基于序列公共稀疏误差对称差异度分割的健壮脑肿瘤良恶性分类方法

The Method of Robust Brain Tumor Classification Based on Symmetric Difference Segmentation with Sequence Public Sparse Error

脑肿瘤已经成为全球范围内致死率最高的疾病之一。术前利用计算机辅助诊断技术对CT或MR图像进行辅助诊断也是当前各国研究的热点。利用序列图像克服单帧图像肿瘤区域信号局限,并通过脑左右半球间的公共稀疏误差对称差异度计算可以更加精准地获得脑肿瘤的显著性区域。进而在此基础上所提取的健壮形状及纹理特征能够对脑肿瘤良恶性进行更加准确的识别。最终实验也表明该方法具有良好的性能。

Brain cancer is one of the deadliest diseases in the world.The noninvasive diagnosis of CT or MR by computer aided diagnos.tic technique is also a hot topic.In the method of using sequential images to overcome the signal limitation of tumor region in single frame image,the significance region of brain tumor can be calculated more accurately through the symmetric differ.ence degree of sparsity error between the left and right hemispheres.Moreover,the robust shape and texture features extract.ed on this basis can more accurately identify the benign and malignant brain tumors.Finally,experiments show that this method has good classification performance.