对 MRI 图像中的尾状核、壳核、苍白球及其背景区域的分割。最终分类的效果除了与设计的分类器有关外,还与从磁共振图像中提取的特征向量有关。结合磁共振图像的特点,采用纹理特征提取和灰度特征提取的方法,纹理提取方法利用灰度共生矩阵来提取图像的局部统计特征。每一个样本点共提取 58 维特征向量。由于高维的图像特征向量严重影响计算速度,降低分割速度,所以在本文中分别采取主成分分析(Principal Component Analysis,
3、 PCA)和粗糙集(Rough Sets, RS)的方法来进行降维处理,大大提高了分割速度。实验结果表明,在分割 MRI 脑部多目标组织时,无论是分割速度还是分类准确率,粗糙集方法均优于主成分分析算法。为了分析和验证所提出的支持向量机分割算法的实际效果,同时采用 K-均值聚类(K-Means Clustering)算法、模糊 C-均值聚类(Fuzzy C-Mean, FCM)、K-最近邻(K-Nearest Neighbor, KNN)算法、贝叶斯分类器(Bayes Classifier, BC)算法和径向基神经网络(Radial Basis Function Neural Network,R
4、BFNN)算法分别来对目标区域进行自动分割,从而有利于非常客观地说明支持向量机分割算法的优越性。最后,根据虚警、漏警概率以及分类正确率指标对这六种方法的分类效果进行对比分析。实验结果表明无论是否采用降维处理,采用多分类支持向量机进行多目标分割的分割正确率均优于以上五种方法。【英文摘要】The boundary of encephalic tissue is highly complicated and irregular in head magnetic resonance image, and the number of samples is limited. Its the reason
5、that the traditional segmentation methods based on the empirical risk minimization is not suitable. Support vector machine based on statistical learning theory is a supervised classification method, which follows the structural risk minimization principle, shows many special advantages in resolving
6、the small sample set, nonlinear and high dimensional pattern recognition problems. Therefore, this paper carries out the research of support vector machine to segment caudatum, putamen and pallidum region in brain magnetic resonance imaging(MRI).Support vector machine was originally used for two-cla
7、ssification, a multi-classification classifier can be constructed by a few two-classification classifiers using directed acyclic graph. And the multi-classification classifiers can well segment caudatum, putamen, pallidum and background region of the MRI image.In addition to the classifier, the fina
8、l classification effect also has an important relationship with the feature vector extracted from the brain MRI images. The texture features and gray features are extracted as the feature vectors in the experiments, Texture features are extracted from gray co-occurrence matrix. The total dimensional
9、 number of each sample point is 58.Since the high dimensional feature vectors seriously impact the calculation speed, and reduce the segmentation speed, the principal component analysis and the rough sets are adopted respectively to reduce the dimension of feature vectors. A great deal of experiment
10、s shows that rough set is better than principal component analysis in the speed and the result of the segmentation.In order to analyze and verify the actual effect of the proposed segmentation algorithm based on SVM, k-means clustering, fuzzy c-mean segmentation, k-nearest neighbor, Bayes classifier
11、, and radial basis function neural network are respectively adopted. The false alarm probability, false dismissal probability and the segmentation accuracy are used as objective indicators. The comparison analysis can objectively indicate the validity of the proposed segmentation algorithm. Experime
12、ntal results show that whether or not to adopt dimensional reduction processing, the proposed segmentation algorithm is better than the five methods above.【关键词】多分类支持向量机 磁共振图像 图像分割 主成分分析 粗糙集【英文关键词】multi-class support vector machine magnetic resonance imaging image segmentation principal component ana
13、lysis rough sets【目录】支持向量机在脑部 MRI 图像微小多目标分割的应用 摘要 8-9 Abstract 9-10 第 1 章 绪论 11-15 1.1 医学图像分割技术的发展及研究现状 11-12 1.2 支持向量机技术的发展 12-13 1.3 本文研究工作的动机与意义 13-15 第 2 章 基于统计学习理论的支持向量机算法 15-21 2.1 二分类支持向量机 15-18 2.1.1 线性可分 16-17 2.1.2线性不可分 17-18 2.2 参数优化 18-21 2.2.1 试凑法 18-19 2.2.2 交义验证法 19-21 第 3 章 基于多分类支持向量机
14、的脑部 MRI 图像多目标分割方法 21-29 3.1 多分类支持向量机算法 21-23 3.2特征向量提取 23-24 3.3 降维处理方法 24-29 3.3.1 主成分分析算法 24-26 3.3.2 粗糙集算法 26-27 3.3.3 降维方法效果对比 27-29 第4 章 脑部 MRI 图像多目标分割效果对比分析 29-40 4.1常规分割算法 29-34 4.1.1 K-均值聚类分割算法 29-30 4.1.2 模糊 C-均值聚类分割算法 30-32 4.1.3 K-最近邻分割算法 32 4.1.4 贝叶斯分类器分割算法 32-33 4.1.5 径向基神经网络分割算法 33-34 4.2 分割效果对比 34-40 4.2.1 未降维分割效果 36-37 4.2.2 主成分分析降维后的分割效果 37-38 4.2.3 粗糙集降维后的分割效果 38-40 第 5 章 总结与展望 40-42 5.1 课题的主要研究工作 40-41 5.2 研究工作展望 41-42 参考文献 42-44 致谢 44-45 攻读学位期间发表的学术论文及参与的项目 45-46 学位论文评阅及答辩情况表 46
