Towards Precision Medicine: Deep Learning Models for Cervical Cancer Detection and Prevention

Cervical cancer poses a significant danger to the lives of women globally and with the changing lifestyle choices leading to a rise in the number of detected cases, early detection and treatment are imperative for combating this disease, particularly in developing countries. Cervical cancer often presents as squamous cell carcinoma or adenocarcinoma. The development of more accurate diagnostic and therapeutic models to address this challenge has been greatly helped by the additional input of Machine Learning. Literature suggests that the use of magnetic resonance imaging (MRI) in conjunction with convolutional neural networks (CNN) enhances the efficiency of cervical cancer detection and prevention. These approaches enable the identification of cancerous cells, with specialized models such as CNN. However, the computational demands of image processing and preprocessing pose significant challenges. In response to these challenges, our research proposes an efficient deep-learning model tailored for cervical cancer detection in high-resolution cervical images. This model incorporates three distinct feature selection modules—low-variance feature extraction, - dimensional feature selection, and recursive feature elimination— while integrating multiple views of cervical images. We evaluated various classification models, including ResNet, DenseNet, and VGG, for cervical cancer diagnosis. Simulation results are considered to validate the proposed model's accuracy. Using AI and ML techniques, our model aims to contribute to advancing cervical cancer diagnosis and treatment, ultimately improving the efficiency in treating cervical cancer.