Classification of Cervical Precancerous Cells Using Gabor Filter and Haar Wavelet Features Based on Neural Networks

Cervical cancer remains a major global health concern, particularly in low-resource regions. This study explored the classification of cervical precancerous cell images using texture-based feature extraction and Artificial Neural Networks (ANNs). Gabor Filter and Haar Wavelet Transform were applied...

Full description

Saved in:
Bibliographic Details
Published in:2024 Beyond Technology Summit on Informatics International Conference (BTS-I2C) pp. 316 - 321
Main Authors: Jusman, Yessi, Resky Pahlevi, Nanda, Intan Rahmawati, Maryza, Aila Mat Zin, Anani, Hussain, Faezahtul Arbaeyah, Winiarti, Sri
Format: Conference Proceeding
Language:English
Published: IEEE 19.12.2024
Subjects:
Accuracy
Biological neural networks
Cervical cancer
Classification
Computer architecture
Feature extraction
Gabor Filter
Gabor filters
Haar Wavelet
Microprocessors
Neural Networks
ROC
Testing
Training
Wavelet transforms
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Cervical cancer remains a major global health concern, particularly in low-resource regions. This study explored the classification of cervical precancerous cell images using texture-based feature extraction and Artificial Neural Networks (ANNs). Gabor Filter and Haar Wavelet Transform were applied to extract critical morphological and textural features, which were classified using Multilayer Perceptrons (MLPs) trained with One Step Secant (OSS) and Gradient Descent with Momentum and Adaptive Learning Rate Backpropagation (GDX) algorithms. The best performance was achieved using Haar Wavelet features with OSS-MLP, attaining a training accuracy of 79.3% with 15 hidden neurons. Evaluation through Receiver Operating Characteristic (ROC) curves demonstrated OSS-MLP's capability to leverage texture features for classification. This research emphasizes the potential of integrating advanced feature extraction with optimized neural networks to develop a reliable and automated diagnostic system for cervical cancer, providing a promising solution for efficient and accurate screening.
DOI:10.1109/BTS-I2C63534.2024.10941777