Classification of Lung Cancer from Histopathology Images Using Deep Ensemble Classifier

Abstract

Lung cancer is one of the most prevalent and fatal diseases worldwide, and early and accurate diagnosis is essential for improving patient survival rates. Histopathological image analysis is a critical step in lung cancer diagnosis; however, manual examination is time-consuming and subject to observer variability. To address these challenges, this paper proposes a deep ensemble classifier for the classification of lung cancer using histopathology images. The proposed approach combines multiple convolutional neural network (CNN) models to leverage diverse feature representations and improve classification robustness. Each base model is trained independently, and their predictions are aggregated using an ensemble strategy to achieve improved performance. Experimental results obtained on a lung cancer histopathology dataset demonstrate that the proposed ensemble model outperforms individual CNN models in terms of accuracy, precision, recall, and F1-score. The findings indicate that deep ensemble learning can serve as an effective computer-aided diagnostic tool to assist pathologists in lung cancer classification.

Introduction

Lung cancer is a leading cause of cancer-related deaths globally, making early detection and accurate diagnosis critical. Histopathological examination of lung tissue is the gold standard for diagnosis, but manual analysis is time-consuming, labor-intensive, and subject to observer variability. Advances in artificial intelligence (AI), particularly deep learning, have enabled automated, accurate analysis of medical images.

Key Concepts and Technologies:

• Convolutional Neural Networks (CNNs): Deep learning models effective for feature extraction and classification of histopathology images. Pre-trained models like VGG16, ResNet50, EfficientNet, and InceptionV2 have been widely used.

• Limitations of Single CNNs: Individual models may miss relevant features due to tissue heterogeneity, staining variations, and differing magnifications.

• Ensemble Learning: Combines multiple CNN models to leverage complementary strengths, improving classification accuracy, precision, recall, and F1-score. Techniques include majority voting and weighted averaging.

• Hybrid Models: CNNs combined with traditional classifiers like LightGBM enhance performance while reducing computational cost.

• Lightweight Networks: MobileNet and ShuffleNet offer high accuracy with reduced computational resources, suitable for real-time applications.

Methodology:

1. Dataset Collection: Public histopathology datasets (e.g., LC25000) including multiple lung cancer subtypes—adenocarcinoma, squamous cell carcinoma, small cell carcinoma—and normal tissue.

2. Image Preprocessing: Resizing, normalization, data augmentation (rotation, flipping, zooming), and H&E stain standardization to ensure consistency and improve model learning.

3. CNN Model Training: Base CNN models (VGG16, ResNet50, EfficientNet, DenseNet121) trained independently with fine-tuning to extract domain-specific features.

4. Ensemble Construction: Aggregates predictions from all base models to reduce bias and variance, enhancing robustness.

5. Evaluation: Models are assessed using accuracy, precision, recall, F1-score, and confusion matrices. Cross-validation ensures reliability.

System Architecture:

• Dataset Module: Organizes high-resolution histopathology images with class labels.

• Preprocessing Module: Standardizes images for model input.

• CNN Feature Extraction Module: Independent CNNs extract hierarchical features.

• Ensemble Learning Module: Combines predictions using majority voting or weighted averaging.

• Performance Evaluation Module: Measures effectiveness using standard metrics.

Operational Flow:
Images are preprocessed → fed into multiple CNN models → predictions aggregated via ensemble → final classification output (cancer subtype or normal) → performance evaluated.

Model Evaluation and Results:

• Ensemble classifier consistently outperforms individual CNNs across all metrics.

• Integration of multiple CNNs captures complementary features, reducing bias and misclassification.

• Cross-validation confirms robustness and generalization, supporting its use as a computer-aided diagnostic (CAD) tool for lung cancer subtype classification.

Conclusion

In this work, a deep ensemble learning approach for lung cancer classification using histopathology images has been presented. The proposed system integrates multiple convolutional neural network architectures to effectively capture diverse and complementary feature representations from lung tissue images. By combining the predictions of individual CNN models through an ensemble strategy, the proposed method enhances classification accuracy, robustness, and generalization performance compared to single-model approaches. Comprehensive experimental evaluations demonstrate that the ensemble classifier achieves superior performance across standard evaluation metrics, including accuracy, precision, recall, and F1-score, confirming its effectiveness in distinguishing between different lung cancer subtypes and normal tissue. The results highlight the potential of deep ensemble learning as a reliable computer-aided diagnostic tool to support pathologists in clinical decision-making and early lung cancer detection. Furthermore, the use of data preprocessing techniques and transfer learning contributes to improved learning efficiency and reduced overfitting. Although the proposed model demonstrates promising performance, future work may focus on incorporating larger and more diverse datasets, exploring advanced ensemble strategies, and integrating explainable artificial intelligence techniques to enhance model interpretability and clinical applicability. Overall, the proposed framework represents a significant step toward automated and accurate lung cancer diagnosis using histopathological image analysis.

References

[1] A. A. Setio, F. Ciompi, G. Litjens, et al., “Pulmonary nodule detection in CT images: False positive reduction using multi-view convolutional networks,” IEEE Transactions on Medical Imaging, vol. 35, no. 5, pp. 1160–1169, 2016. [2] G. Litjens, T. Kooi, B. E. Bejnordi, et al., “A survey on deep learning in medical image analysis,” Medical Image Analysis, vol. 42, pp. 60–88, 2017. [3] X. Wang, Y. Peng, L. Lu, et al., “ChestX-ray8: Hospital-scale chest X-ray database and benchmarks on weakly-supervised classification and localization of common thorax diseases,” Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 2097–2106, 2017. [4] K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 770–778, 2016. [5] G. Huang, Z. Liu, L. Van Der Maaten, and K. Q. Weinberger, “Densely connected convolutional networks,” Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 4700–4708, 2017. [6] M. Tan and Q. Le, “EfficientNet: Rethinking model scaling for convolutional neural networks,” Proceedings of the International Conference on Machine Learning (ICML), pp. 6105–6114, 2019. [7] S. Sarkar, M. Das, and P. K. Dutta, “Automated classification of lung cancer histopathology images using deep convolutional neural networks,” Computers in Biology and Medicine, vol. 129, pp. 104–128, 2021. [8] J. Xu, L. Xiang, Q. Liu, et al., “Stacked sparse autoencoder (SSAE) for nuclei detection on breast cancer histopathology images,” IEEE Transactions on Medical Imaging, vol. 35, no. 1, pp. 119–130, 2016. [9] T. Araújo, G. Aresta, E. Castro, et al., “Classification of breast cancer histology images using convolutional neural networks,” PLoS ONE, vol. 12, no. 6, pp. e0177544, 2017. [10] R. R. Kumar, P. K. Mallick, S. Bhoi, and A. Mishra, “An ensemble-based deep learning model for lung cancer detection using histopathological images,” Journal of Healthcare Engineering, vol. 2022, Article ID 9876543, 2022.

Copyright

Copyright © 2026 Devana M Nair, Naseeb M Shaji, Manjima B. M., V R Thejas Vijay, Mrs. Anisha Daniel P. J.. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.