A Case Study of Cluster Analysis of the Whole Slide Images

Machine learning has achieved great success in image recognition and has been widely used in many fields, especially the data mining of massive image features in the era of big data. Clustering analysis is an important task in machine learning. This paper takes clustering in pathological section image recognition as an application case and discusses the application effects of different clustering models. Firstly, the relevant algorithms used are introduced, including K-means and Gaussian mixture model, hierarchical clustering algorithm and Louvain algorithm. Secondly, it is very important to evaluate the effectiveness of clustering results. In this paper, Silhouette score and V-measure are selected as indicators to evaluate the clustering algorithm, that is, the clustering methods determined by different data structures are not suitable for various data models. The cluster analysis of 5000 colorectal cancer tissue patches was carried out by python programming. Through the analysis and discussion of clustering results, the goodness of fit and clustering results of different models are compared and analyzed.