Oral cancer represents a significant global health burden, where early and accurate diagnosis is essential to improve survival rates and treatment outcomes. Conventional diagnostic approaches are often manual, time-consuming, and susceptible to variability, which can delay detection and clinical intervention. To overcome these limitations, this study proposes an enhanced deep learning framework for automated oral cancer classification that integrates DenseNet-201 for deep feature extraction and a Multilayer Perceptron (MLP) for final decision-making. To further improve performance, a novel Quantum-Enhanced Chaotic Differential Evolution (QCDE) algorithm is introduced for hyperparameter optimization, where chaotic mapping enhances population diversity and quantum-inspired strategies strengthen the exploration-exploitation balance, thereby accelerating convergence and reducing overfitting. The effectiveness of QCDE is validated using the CEC2022 benchmark test suite to demonstrate its optimization capability and robustness. By combining QCDE-driven optimization, deep feature representation, and advanced image processing techniques, the proposed framework provides a reliable, efficient, and accurate solution for early oral cancer detection. Experimental results demonstrate that QCDE-MLP achieves 96.43 % accuracy, surpassing competing metaheuristic-optimized classifiers.