小波在基于CT图像的肝脏疾病良恶性分类中的研究

The Research of Wavelet in Classification of Benign and Malignant Liver Tumor in CT Images

肝癌是一种常见的恶性肿瘤,近年来发病率呈缓慢上升的趋势,病死率也随之上升。文章利用小波在特征提取和模式识别方面的独特优势,提取了基于小波和灰度共生矩阵的纹理特征,结合遗传算法进行特征选择和优化,用KNN分类器设计出高精确度的肝脏疾病良恶性分类器。采用肝脏CT平扫图像,将肝癌与其他的良性病变进行分类,探讨了小波的不同性质及特征提取方式对分类结果的影响,对小波在肝脏CT图像良恶性分类中的研究有指导意义。

Hepatocellular carcinoma（HCC） is a common malignant rumor, in recent years, and the rate of liver diseases was slowly rising, and the mortality rate also rise. In this paper, by using the unique advantage of wavelet in feature extraction and pattern recognition, the texture features based on wavelet and spatial gray level co-occurance matrix are extracted. Combining the genetic algorithm to select and optimize the features, KNN classifier is used to define an optimal computer-aided diagnosis （CAD） architecture for the classification of liver tissue from non-enhanced computed tomography （CT） images into HCC and benign tumor. This paper explores the effect of the different characters of wavelet and feature extraction methods to the classification results, this paper gives guidance to the research of wavelet in classification of benign and malignant liver tumor in CT images.