An Enhanced Feature Fusion Network Model for Cervical Cancer Prediction
B. Pragatheeswari, Jameela Mary, S. Roshni Priya, C.R. Dhivyaa · 2024
Healthcare providers and researchers that work with patients who have cervical cancer face a significant challenge because it is one of the world's most common causes of death. Worldwide, cervical cancer is one of the major causes of death among women according to the World Health Organization (WHO). Pap smear, also called the Pap Test, is used to examine the cells present in the cervix for abnormal cells. However, manual examination of these pap smears is prone to human error. This work aims to explore the use of deep learning models to detect cancer from pap smear images and develop an ensemble model to improve the accuracy. Our proposed model uses pretrained CNN models through transfer learning to extract features, perform feature concatenation followed by dimensionality reduction and hyperparameter tuning, which are then fed into a neural network for multiclass classification. Hyperparameters are fine-tuned with the help of Bayesian optimization to improve the model's performance. Our proposed model has achieved an accuracy of 96.4%, sensitivity of 96.41% and specificity of 96.66%.