Advanced DEEPCNN breast cancer mammogram image detection and classification with butterfly optimisation algorithm

Abstract

A major aspect influencing human health is breast cancer. Mammography, fine needle aspiration, and surgical biopsy are some of the evolving diagnosis methods for this problem. Pathology images are used to diagnose breast cancer. Breast tumour surgery allows doctors to microscopically study breast tissue. Traditional methods use a cuckoo-optimised radial basis neural network. Earlier RBN algorithms handled feature extraction and reduction differently. To reduce unneeded complexity, outperform convolutional neural network (CNN) for feature extraction and classification. The Butterfly optimisation technique suggests a CNN. Zernike moments' scale, interpretation, and rotation similarity lets us bypass numerous pre-processing steps. The picture dataset was created from tumour treatment archives. The Butterfly optimisation method feeds the DCNN training data. DCNN removes, reduces, and classifies features. By determining the number of historical periods and training images for Deep CNN, optimisation improves efficiency and reduces error rates. This approach projects normal, benign, and malignant. The model achieves sensitivity, accuracy, specificity, F1 score, and recall by contrasting RBF with cuckoo search.