Multi Resolution Based CAD for Histopathological Tissue Classification

Abstract

Cancer being one of the most fatal diseases, any research for the early diagnosis of the disease is very important. Nowadays, Malignancy in the breast is one among the significant reasons for death in women. Breast cancer detection using histopathology images is done by detecting abnormalities like masses, calcifications and change in the structure of nucleus. Breast cancer can be described as the malignant growth of cells in the breast which spread to other parts of the body if not properly treated. newline newlineIt has been noted that histopathology images can be used to detect cancer up to 90%. Computer Aided Diagnosis (CAD) helps to prevent and reduce the limitation in human observation. Histopathologist carefully studies the digital histology images after image production of the specimen is carried out. But detailed analysis of large number of slides manually is an intensive task which is not only time consuming but also require high labor force. The recognition of histopathological tissue patterns in a Whole Slide Image (WSI) of a CAD system provides a numerical support for diagnosis as well as constitutes an important atmosphere for quantitative second opinion. CAD along with digital image processing and artificial intelligence can be used to improve the accuracy of diagnosis of any disease. newline This work is on the accurate method for the early diagnosis of breast cancer from histopathology images. The BreaKHis _v1 data set is utilized in this work. The method includes pre-processing, post processing, K-means segmentation, feature vector extraction and classification. Intensity and the texture features of the histopathology images are taken out and combined with multi resolution features such as Wave atom feature, Wavelet and Contourlet feature. newline The proposed work suggests a technique for detecting breast cancer from Immuno Histo Chemistry (IHC) images having two main parts. newline newlineand#61607; Robust breast cancer detection using multi-resolution features newlineand#61607; Automatic Cell Nucleus Breast Cancer Detection System newline