改进FCMS算法及其在颅内肿瘤图像分割的研究

为了提高脑部肿瘤的磁共振成像(MRI)在肿瘤分割方面的精度和分割效率,提出了自适应阈值蚁群模糊聚类算法(TSAG_PnFCMS)。针对传统的模糊c均值聚类(FCMS)算法对噪声敏感,以及MRI图像中存在属性不同的样本点,在聚类过程中,将不同属性样本点的相关系数作为权重融入到欧氏距离的计算,提高聚类精度;针对蚁群算法容易陷入局部最优,提出一种自适应阈值蚁群算法,提高算法的全局搜索能力,将自适应阈值蚁群算法与改进的模糊聚类算法相结合,提高系统的分割精度和抗噪声性能,使得最终的分割效果达到最优。通过轮廓系数、目标函数收敛结果以及迭代时间进行实验仿真对比,表明改进算法的有效性,可见算法为颅内肿瘤图像的分割提供了可靠的技术手段。

Improved FCMS algorithm and its study on intracranial tumor image segmentation

An adaptive threshold ant colony fuzzy clustering algorithm (TSAG_PnFCMS) was proposed to improve the accuracy and segmentation efficiency of MRI in tumor segmentation. In view of the traditional FCMS algorithm"s sensitivity to noise and the existence of sample points with different attributes in MRI images, in the clustering process, the correlation coefficients of sample points with different attributes are used as weights in the calculation of Euclidean distance to improve the clustering accuracy. An adaptive threshold ant colony algorithm is proposed to improve the global search ability of the algorithm. The adaptive threshold ant colony algorithm is combined with the improved fuzzy clustering algorithm to improve the segmentation accuracy and anti-noise performance of the system, so as to achieve the optimal segmentation effect. The simulation results of contour coefficient, convergence result of target function and iteration time show the effectiveness of the proposed algorithm, which provides a reliable technical means for intracranial tumor image segmentation.