基于Tamura纹理特征提取和SVM的多模态脑肿瘤MR图像分割

Brain Tumor Segmentation on Multi-Modality Magnetic Resonance Images Based on Tamura Texture Feature and SVM Model

在Tamura纹理特征和支持向量机(SVM)算法基础上提出一种多模态脑肿瘤图像分割算法.将4种模态下的多序列核磁共振图像(MRI)的局部灰度特征与Tamura纹理度量相结合,尽可能提取足够多的图像信息;在SVM模型中输入已知样本并进行训练;用训练好的SVM模型处理其他脑肿瘤图像.实验通过对20例患者的图像进行展开,从实验数据来看,提出的方法可以精准有效地分割出脑肿瘤区域,得到脑肿瘤的边界,并且对脑肿瘤图像的差异性表现出较强的自适应能力.

A segmentation algorithm of brain tumor MR image based on Tamura texture feature and SVM model is proposed.The local grayscale features of four modal MR images(MRI)are combined with the Tamura texture metrics in the algorithm.The information in the image is extracted as much as possible.The known samples are input into the SVM model and classifier training is performed.The other brain tumor images are processed with the trained SVM model.The experiments was performed on the images of 20 patients.From the obtained data,the method proposed can segment the brain tumor region accurately,effectively and show strong self-adaptability to the difference of the brain tumor images.