Automated Hybrid Model-Based Lung Cancer Prediction Using Histopathology Images with Deep Learning

Abstract

Lung cancer is a major cause of death around the world. Finding it early can help save lives. Doctors usually look at tissue images under a microscope to find lung cancer, but this takes time and can sometimes lead to mistakes. In this work, we created a computer system that uses deep learning to help detect lung cancer automatically. We use special AI models called GANs to make the images clearer and create more samples for training. Our system looks at three types of lung tissue: normal, adenocarcinoma, and squamous cell carcinoma. To improve results, we combine three deep learning models—EfficientNet, ResNet, and DenseNetthat work together to make better decisions. We trained the system using both real and AI-created images. The results showed higher accuracy and better performance than using just one model. We also added a language model to help doctors with decision-making and used tools to monitor the system in real time. The system is easy to set up and scale using Docker and Kubernetes. Our goal is to help doctors diagnose lung cancer faster and more accurately. In the future, we plan to use more data, improve how the model explains its results, and keep patient data safe using privacy-friendly methods.

Introduction

Objective:
To improve the early and accurate detection of lung cancer using an automated deep learning system that analyzes histopathology images. This system aims to overcome the challenges of manual diagnosis, such as time consumption, inter-observer variability, and limited scalability.

Key Components of the Proposed System

1. Image Augmentation with GANs:

   • Uses Generative Adversarial Networks (GANs) to generate synthetic, high-quality histopathology images.

   • Enhances training data diversity and reduces class imbalance.

2. Three-Class Classification:

   • Focused on three tissue types:

     • Normal lung tissue (lung_n)

     • Lung adenocarcinoma (lung_aca)

     • Lung squamous cell carcinoma (lung_scc)

3. Ensemble Deep Learning Model:

   • Combines predictions from three CNN models:

     • EfficientNet – lightweight, scalable, high accuracy

     • ResNet – uses residual connections to handle deep networks

     • DenseNet – uses dense connections for efficient feature reuse

   • Ensemble uses a voting mechanism for final prediction.

4. LLM Integration for Clinical Support:

   • A Large Language Model (e.g., BioGPT) generates explanations and clinical summaries of results to assist medical professionals.

5. Containerized and Scalable Deployment:

   • Uses Docker and Kubernetes (Minikube) to deploy the system as microservices.

   • Ensures portability, scalability, and easy system maintenance.

6. Real-Time Monitoring:

   • Prometheus and Grafana monitor system performance (e.g., inference time, accuracy, resource usage).

7. User Interface & Interaction:

   • Web interface allows doctors to upload images and view diagnostic outputs with visual explanations and analytics.

8. Privacy & Federated Learning:

   • Plans to integrate federated learning for privacy-preserving training across hospitals without sharing raw data.

Performance Evaluation

• Metrics Used: Accuracy, Precision, Recall, F1-Score

• Results:

• Impact of GAN-Augmented Data:

  • Boosted model accuracy by 3–5% compared to training on real images alone.

Conclusion

In conclusion, by leveraging the individual strengths of EfficientNet, ResNet, and DenseNet, we have built a powerful and accurate lung cancer classification system. EfficientNet handles computational efficiency, ResNet ensures deep feature learning, and DenseNet promotes rich feature reuse. When combined, these models create an ensemble that outperforms individual networks in terms of precision, recall, and overall accuracy.This multi-model approach is especially valuable in critical applications like medical diagnostics, where every prediction matters.

References

[1] Burhanettin Ozdemir, Emrahaslan ,Ishak Pacal, \" Attention Enhanced InceptionNeXt-Based Hybrid Deep Learning Model for Lung Cancer Detection,\" IEEE Access, Vol. 13, pp.10.1109/ACCESS.2025.3539122 (2025). [2] Alaa Wehbe, Silvana Dellepiane , Irene Minetti, \" Enhanced Lung Cancer Detection and TNM Staging Using YOLOv8 and TNMClassifier: An Integrated Deep Learning Approach for CT Imaging\" IEEE Access, Vol.12, pp 10.1109/ACCESS.2024.3462629 (2024). [3] Tehnan I. A. Mohamed, Absalom EL-Shamir Ezugwu, \" Enhancing Lung Cancer Classification and Prediction With Deep Learning and Multi-Omics Data,\" IEEE Access, Vol. 12, pp 10.1109/ACCESS.2024.3394030, (2024). [4] Pushkar Sathe ,Alka Mahajan , Deepak Patkar, Mithusha Verma, \" End-to-End Fully Automated Lung Cancer Screening System,\" IEEE Access, Vol. 12, pp 10.1109/ACCESS.2024.3435774 (2024). [5] Muhammad Imran, Ahmet e. Topco, Bushrahaq, Ersin Elbasi , Wei Shao \"Transformer Based Hierarchical Model for Non-Small Cell Lung Cancer Detection and Classification \" IEEE Access,Vol.12,pp10.1109/ACCESS.2024.3449230 (2024). [6] Divya Khanna, Arun Kumar, A Shahid Ahmad Bhat, \"Volatile Organic Compounds for the Prediction of Lung Cancer by Using Ensembled Machine Learning Model and Feature Selection\" IEEEAccess,Vol.13,pp10.1109/ACCESS.2025.3527027 (2024). [7] Marian Magdy Amin, Ahmed S. Ismail, Masoud E. Shaheen,\" Multimodal Non-Small Cell Lung Cancer Classification Using Convolutional Neural Networks,\" IEEE Access, Vol.12, pp 10.1109/ACCESS.2024.3461878 (2024). [8] Naglaa F. Noaman1, Bassam M. Kanber Licheng Jiao, Ahmadalsmadi, Mutasem K. Alsmadi, \" Advancing Oncology Diagnostics: AI-Enabled Early Detection of Lung Cancer Through Hybrid Histological Image Analysis \" IEEE Access,Vol.12,pp10.1109/ACCESS.2024.3397040 (2024). [9] P. Deepa C.Gunavathi, \" A Novel Aggregated Multiple Imputation Approach for Enhanced Survival Prediction and Classification on Breast Cancer and Lung Cancer Data, \" IEEE Access, Vol. 12, pp 10.1109/ACCESS.2024.3516837 (2024). [10] Débora V. C. Lima, ,Patrick Terrematte ,Beatriz Stransky, Adrião D. D. Neto, \" An Integrated Data Analysis Using Bioinformatics and Random Forest to Predict Prognosis of Patients With Squamous Cell Lung Cancer\" IEEE Access,Vol.12,pp10.1109/ACCESS.2024.3392277 (2024) [11] Mahmoudragab, , Iyad Katib , Sanaa Abdullah Sharaf, Fatmah Yousef Assiri, Diaa Hamed, Abdullah Al-Malaise Al-Ghamdi, \" Self-Upgraded Cat Mouse Optimizer With Machine Learning Driven Lung Cancer Classification on Computed Tomography Imaging\", IEEE Access, Vol. 11, pp 10.1109/ACCESS.2023.3313508, 2023. [12] Rabbia Mahum, Abdulmaliks. Al-Salman,\" Lung-RetinaNet: Lung Cancer Detection Using a RetinaNet With Multi-Scale Feature Fusion and Context Module \", IEEE Access, Vol.11 pp 10.1109/ACCESS.2023.3281259, 2023. [13] Iftikhar Naseer, Sheeraz Akram, Tehreem Masood, Muhammad Rashid, Arfan Jaffar, \" Lung Cancer Classification Using Modified U-Net Based Lobe Segmentation and Nodule Detection \", IEEE Access, Vol. 11, pp 10.1109/ACCESS.2023.3285821, 2023. [14] Hamed Alqahtani, Eatedal Alabdulkreem, Faiz Abdullah Alotaibi, Mrim M. Alnfiai, Chinu Singla , Ahmeds.Salama, \" Improved Water Strider Algorithm With Convolutional Autoencoder for Lung and Colon Cancer Detection on Histopathological Images\", IEEE Access, Vol.12, pp 10.1109/ACCESS.2023.3346894, 2023. [15] Abdul Rehman Chishti, Abdul Aziz, Muhammadaliqureshi, Muhammadnawazabbasi, Durria Abbasi, Sheikh Sharif Iqbal, Azzedine Zerguine, Abdullah M. Algarni, Rifaqat Hussain, \" Advances in Antenna-Based Techniques for Detection and Monitoring of Critical Chronic Diseases A Comprehensive Review\", IEEE Access, Vol. 11, pp 10.1109/ACCESS.2023.3316149, 2023. [16] Matteo Tortora, Ermanno Cordelli, Rosa Sicilia, ,Lorenzo Nibid ,Edyippolito, Giuseppe Perrone, Sara Ramella, Paolo Soda \" RadioPathomics Multimodal Learning in Non-Small Cell Lung Cancer for Adaptive Radiotherapy\", IEEE Access, Vol. 11, pp 10.1109/ACCESS.2023.3275126, 2023. [17] Qian-Lin Lian, Xiang-You Li, Bing Lu, Chen-Wei Zhu, Jiang-Tao Li, Jian-Jun Chen \" Identification of Lung Tumors in Nude Mice Based on the LIBS With Histogram of Orientation Gradients and Support Vector Machine\", IEEE Access, Vol. 11, pp 10.1109/ACCESS.2023.3342105. 2023. [18] Seungwon Oh, Jaena Im, Sae-Ryung Kang, In-Jae Oh, Andmin-Sookim, \" PET-Based Deep-Learning Model for Predicting Prognosis of Patients With Non-Small Cell Lung Cancer\" , IEEE Access, Vol. 9, pp 10.1109/ACCESS.2021.3115486. 2021. [19] Francisco Silva, Tania Pereira, Joana Morgado, Julieta Frade, José Mendes, Cláudia Freitas, Eduardo Negrão, Beatriz Flor De Lima, Miguel Correia Da Silva ,António J. Madureira, Isabel Ramos, Venceslau Hespanhol, José Luís Costa, António Cunha, Hélder P. Oliveira, \" EGFR Assessment in Lung Cancer CT Images: Analysis of Local and Holistic Regions of Interest Using Deep Unsupervised Transfer Learning\", IEEE Access, Vol.9, pp 10.1109/ACCESS.2021.3070701. 2021. [20] Mohammada.Alzubaidi, Mwaffaq Otoom, Hamzajaradat, \" Comprehensive and Comparative Global and Local Feature Extraction Framework for Lung Cancer Detection Using CT Scan Images \", IEEE Access, Vol.9, pp 10.1109/ACCESS.2021.3129597. 2021

Copyright

Copyright © 2025 Geetha T, Riyaz Ahamed , Santhosh R, Yuvaraj K, Murganswami M. 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.