APPLICATION OF LENET-5-TYPE NEURAL NETWORK ENSEMBLES FOR DIAGNOSING PNEUMONIA IN LUNG X- RAY IMAGES

D. N. Tumakov1, A. R. Berdigulov1, D. I. Khairullina1, I. I. Bariev2

About the authors

D. N. Tumakov1, A. R. Berdigulov1, D. I. Khairullina1, I. I. Bariev2

Бердигулов А. Р. – магистрант института вычислительной математики и информационных технологий, ФГБОУ ВО К(П)ФУ, кафедра прикладной математики и искусственного интеллекта, 420008, Казань, ул. Кремлевская, 18. Область научных интересов: нейронные сети, машинное зрение, искусственный интеллект в медицине.

Хайруллина Д. И. – магистрант института вычислительной математики и информационных технологий, ФГБОУ ВО К(П)ФУ, кафедра прикладной математики и искусственного интеллекта, 420008, Казань, ул. Кремлевская, 18. Область научных интересов: нейронные сети, машинное зрение, кластеризация.

Бариев И. И. – директор, АНО ВО «Университет Иннополис», 420500, Иннополис, ул. Университетская, 1. Область научных интересов: нейронные сети, искусственный интеллект в медицине.

Download PDF

UDC: 004.032.26, 004.932.2
DOI: 10.24412/2413-7383- 2025-3-38-34-50
Language: Russian

Abstract: Ensembles of shallow neural networks for identifying pneumonia from X-ray images are considered. The ensemble's resulting response is constructed according to the principle: if at least one of its neural networks identifies pneumonia, the ensemble also ret urns the answer "pneumonia." This approach to aggregating the results of individual neural networks maximizes the recall of the "pneumonia" class. To minimize the deterioration in precision that is inevitable with this aggregation approach, the ensemble's neural networks are all of the same type —LeNet-5. Six neural network architectures are proposed, each receiving 256 by 256 pixel images as input and having different convolutional matrix sizes. The architectures of the networks considered also have differe nt depths. On the test set, the recall for the optimal ensembles was 0.991, while the precision varied between 0.934 and 0.936.

Keywords: neural network ensembles, LeNet-5, recall, precision, pneumonia.

References

1. – P. 1589–1601. – DOI: 10.1007/s12559-020-09787-5. References
2. Balasubramaniam S. A modified LeNet CNN for breast cancer diagnosis in ultrasound images / S. Balasubramaniam, Y. Velmurugan, D. Jaganathan, S. Dhanasekaran // Diagnostics. – 2023. – Vol. 13, No. 17. – P. 2746. – DOI: 10.3390/diagnostics13172746.
3. Devesh R. Comparative analysis of machine learning algorithms in pneumonia detection / R. Devesh, V. Harsh, A. Mohd, S. Sonal // Innovative Computing & Communication (ICICC 2024): Proceedings of the International Conference. – 2024. – DOI: 10.2139/ssrn.4776487. 1Т Тумаков Д. Н. Бердигулов А. Р. Хайруллина Д. И. Бариев И. И.
4. Demetgul M. Intelligent monitoring of linear stages with ensembles of improved LeNet DCNN and random forest classifiers / M. Demetgul, Y. Zhao, I.N. Tansel, J. Fleischer // Measurement and Control. – 2025. – [In press]. – DOI: 10.1177/00202940251317395.
5. Hansen L. K. Neural network ensembles / L. K. Hansen, P. Salamon // IEEE Transactions on Pattern Analysis and Machine Intelligence. – 1990. – Vol. 12, No. 10. – P. 993–1001. – DOI: 10.1109/34.58871.
6. Kayumov Z. Convolution neural network learning features for handwritten digit recognition / Z. Kayumov, D. Tumakov // East-West Design and Test Symposium (EWDTS 2020): Proceedings of the 2020 IEEE. – 2020. – P. 330–334. – DOI: 10.1109/EWDTS50664.2020.9224822.
7. Kayumov Z. Hierarchical convolutional neural network for handwritten digits recognition / Z. Kayumov, D. Tumakov, S. Mosin // Procedia Computer Science. – 2020. – Vol. 171. – P. 1927–1934. – DOI: 10.1016/j.procs.2020.04.206.
8. Tiar M. L. Pneumonia Classification using deep learning: a comparative study / M. L. Tiar, N. Terki, J.J. Dominguez-Jimenez // Image and Signal Processing and their Applications (ISPA): Proceedings of the 8th International Conference. – 2024. – P. 1–6. – DOI: 10.1109/ISPA59904.2024.10536711.
9. Tuliabaeva D. On the recognition of weakly blurred, highly contrasting objects by neural networks / D. Tuliabaeva, D. Tumakov, A. Egorchev // AIP Conference Proceedings. – 2024. – Vol. 3243. – Art. No. 020087. – DOI: 10.1063/5.0247350.
10. Verdhan V. Image classification using LeNet / V. Verdhan // Computer Vision Using Deep Learning: Neural Network Architectures with Python and Keras. – Apress, 2021. – 308 p.
11. Yang Y. Pneumonia Recognition by Deep Learning: A Comparative Investigation / Y. Yang, G. Mei // Applied Sciences. – 2022. – Vol. 12, No. 9. – P. 4334. – DOI: 10.3390/app12094334.
12. Schapire R. E. The strength of weak learnability / R.E. Schapire // Machine Learning. – 1990. – Vol. 5, No. 2. – P. 197–227. – DOI: 10.1007/BF00116037.
13. Breiman L. Random forests / L. Breiman // Machine Learning. – 2001. – Vol. 45. – P. 5–32. – DOI: 10.1023/A:1010933404324.
14. Freund Y. A decision-theoretic generalization of on-line learning and an application to boosting / Y. Freund, R.E. Schapire // Journal of Computer and System Sciences. – 1997. – Vol. 55, No. 1. – P. 119–
15. – DOI: 10.1006/jcss.1997.1504.
16. Dehaerne E. Optimizing YOLOv7 for semiconductor defect detection / E. Dehaerne, B. Dey, S. Halder, S. De Gendt // Proceedings of SPIE. – 2023. – Vol. 12496. – P. 635–642. – DOI: 10.48550/arXiv.2302.09565.
17. Chung A. Cloud computed machine learning based real-time litter detection using micro-UAV surveillance / A. Chung, D.Y. Kim, E. Kwok, M. Ryan, E. Tan, R. Gamadia // MIT Undergraduate Research Technology Conference (URTC): Proceedings of the IEEE. – 2018. – P. 1–4. – DOI: 10.1109/URTC45901.2018.9244800.
18. Chernukhin A.V. Postroyeniye modeli prediktivnoy analitiki neispravnostey promyshlennogo oborudovaniya / A.V. Chernukhin. E.A. Bogdanova. T.V. Savitskaya // Programmnyye produkty i sistemy. – 2024. – T. 37. № 2. – S. 254–261. – DOI: 10.15827/0236-235X.142.254-261.
19. Sasaki K. Lightweight and accurate YOLOv7-based ensembles with knowledge distillation for urinary sediment detection / K. Sasaki, H. Nishikawa, I. Taniguchi, T. Onoye // IEEE Access. – 2025. – Vol. 13. – P. 93468–93480. – DOI: 10.1109/ACCESS.2025.3574169.
20. Viola P. Robust real-time face detection / P. Viola, M.J. Jones // International Journal of Computer Vision. – 2004. – Vol. 57, No. 2. – P. 137–154. – DOI: 10.1023/B:VISI.0000013087.49260.fb.
21. Bessarabov N.A. Avtomatizirovannaya razmetka izobrazheniy s bespilotnogo letatelnogo apparata pri pomoshchi predobuchennykh modeley iskusstvennogo intellekta / N.A. Bessarabov. A.A. Sapozhnikov. D.V. Tatarnikov. R.R. Tyugunov. A.M. Tsyganov // Programmnyye produkty i sistemy. – 2025. – T. 38. № 2. – S. 251-260. – DOI: 10.15827/0236-235X.150.251-260.
22. Nanni L. Ensemble of convolutional neural networks for bioimage classification / L. Nanni, S. Ghidoni, S. Brahnam // Applied Computing and Informatics. – 2021. – Vol. 17, No. 1. – P. 19–35. – DOI: 10.1016/j.aci.2018.06.002.
23. Balaji S. A context for effective prediction and classification of diabetic retinopathy disease using deep ensemble AlexNet & LeNet classifier / S. Balaji, B. Karthik // Computing for Sustainable Global Development (INDIACom): Proceedings of the 11th International Conference. – 2024. – P. 414-421. – DOI: 10.23919/INDIACom61295.2024.10498405.
24. Rajeashwari S. Enhancing pneumonia diagnosis with ensemble-modified classifier and transfer learning in deep-CNN based classification of chest radiographs / S. Rajeashwari, K. Arunesh // Biomedical Signal Processing and Control. – 2024. – Vol. 93. – Art. ID 106130. – DOI: 10.1016/j.bspc.2024.106130. Применение ансамблей нейронных сетей типа LeNet-5… 1Т
25. Zhou T. The ensemble deep learning model for novel COVID-19 on CT images / T. Zhou, H. Lu, Z. Yang, S. Qiu, B. Huo, Y. Dong. Applied Soft Computing. 2021. Vol. 98. Art. ID 106885. DOI: 10.1016/j.asoc.2020.106885.
26. Osagie E. Ensemble learning for medical image character recognition based on enhanced LeNet-5 / E. Osagie, W. Ji, N. Helian. Computational Intelligence in Bioinformatics and Computational Biology (CIBCB): Proceedings of the 2023 IEEE Conference. 2023. P. 1–8. DOI: 10.1109/CIBCB56990.2023.10264911.
27. Jaganathan D. Concatenated modified LeNet approach for classifying pneumonia images / D. Jaganathan, S. Balsubramaniam, V. Sureshkumar, S. Dhanasekaran. Journal of Personalized Medicine. 2024. Vol. 14, No. 3. Art. ID 328. DOI: 10.3390/jpm14030328.
28. Meir Y. Efficient shallow learning as an alternative to deep learning / Y. Meir, O. Tevet, Y. Tzach, S. Hodassman, R.D. Gross, I. Kanter. Scientific Reports. 2023. Vol. 13. Art. ID 5423. DOI: 10.1038/s41598- 023-32559-8.
29. Pikalev Ya.S. O neyronnykh arkhitekturakh izvlecheniya priznakov dlya zadachi raspoznavaniya obyektov na ustroystvakh s ogranichennoy vychislitelnoy moshchnostyu / Ya.S. Pikalev. T.V. Ermolenko Problemy iskusstvennogo intellekta. 2023. № 3(30). S. 44-54.
30. Sharma S. A systematic literature review on deep learning approaches for pneumonia detection using chest X-ray images / S. Sharma, K. Guleria. Multimedia Tools and Applications. 2024. Vol. 83. P. 24101– 24151. DOI: 10.1007/s11042-023-16419-1.
31. Singh S. Pneumonia classification using quaternion deep learning / S. Singh, B.K. Tripathi. Multimedia Tools and Applications. 2022. Vol. 81. P. 1743–1764. DOI: 10.1007/s11042-021-11409-7.
32. Hasan M.D.K. Deep learning approaches for detecting pneumonia in COVID-19 patients by analyzing chest X-Ray images / M.D.K. Hasan [и др.] Mathematical Problems in Engineering. 2021. Art. ID 9929274. 8 p. DOI: 10.1155/2021/9929274.
33. Ibrahim A.U. Pneumonia classification using deep learning from chest X-ray images during COVID-19

Issue: 3(38)'2025
Section: ARTIFICIAL INTELLIGENCE AND MACHINE LEARNING
How to cite: D. N. Tumakov1, A. R. Berdigulov1, D. I. Khairullina1, I. I. Bariev2. APPLICATION OF LENET-5-TYPE NEURAL NETWORK ENSEMBLES FOR DIAGNOSING PNEUMONIA IN LUNG X- RAY IMAGES // Problems of Artificial Intelligence. - 2025. - № 3 (38). - P. 34-50. - https://paijournal.guiaidn.ru/en/2025/3(38)-4.html

Back to issue 3(38)'2025