Breast Histopathological Whole Slide Image For Retrieval Using Latent Dirichlet Allocation

Whole Slide Image (WSI) has become the major carrier of visual and diagnostic information. Recovery of content-based images among WSIs may help to diagnose an unknown pathological image by finding its similar regions in WSIs with diagnostic information based on deep learning and local morphology nuclei statistical function. This research work implements the breast histopathological image approach, extracting the features such as Gabor and LSFN from the given image and splits the region based on the threshold (detha, beta). Based on threshold value, the matched regions are selected called candidate region. Then, the LDA model is utilized to obtain a histogram for each feature of every region. For this training and testing images are utilized. In this proposed system focused on clustering method such as NN- clustering algorithm that has been widely used for medical image segmentation. Local statistical feature of nuclei (LSFN) is presented to describe morphology and distribution pattern of nuclei is utilized for texture information. The algorithms have been implemented and tested on WSI images. The comparison is made with existing conventional NN-clustering method