Ensemble Deep Learning Framework for Oral Squamous Cell Carcinoma (OSCC) Classification Using Modified VGG and CNN Features with XGBoost

OSCC is among the most widely diagnosed cancers globally, with early detection being vital for enhancing patient outcomes. This study introduces a hybrid deep learning framework for OSCC classification, combining a Modified VGG (Visual Geometry Group) network and Convolutional Neural Networks (CNN) for feature extraction, followed by the application of XGBoost for classification. The Modified VGG network is adapted to enhance feature extraction from oral lesion images, capturing intricate patterns that traditional methods might overlook. The CNN architecture is fine-tuned to complement this by learning deeper hierarchical features. These extracted features are then passed to XGBoost, an ensemble learning method known for its high performance in classification tasks, to generate reliable predictions. The proposed methodology is evaluated on a comprehensive OSCC dataset, showing improved effectiveness in terms of classification accuracy, sensitivity, and specificity compared to conventional methods. The integration of deep learning-based feature extraction and XGBoost's robust classification capabilities offers a promising approach for early OSCC detection, facilitating clinical decision- making and improving patient outcomes.