CNN-RNN framework for lung cancer classification using CT imaging and GAN-based augmentation

Abstract

Lung cancer remains one of the leading causes of cancer-related deaths worldwide, and early identification of malignant abnormalities plays an important role in improving patient survival rates. However, accurate lung cancer classification using CT imaging remains challenging because of limited dataset availability, class imbalance, overlapping lesion characteristics, and lack of interpretability in existing deep learning systems. This study presents a GenAI-driven CNN–RNN framework for explainable lung cancer classification using CT imaging and GAN-based augmentation. The proposed framework integrates convolutional neural networks for spatial feature extraction, LSTM-based recurrent learning for sequential dependency analysis, GAN-assisted augmentation for improving minority class representation, and attention-guided feature fusion for enhanced classification performance. The experimental evaluation was conducted using the publicly available IQ-OTH/NCCD lung cancer CT imaging dataset containing Normal, Benign, and Malignant categories. During preprocessing, normalization, resizing, and image enhancement operations were applied to improve image consistency before training. The framework was trained using the Adam optimizer with 50 epochs and evaluated using 5-fold cross-validation. Experimental results demonstrated that the proposed framework achieved an accuracy of 92.84%, precision of 91.76%, recall of 90.42%, F1-score of 91.08%, and ROC-AUC value of 0.94. Grad-CAM and SHAP visualization methods further improved interpretability by highlighting important lesion regions influencing prediction outcomes. The obtained findings suggest that the proposed CNN–RNN framework can support CT image-based lung cancer classification under limited dataset conditions for CT image-based lung cancer classification under limited medical imaging conditions.