Deep Feature Extraction of Pap Smear Images Based on Convolutional Neural Network and Vision Transformer for Cervical Cancer Classification

Cervical cancer is a malignant disease that women commonly experience. This cancer can be prevented if screening is carried out early using the pap smear method. The pap smear technique yields a subjective diagnosis. An appropriate decision-making method is needed to overcome this obstacle, such as using a computer-based diagnosis method and applying machine learning. We apply a combination of deep feature extraction using transfer learning from convolutional neural network models and vision transformers to obtain local and global features. Local and global features can represent an image's more comprehensive variety of features. The combined features are then reduced using two steps, principal component analysis and linear discriminant analysis, to obtain a representation of the essential features of the data. The reduced features are then analyzed using several classifiers, including SVM, K-NN, MLP, and LR. The proposed framework was evaluated on three publicly accessible datasets, namely Herlev, Mendeley LBC, and SIPaKMeD, achieving classification accuracies of 97.83% (SVM, K-NN, MLP, and LR), 100% (SVM, K- NN, MLP, and LR), and 98.52% (SVM, K-NN, and LR) respectively.