Mammograms Classification Using ELM Based on Improved Sunflower Optimization Algorithm

Abstract

Abstract To assist specialists in detecting breast cancer on mammograms with better accuracy and less time consuming, this paper proposes an approach based on improved sunflower optimization algorithm (ISFO) and extreme learning machine (ELM). Firstly, features were extracted by using lifting scheme and gray-level co-occurrence matrix (GLCM). Then, the parameters of ELM were optimized by (ISFO) to obtain the final classification results. Finally, in order to avoid overfitting, the proposed model’s performance was evaluated with k-fold random stratified cross validation, and the experiments compared the model with other models on MIAS datasets. The experimental results show that the proposed model has higher classification accuracy, shorter learning time and stronger robustness on mammograms classification task. Thus, this method could be a promising application in bio-medical and provide a basis for the early diagnosis of breast cancer.