中文摘要：

针对传统FCM算法对脑肿瘤图像分割时对噪声敏感且没有考虑图像空间信息出现分割结果不准确的问题,提出一种新的基于空间模糊核聚类（Spatial Kernel Fuzzy C-means,SKFCM）的脑肿瘤MR图像分割方法。首先,算法引入核函数,将数据由低维特征空间转化为高维特征空间,提高了不同类别特征差异。其次,根据图像目标区域梯度差异,对KFCM算法进行改进,将图像空间信息引入到KFCM算法,用目标区域方差总和与边界梯度均值倒数之和作为新的目标函数。最后,对脑肿瘤图像的样本特征进行迭代优化,实现了脑肿瘤的精确分割。算法分割精确度得到了提高,并且通过图像的空间信息增加了方法的鲁棒性。实验结果表明,改进后算法对脑肿瘤MR图像不同类别间区分度高且具有较强噪声抑制能力,实现了较好的分割效果。

英文摘要：

Aiming at the problem of the inaccuracy of spilting due to the sensitiveness to noise and no consideration of the spatial information when segmenting the brain tumor images by the traditional FCM algorithm, a new segmentation method based on spatial fuzzy kernel clustering for brain tumor MR image is proposed. Firstly, the kernel function is introduced into the algorithm which transforms the data from the low dimensional feature space into the high dimensional feature space, and the difference of category features is enhanced. Then, the KFCM algorithm is improved based on the difference of the image target area gradient. The spatial information of images is introduced into the KFCM, and the sum of the target area variance and the sum of the borderline gradient mean reciprocal are used as the new objective function. Finally, the sample characteristics of brain tumor images are optimized iteratively to achieve the accurate segmentation of brain tumor. The precision and robustness of the algorithm have been improved by increasing the space information of the image. The experimental results show that the improved algorithm has high degree of differentiation in different categories and strong noise suppression capability for brain tumor MR images, and achieves good segmentation effect.