Modeling of Aquila Optimizer with Hybrid ResNet-DenseNet enabled Breast Cancer Classification on Histopathological Images

Abstract

Breast cancer (BC) is the most common cancer amongst women that threatens the health of women, initial diagnosis of BC becomes essential. Though there were several means to diagnose BC, the standard way is pathological analysis. Precise diagnosis of BC necessitates experienced histopathologists and needs more effort and time for completing this task. Recently, machine learning (ML) was successfully implemented in text classification, image recognition, and object recognition. With the emergence of computer aided diagnoses (CAD) technology, ML was effectively implemented for BC diagnosis. Histopathological image classification depends on deep learning (DL), particularly convolution neural network (CNN), which frequently needs a large amount of labelled training models, whereas the labelled data was hard to obtain. This study develops an Aquila Optimizer(AO) with Hybrid ResNet-DenseNet Enabled Breast Cancer Classification on Histopathological Images (AOHRD-BC2HI). The proposed AOHRD-BC2HI technique inspects the histopathological images for the diagnosis of breast cancer. To accomplish this, the presented AOHRD-BC2HI technique uses hybridization of Resnet with Densenet (HRD) model for feature extraction. Moreover, the HRD method can be enforced for feature extracting procedure in which the DenseNet (feature value memory by concatenation) and ResNet (refinement of feature value by addition) were interpreted. For BC detection and classification, the DSAE model is utilized. The AO algorithm is exploited to improve the detection performance of DSAE model. The experimental validation of the presented AOHRD-BC2HI approach is tested using benchmark dataset and the results are investigated under distinct measures.Also the proposed model achieved the accuracy of 96%. The comparative result reports the improved performance of the presented AOHRD-BC2HI technique over other recent methods.