Abstract—Cervical Cancer (CC) remains the fourth most typical cancer internationally. Whole-Slide Images (WSIs) remain a significant benchmark for CC prognosis. Missed prognoses and mis- prognoses frequently happen because of the higher similarity in pathological cervical images, the higher quantity of readings, the lengthy reading duration and the pathologists inadequate experience degrees. The prevailing paradigms possess inadequate Feature Extraction (FE) and portrayal abilities and they are burdened with inadequate pathological classification. Hence, this study initially proffers a new FE network called NASNet alongside a genetic algorithm- based feature selection procedure. Next, the chosen features will be sent as input into the ensemble classifier to classify 4 classes—Negative for Intra-Epithelial Malignancy (NILM), Squamous Cell Carcinoma (SCC), Low Squamous Intra-Epithelial Lesion (LSIL) and High Squamous Intra-Epithelial Lesion (HSIL). The database will be split into a training set (90%) and a test set (10%). The proffered network is called Genetic NASNet Ensemble Classifier (GenNASNet_EC) and is correlated with the prevailing methodologies concerning the Accuracy, Precision, Recall, Specificity, FPR, FNR and F1-Score. Consequently, it is observed that the proffered GenNASNet_EC attains the Accuracy of 98.02%, Precision of 97.56%, Recall of 98.02%, Specificity of 99.34%, FPR of 0.0066%, FNR of 0.0198% and AUC of 98.35%.
 
Keywords—cervical cancer, squamous cell carcinoma, high squamous intra-epithelial lesion, genetic NASNet ensemble classifier, pathological cervical images


Copyright © 2023 by the authors. This is an open access article distributed under the Creative Commons Attribution License (CC BY-NC-ND 4.0), which permits use, distribution and reproduction in any medium, provided that the article is properly cited, the use is non-commercial and no modifications or adaptations are made.