基于Ncut分割和SVM分类器的医学图像分类算法

为解决医疗诊断中由于疲劳和主观因素影响导致的诊断错误,本文提出了基于Ncut分割方法的医学CT图像的分割、特征提取和诊断的新方案.将Ncut分割方法应用于脑CT图像.先进行图像分割,提取感兴趣区域,再从边缘、灰度,纹理三方面提取特征,最后利用支持向量机(SVM)对图像进行分类,为医生的诊断提供参考.从表格化的分类结果看,所提方案有较大的应用价值.

Ncut-Based Segmentation and SVM Classifier for Medical Image Classification

To identify tumor part on a computer tomography(CT)image,this paper proposes a novel computer aided diagnosis(CAD)scheme based on normalized cut(Ncut)image segmentation and support vector machine(SVM)classifier.Firstly,the Ncut segmentation method is used to perform the segmentation and to obtain the region of interest(ROI).Then,such image features like histogram,gray level co-occurrence matrix to construct the feature space are extracted.Finally,SVM classifier is trained and used to perform the classification.The classification results show that the new scheme can provide useful help for better diagnosis.Thus,the method can solve the problem of medical diagnosis error caused by the human fatigue and the subjective factor.