Deep learning-based decision support system for cervical cancer identification in liquid-based cytology pap smears
Ghada Atteia1, Maali Alabdulhafith1, Hanaa A Abdallah1
1Department of Information Technology, College of Computer and Information Sciences, Princess Nourah bint Abdulrahman University, P.O. Box 84428, Riyadh 11671, Saudi Arabia.
Summary
A new AI system accurately detects cervical cancer from Pap smear images using deep learning. This advanced computer-assisted diagnosis significantly improves early detection rates and reduces diagnostic time for women's cancer.
Area of Science:
- Oncology
- Medical Imaging
- Artificial Intelligence
Background:
- Cervical cancer, a leading cause of cancer death in women, is primarily caused by persistent human papillomavirus infection.
- Liquid-based cytology (LBC) is standard for early detection, but microscopic evaluation is time-consuming and prone to observer variability.
- Artificial Intelligence (AI) offers potential to expedite diagnosis and improve accuracy in cervical cancer screening.
Purpose of the Study:
- To develop a novel deep learning-based decision support system for identifying cervical cancer in LBC smear images.
- To enhance diagnostic efficiency and accuracy compared to traditional methods.
Main Methods:
- A hybrid feature reduction and optimization module was developed, combining a sparse Autoencoder with Binary Harris Hawk optimization.
- Three pretrained Convolutional Neural Networks (CNNs) extracted features, which were then refined by the optimization module.
- A Bayesian-optimized K Nearest Neighbors (KNN) classifier was used for cervical cancer classification.
Main Results:
- The system achieved a high classification accuracy of 99.9%.
- It demonstrated excellent performance in detecting cervical cancer stages with 99.8% sensitivity.
- The system identified the absence of cervical cancer stages with a specificity of 99.9%.
Conclusions:
- The proposed AI system significantly outperforms existing deep learning methods for cervical cancer identification.
- This advanced system promises to reduce diagnostic delays and improve patient outcomes in cervical cancer screening.
