Deep Learning Based Approach to Diagnose Lung Cancer Using CT Scan Images

This paper presented an IEEE-style research article on a deep learning-based approach to diagnose lung cancer using CT scan images. The proposed Phase II system introduced an optimized CNN architecture to improve lung cancer classification while reducing computational complexity. The system includes CT image acquisition, preprocessing, augmentation, CNN feature extraction, binary classification, result visualization, and interpretability support using Grad-CAM. The experimental results show that the proposed optimized CNN achieved 97.6% accuracy, 96.8% precision, 97.2% recall, 98.1% specificity, and 97.0% F1-score. These results indicate that the model can effectively classify CT images into cancerous and non-cancerous categories. Compared with Phase I models such as 2D CNN, 3D CNN, and ResNet50, the optimized CNN achieved better accuracy while maintaining efficiency and real-time potential. The study demonstrates that a carefully designed CNN can provide a strong balance between accuracy, interpretability, and deployment feasibility. The inclusion of visualization and confidence output improves user trust and supports clinical review. Overall, the proposed system has strong potential as a computer-aided diagnosis tool for early lung cancer detection.

[1] W. Shen, M. Zhou, F. Yang, C. Yang, and J. Tian, "Multi-scale convolutional neural networks for lung nodule classification," Proc. Int. Conf. Information Processing in Medical Imaging, pp. 403-412, 2016. [2] R. Kumar and S. Gupta, "Lung cancer detection using 2D CNN on CT scan images," International Journal of Computer Applications, vol. 167, no. 9, pp. 14-21, 2017. [3] S. Hussein, R. Gillies, K. Cao, Q. Song, and U. Bagci, "3D convolutional neural networks for lung nodule classification using multi-view representation," IEEE Transactions on Medical Imaging, vol. 37, no. 2, pp. 1030-1040, 2018. [4] A. Paul and S. Bhattacharya, "Hybrid CNN-SVM architecture for improving lung cancer classification accuracy," Procedia Computer Science, vol. 165, pp. 175-183, 2019. [5] D. S. Kermany, K. Zhang, and M. Goldbaum, "Transfer learning using pre-trained CNNs for lung disease classification," Cell, vol. 172, no. 5, pp. 1122-1134, 2019. [6] P. Sathish and R. Nithya, "3D residual CNN model for automatic detection and segmentation of lung nodules," Neural Computing and Applications, vol. 32, no. 15, pp. 11275-11288, 2020. [7] A. Gomez and M. Lopez, "Comparative study of 2D and 3D convolutional neural networks for lung cancer detection from CT scans," Computer Methods and Programs in Biomedicine, vol. 190, pp. 105-118, 2020. [8] X. Li, L. Zhao, and H. Wang, "Attention-based ResNet architecture for lung nodule detection in CT images," Pattern Recognition Letters, vol. 145, pp. 67-75, 2021. [9] T. Zhang and X. Wang, "Attention-guided CNN for enhanced lung nodule detection in CT images," Computerized Medical Imaging and Graphics, vol. 98, p. 102030, 2022. [10] M. Z. Alom, M. Hasan, C. Yakopcic, T. M. Taha, and V. K. Asari, "Recurrent convolutional neural network for lung cancer detection using volumetric CT scans," IEEE Access, vol. 10, pp. 18320-18329, 2022.