基于近似密度函数的医学图像聚类分析研究

Research on Medical Image Clustering Based on Approximate Density Function

摘要: 针对医学图像数据难以用数学模型来表述和聚类的问题，提出一种基于近似密度函数的医学图像聚类分析方法.该方法采用核密度估计模型来构造近似密度函数，利用爬山策略来提取聚类模式.基于真实的人体腹部医学图像数据集的实验结果表明，该方法可以取得较好的聚类效果.

Abstract: It is difficult to represent and cluster medical image data by mathematic model. In order to address this problem, an medical image clustering analysis method based on approximate density function is designed. This method uses kernel density estimation model to construct the approximate density function, and takes hill climbing strategy to extract clustering patterns. Results of experiments show that it can achieve good effect on real human abdomen medical images.