Diagnosis of Breast Cancer on Microscopic-Histopathology Images using Texture Analysis

Abstract

Cancer research is a major public health priority in the world due to its high incidence, diversity and mortality. Despite great advances in this field during recent decades, the high incidence and lack of specialists have proven that one of the major challenges is to achieve early diagnosis. Improved early diagnosis, especially in developing countries, plays a crucial role in timely treatment and patient survival. The main objective of this study was to diagnose breast cancer on microscopic images by means of image analysis using texture features and Linear Discriminant Analysis (LDA) as classifier. The algorithm was validated on BreakHis dataset and pathpedia images versus the ground truth. The results of this study showed: the normal breast tissue (Fat, connective tissue and glandular) was identified with accuracy of 99.4% .Then the breast tissues were classified as normal and carcinoma tissue with and accuracy of 98.5%. Also, carcinoma was classified into ductal carcinoma and lobular carcinoma and accuracy was (93.4 %), and accuracy was obtained 99.1% for classification of ductal carcinoma into general ductal carcinoma, mucinous or papillary. After that, the breast tissues on the microscopic images were classified into normal and benign tissue with the accuracy of 94.1%. Also benign tissues were classified into adenosis, fibroadenomas, phyllodes tumor and tubular adenoma and accuracy were 93.8% with ANN. Finally, this study showed that the proposed algorithm can be used to classify the breast tissues into normal, benign and carcinoma with an accuracy of 90.9%, 93.9% and 99.2% using LDA, Tree, and ANN respectively. The all results of this study are highest accuracy of classification tissues when compare with previous studies.