Automated Multi-Class Lung Cancer Classification from CT scans Using Deep Learning: A Comprehensive CNN-Based Framework with Advanced Preprocessing and Performance Analysis
DOI:
https://doi.org/10.64882/ijrt.v14.i2.1275Keywords:
Lung cancer detection, Convolutional Neural Network, CT image classification, deep learning, CLAHE, medical image preprocessing, adenocarcinoma, squamous cell carcinoma, large cell carcinoma, transfer learning, automated diagnosisAbstract
Lung cancer is one of the deadliest cancers worldwide, and the chance of a successful outcome is strongly related to the stage at which the disease is diagnosed. Manual radiological assessment of Computed Tomography (CT) images is the current clinical practice, which is limited by inter-observer variability, workload and lack of specialist services in underserved areas. This work aims to provide a complete automated deep learning framework for classifying the pulmonary malignancy into four classes from CT scan data. Methods: This work presents a complete automated framework for 4-class classification of pulmonary malignancy from CT scan data using deep learning. The pipeline incorporates state-of-the-art pre-processing techniques such as Median Filtering, Histogram Equalization, Contrast-Limited Adaptive Histogram Equalization (CLAHE), and morphological image operations to optimize the diagnostic-relevant features before training the model. The Convolutional Neural Network (CNN) architecture is created with the purpose of training and testing on a curated set of 1,000 chest CT images divided into adenocarcinoma, large cell carcinoma, squamous cell carcinoma and normal lung tissue using stacked convolutional blocks, ReLU activations, Max-Pooling, Dropout regularisation (rate = 0.4), and Softmax output layer. An early stopping regime with patience of 20 epochs and the Adam optimiser with a learning rate of 1×10⁻⁴, and categorical cross-entropy loss function are used. Results: Experimental evaluation results are 90.60% testing accuracy and 72.22% validation accuracy, 0.2203 training loss and 0.8599 validation loss. As shown in the proposed model, the performance of the model is competitive with existing transfer learning baselines such as VGG16, ResNet50, InceptionV3 and ConvNeXt, with the model outperforming the others in the squamous cell carcinoma and normal tissue classes. Conclusions: It is proved that custom CNN architecture with strict preprocessing methods can be a clinically viable automated diagnostic support framework. The future directions involve 3D volumetric CNN integration, interpretability with Grad-CAM, Federated multi-institutional training, and prospective clinical validation.
References
Aberle, D. R., et al. (2011). Reduced lung-cancer mortality with low-dose computed tomographic screening. New England Journal of Medicine, 365(5), 395–409.
Alshamrani, K., & Alshamrani, H. A. (2022). Classification of chest CT lung nodules using collaborative deep learning model. Journal of Multidisciplinary Healthcare, 15, 1083–1095.
Anthimopoulos, M., et al. (2016). Lung pattern classification for interstitial lung diseases using a deep convolutional neural network. IEEE Transactions on Medical Imaging, 35(5), 1207–1216.
Ardila, D., et al. (2019). End-to-end lung cancer screening with deep learning on low-dose CT. Nature Medicine, 25(6), 954–961.
Armato, S. G., et al. (2011). The lung image database consortium (LIDC) and image database resource initiative (IDRI). Medical Physics, 38(2), 915–931.
Baduge, S. K., et al. (2022). Artificial intelligence and smart vision for building and construction 4.0. Expert Systems with Applications, 206, 117825.
Campanella, G., et al. (2019). Clinical-grade computational pathology using weakly supervised deep learning. Nature Medicine, 25, 1301–1309.
Cho, J., et al. (2020). Classifying lung cancer subtypes with VGG-16 CNN. Applied Sciences, 10(8), 2934.
de Koning, H. J., et al. (2020). Reduced lung-cancer mortality with volume CT screening in a randomized trial. New England Journal of Medicine, 382(6), 503–513.
Dou, Q., et al. (2017). Automated pulmonary nodule detection via 3D ConvNets with online sample filtering and hybrid-loss residual learning. MICCAI, 2017.
El-Baz, A., et al. (2013). Computer-aided detection/diagnosis of pulmonary nodules in thoracic CT scans. Artificial Intelligence in Medicine, 62(1), 1–22.
Erasmus, J. J., et al. (2003). Interobserver and intraobserver variability in measurement of non-small-cell carcinoma lung lesions. European Journal of Radiology, 45(1), 37–44.
Faizi, M. K., et al. (2025). Deep learning–based lung cancer classification of CT images. BMC Cancer, 25, 1056.
Gu, Y., et al. (2015). Multi-scale lung nodule detection by using texture and shape features. Computerized Medical Imaging and Graphics, 42, 1–10.
Hany, M. (2020). Chest CT-Scan images dataset: CT images with different types of chest cancer. Kaggle Datasets.
He, K., Zhang, X., Ren, S., & Sun, J. (2016). Deep residual learning for image recognition. CVPR, 770–778.
Huang, G., et al. (2017). Densely connected convolutional networks. CVPR, 4700–4708.
Janiesch, C., Zschech, P., & Heinrich, K. (2021). Machine learning and deep learning. Electronic Markets, 31, 685–695.
Kingma, D. P., & Ba, J. (2015). Adam: A method for stochastic optimization. ICLR, arXiv:1412.6980.
Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2012). ImageNet classification with deep convolutional neural networks. Advances in Neural Information Processing Systems, 25.
LeCun, Y., et al. (1989). Backpropagation applied to handwritten zip code recognition. Neural Computation, 1(4), 541–551.
Litjens, G., et al. (2017). A survey on deep learning in medical image analysis. Medical Image Analysis, 42, 60–88.
Liu, Z., et al. (2022). A ConvNet for the 2020s. CVPR, 11976–11986.
Masud, M., & Tamne, M. (2023). Deep learning for lung cancer screening based on CT images. Diagnostics, 13, 2617.
Nair, V., & Hinton, G. E. (2010). Rectified linear units improve restricted Boltzmann machines. ICML, 807–814.
Pizer, S. M., et al. (1987). Adaptive histogram equalization and its variations. Computer Vision, Graphics, and Image Processing, 39(3), 355–368.
Shaji, A. K., & Rapheal, B. (2021). Lung cancer prediction using machine learning techniques. IEEE International Conference on Smart Structures and Systems, 1–6.
Shatnawi, M. Q., & Abuein, Q. (2022). Deep learning approach for lung cancer diagnosis using CT scans. Journal of Healthcare Engineering, 2022, Article ID 1004960.
Shen, D., Wu, G., & Suk, H. I. (2017). Deep learning in medical image analysis. Annual Review of Biomedical Engineering, 19, 221–248.
Shen, W., et al. (2017). Multi-scale convolutional neural networks for lung nodule classification. IEEE Transactions on Medical Imaging, 36(6), 1484–1492.
Simonyan, K., & Zisserman, A. (2015). Very deep convolutional networks for large-scale image recognition. ICLR, arXiv:1409.1556.
Szegedy, C., et al. (2016). Rethinking the inception architecture for computer vision. CVPR, 2818–2826.
Tajbakhsh, N., et al. (2016). Convolutional neural networks for medical image analysis: Fine tuning or full training? IEEE Transactions on Medical Imaging, 35(5), 1299–1312.
World Health Organization. (2024). Global Cancer Statistics. WHO Press, Geneva, Switzerland.
Xie, Y., et al. (2019). Knowledge-based collaborative deep learning for benign-malignant lung nodule classification on chest CT. IEEE Transactions on Medical Imaging, 38(4), 991–1004.
Downloads
How to Cite
Issue
Section
License
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
