A Hybrid Deep Learning and Nature-Inspired Optimization Framework for Accurate and Interpretable Lung Cancer Detection from CT Scans

Lung cancer remains one of the leading causes of cancer-related mortality worldwide, primarily due to late diagnosis and complex tumor characteristics. Early detection through medical imaging can significantly improve survival rates. Computed Tomography (CT) scans are widely used for identifying pulmonary nodules; however, manual interpretation by radiologists is time-consuming and susceptible to variability. This study proposes a hybrid deep learning and nature-inspired optimization framework for accurate lung cancer detection from CT images. The proposed approach integrates deep learning-based feature extraction with a nature-inspired optimization algorithm for feature selection, improving classification performance while maintaining model interpretability. CT images undergo preprocessing to remove noise and enhance relevant structures. A convolutional neural network (CNN) extracts deep features representing nodular characteristics. Subsequently, a nature-inspired optimization algorithm selects the most informative features to reduce dimensionality and improve classification accuracy. The optimized features are fed into a classification model to distinguish between malignant and benign lung nodules. Experimental evaluation demonstrates that the proposed hybrid model improves detection accuracy, sensitivity, and computational efficiency compared with traditional deep learning models. The framework also incorporates interpretability mechanisms that highlight significant image regions influencing the classification outcome, assisting clinicians in understanding model predictions. The results indicate that combining artificial intelligence techniques with nature-inspired optimization can significantly enhance automated lung cancer diagnosis systems.