Gray-Level Co-occurrence Matrix Feature-based Classification of Cervical Cytology Images using Neural Networks

Cervical cancer is the second most prevalent cancer among women worldwide, emphasizing the need for accurate and early detection systems. This study proposes an automated classification method for cervical cytology using texture features extracted from RGB images through the Gray-Level Co-occurrence Matrix (GLCM) and classified with a Multilayer Perceptron (MLP) neural network. The Levenberg-Marquardt training algorithm was employed, with the optimal model achieved using 15 hidden neurons, resulting in 97.7% training, 98.8% validation, and 97.1% testing accuracy. Receiver Operating Characteristic (ROC) analysis further validated the model’s strong discriminative performance across all classes. These results demonstrate the effectiveness of texture-based MLP models for enhancing cervical cytology screening and supporting early diagnosis in clinical practice.