基于M带小波变换多重分形的胰腺内镜超声图像分类

Pancreatic Endoscopic Ultrasonography Image Classification Based on Multifractal Feature of M-band Wavelet Transform

提出胰腺内镜超声图像分形特征的提取与分类方法,用于胰腺内镜超声图像的计算机辅助诊断,以提高胰腺癌内镜超声早期诊断的准确性。通过改进基于分形维数的M带小波变换分形特征,引入多重分形维数并进行特征筛选,获得M带小波变换多重分形的特征矢量,采用贝叶斯分类器、支持向量机和AdaBoost等三种不同的分类器进行胰腺内镜超声图像的分类研究。实验表明:基于本研究分形特征矢量的分类,在运行时间和分类准确率上均优于基于传统分形特征的分类。此分类方法对胰腺内镜超声图像具有较高的分类准确性,有望为胰腺癌的临床诊断提供有价值的参考。

In order to enhance the accuracy of the early diagnosis for pancreatic cancer with pancreatic endoscopic uhrasonography （EUS）, a method for the fractal feature extraction and EUS image classification was proposed to be applied in the computer-aided diagnosis of pancreatic EUS images. By modifying the M-band wavelet transform fractal feature based on the fractal dimension, the multifractal dimension was presented with the feature selection to obtain the multifractal feature vector of M-band wavelet transform. Three different classifiers, namely Bayes classifier, Support Vector Machine and AdaBoost, were used to classify the pancreatic EUS images. Experimental results showed that the classification based on the proposed fractal feature outperformed those based on the traditional fractal feature in both executing time and classifying accuracy. The proposed method has the potential to provide physicians a valuable opinion on the diagnosis of pancreatic cancer.