Related Experiment Video
Updated: Jul 10, 2026
Introduction of an Integrated Pathology Image Management, Artificial Intelligence, and Reporting System
Published on: July 11, 2025
An automated cervical pre-cancerous diagnostic system
Nor Ashidi Mat-Isa1, Mohd Yusoff Mashor, Nor Hayati Othman
1Center for Electronic Intelligent System (CELIS), School of Electrical & Electronic Engineering, Universiti Sains Malaysia, Engineering Campus, Nibong Tebal, Penang, Malaysia. ashidi@eng.usm.my
This study developed an automated diagnostic system for cervical pre-cancerous detection using a novel region-growing-based feature extraction (RGBFE) algorithm and a hierarchical hybrid multilayered perceptron (H(2)MLP) neural network. The system achieved high accuracy, demonstrating its potential for improved cervical cancer screening.
Area of Science:
- Medical Imaging
- Biomedical Engineering
- Artificial Intelligence in Healthcare
Background:
- Cervical pre-cancerous conditions require accurate and timely diagnosis for effective treatment.
- Manual screening methods can be subjective and time-consuming.
- Automated systems offer potential for objective and efficient diagnostic processes.
Purpose of the Study:
- To develop an automated diagnostic system for cervical pre-cancerous conditions.
- To propose a new feature extraction algorithm (RGBFE) for cervical cell analysis.
- To introduce a novel artificial neural network (ANN) architecture (H(2)MLP) for improved diagnostic accuracy.
Main Methods:
- Development of a two-part automated diagnostic system: feature extraction and intelligent diagnosis.
- Implementation of the region-growing-based feature extraction (RGBFE) algorithm to extract nucleus and cytoplasm features.
- Design and application of a hierarchical hybrid multilayered perceptron (H(2)MLP) network for classifying cervical pre-cancerous stages (normal, LSIL, HSIL).
- Validation using 550 reported cases.
Main Results:
- The RGBFE algorithm demonstrated strong linear correlation ( > 0.8, approaching 1) with manual feature extraction by cytotechnologists.
- The H(2)MLP network outperformed standard ANNs, achieving 97.50% accuracy, 100% specificity, and 96.67% sensitivity.
- Low false negative (1.33%) and false positive (3.00%) rates were recorded for the H(2)MLP network.
Conclusions:
- An automated diagnostic system for cervical pre-cancerous conditions was successfully developed.
- The RGBFE algorithm is an effective image processing technique for automatic feature extraction.
- The H(2)MLP network architecture provides superior diagnostic performance for cervical pre-cancerous detection.
