Advanced Feature Extraction for Cervical Cancer Image Classification: Integrating Neural Feature Extraction and
Muhammad Amjad Raza1,2, Hafeez Ur Rehman Siddiqui1, Adil Ali Saleem1
1Institute of Computer Science, Khwaja Fareed University of Engineering and Information Technology, Abu Dhabi Road, Rahim Yar Khan 64200, Pakistan.
Sensors (Basel, Switzerland)
|May 14, 2025
Summary
This study introduces a deep learning framework for cervical cancer diagnosis, achieving 99.96% accuracy with KNN. This AI approach enhances early detection, particularly in resource-limited settings.
Area of Science:
- Oncology
- Medical Imaging
- Artificial Intelligence
Background:
- Cervical cancer is a major global health issue, causing high mortality, especially in underserved regions.
- Accurate and early diagnosis is critical for effective treatment and improved patient outcomes.
- Existing diagnostic methods may face limitations in resource-constrained environments.
Purpose of the Study:
- To develop and evaluate an advanced deep learning framework for cervical cancer diagnosis.
- To investigate the efficacy of a novel classification approach integrating feature extraction and interaction learning.
- To assess the performance of various machine learning classifiers for cervical cancer image classification.
Main Methods:
- Utilized a publicly available cervical cancer image dataset.
- Developed a novel classification framework employing a Neural Feature Extractor (NFE) with VGG16 and an AutoInt model.
- Applied machine learning classifiers including KNN, LGBM, and Extra Trees for classification.
- Evaluated computational complexity and prediction times of different models.
Main Results:
- The proposed deep learning framework achieved high diagnostic accuracy.
- K-Nearest Neighbors (KNN) classifier yielded the highest accuracy at 99.96%, followed by LGBM at 99.92%.
- Simpler models like LDA demonstrated faster prediction times, while KNN and LGBM offered superior accuracy.
Conclusions:
- Deep learning frameworks show significant potential for improving cervical cancer classification accuracy.
- The developed methodology offers a promising tool for early cervical cancer detection.
- This approach could be particularly impactful in resource-limited settings for enhancing diagnostic capabilities.
