A Deep Learning-Based Approach for Cervical Cancer Classification Using 3D CNN and Vision Transformer.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Oncology

Background:

• Cervical cancer poses a significant global health challenge, necessitating improved early detection methods.
• Current diagnostic approaches can benefit from advanced computational tools for enhanced accuracy and efficiency.

Purpose of the Study:

• To develop and evaluate a novel deep learning system for accurate cervical cancer classification.
• To leverage 3D Convolutional Neural Networks (CNN) and Vision Transformer (ViT) for robust feature extraction and classification.

Main Methods:

• A hybrid deep learning model combining 3D CNN for spatiotemporal feature extraction and Vision Transformer (ViT) for complex feature learning.
• Integration of a 3D Feature Pyramid Network (FPN) and Squeeze-and-Excitation (SE) block for feature refinement.
• Classification using a Kernel Extreme Learning Machine (KELM) with a Radial Basis Function (RBF) kernel.

Main Results:

• The proposed deep learning model achieved a high classification accuracy of 98.6% on cervical images.
• The system demonstrated effective feature extraction and recalibration, enhancing discriminative power.
• Simulation results validated the model's superiority in cervical cancer classification.

Conclusions:

• The developed deep learning system offers a highly accurate and effective method for cervical cancer classification.
• This model shows significant potential as a supportive diagnostic tool for medical experts in identifying cervical cancer.
• Early and accurate detection through AI can substantially improve patient outcomes in cervical cancer management.