A classification method of epithelial cells and clue cells based on multi-scale texture analysis
Yukun Guo, Xiangru Yu, Jinping Li · 2016
A novel method is proposed for the classification of epithelial cells and clue cells in wet mount microscopic images of vaginal secretions. Generally, there are many complicated ingredients in these images, and different from the gram stain method, no color information can be used. Obviously, the texture information of these cells becomes the main feature in the classification of these ingredients. After careful observation of difference of these two kinds of cells, we found the surface of clue cells is rougher. So, we extract texture features of those cells using the multi-scale texture energy descriptor which can reflect the difference of these two kinds of cells in different scales effectively. After that, the traditional SVM classifier is employed for the classification. The experimental results show that our method can effectively complete the classification task.