一种运用图熵的医学图像聚类方法

Medical Image Clustering Method Based on Graph Entropy

近些年,各种医学影像技术被广泛应用于临床诊断中.由于医院每天都会产生大量的医学图像,如何利用好这些医学图像,对其进行有效聚类,以此来辅助医生对疾病进行诊断,是当前医学图像数据挖掘领域所研究的热点问题之一.本文提出一种医学图像聚类方法,首先,将医学图像集抽象成一个带权无向完全图,之后对其进行稀疏化剪枝处理,以此来对医学图像之间进行更好的相似性描述,最后,又提出了一种运用图熵的带权无向图聚类方法,通过此方法来实现对医学图像的聚类.实验结果表明,本文所提出的聚类方法能够有效对医学图像进行聚类,并在时间损耗及聚类结果等方面表现良好.

Recently,a variety of medical imaging technologies have been used widely in clinical diagnosis. As a large number of medical images are produced everyday,it is a hot issue of data mining on medical image at the moment that howto make full use of these medical images and cluster efficiently to help doctors to diagnose. In this paper,a medical image clustering method is proposed. Firstly,medical image dataset is represented as a weighted,undirected and completed graph. Secondly,the graph is sparsified and pruned. This model can describe the similarity between medical images very well. Finally,weighted and undirected graph clustering method based on graph entropy is proposed to cluster these medical images. The experimental results showthat this method can cluster medical images efficiently and run well in time complexity and clustering results.