Deep Learning Model for Pneumothorax Detection in Chest Radiographs: A Multicenter Retrospective Cross-Sectional Study

Document Type : Original Article

Authors

1 Department of Emergency Medicine, School of Medicine, Mashhad University of Medical Sciences, Mashhad, Iran

2 Radiation Sciences Research Center, AJA University of Medical Sciences, Tehran, Iran

3 Department of Radiology, Faculty of Medicine, AJA University of Medical Sciences, Tehran, Iran

4 Department of Emergency Medicine, Be-sat Hospital, AJA University of Medical Sciences, Tehran, Iran

5 Department of Interventional Radiology and Radiation Sciences Research Center, Aja University of Medical Sciences, Tehran, Iran

10.30476/ijvlms.2025.106465.1333

Abstract

Background: Pneumothorax is a common clinical condition characterized by the presence of air within the pleural space, occurring in about half of chest trauma cases. Its clinical presentation ranges from asymptomatic cases to severe conditions causing hemodynamic instability or death. Deep learning models offer transformative potential for both clinical diagnosis and medical education through automated detection and interactive training tools. This study sought to evaluate deep learning models for detecting pneumothorax in Chest Radiographs (CXRs), assessing their diagnostic accuracy and potential to enhance medical education.
Methods: This retrospective cross-sectional study was conducted between February 2022 and September 2023 to assess the performance of four deep learning models for pneumothorax detection: Mask Region-based Convolutional Neural Network (Mask R-CNN), Deep Labelling version 3 (DeepLabv3), You Only Look Once version 8 (YOLOv8), and the U-shaped CNN model (U-Net). The evaluation was conducted using 20,000 chest X-ray images sourced from three hospitals in Iran, along with three open source datasets, including PTX-498, PTX-227, and SIIM-ACR-Pneumothorax. Images were labeled by consensus from two radiologists and two traumatologists. Rather than applying a conventional percentage-based split, a tiered data strategy was applied: internal datasets for training and validation, and external datasets (CheXpert and NIH) for independent testing to verify generalizability. Each model was trained to detect pneumothorax by extracting features and performing segmentation. Performance was evaluated using sensitivity, specificity, precision, recall, and F1-score. The outputs were analyzed for integration into virtual learning platforms to train medical students in diagnosing pneumothorax.
Results: The YOLOv8 algorithm showed the best performance for detecting and localizing pneumothorax, achieving an F1 score of 0.68. The final model’s precision was 0.79, and a recall of 0.60, and it worked best on chest X-ray images with 1024x1024 resolution, particularly showing greater accuracy in identifying larger pneumothoraces.
Conclusion: Integration of YOLOv8 into medical education has the potential to improve diagnostic training via interactive AI-based simulations. However, challenges remain in detecting smaller pneumothoraces, highlighting the need for further optimization.

Highlights

Reza Ibrahimi (Google Scholar)

Mohammad Reza Azimi-Aval (Google Scholar)

Keywords

References

  1. Swierzy M, Helmig M, Ismail M, Rückert J, Walles T, Neudecker J. [Pneumothorax]. Zentralbl Chir. 2014;139 Suppl 1:S69-86; quiz S7. German. doi: 10.1055/s-0034-1383029. PubMed PMID: 25264729.
  2. DeMaio A, Semaan R. Management of Pneumothorax. Clin Chest Med. 2021;42(4):729-38.
  3. Tran J, Haussner W, Shah K. Traumatic Pneumothorax: A Review of Current Diagnostic Practices And Evolving Management. J Emerg Med. 2021;61(5):517-28. PubMed PMID: 34470716.
  4. Ince A, Ozucelik DN, Avci A, Nizam O, Dogan H, Topal MA. Management of pneumothorax in emergency medicine departments: multicenter trial. Iran Red Crescent Med J. 2013;15(12):e11586. PubMed PMID: 24693384; PubMed Central PMCID: PMC3955499.
  5. Alghnam S, Aldahnim MH, Aldebasi MH, Towhari JA, Alghamdi AS, Alharbi AA, Almarhabi YA, Albabtain IT. The incidence and predictors of pneumothorax among trauma patients in Saudi Arabia. Findings from a level-I trauma center. Saudi Med J. 2020;41(3):247-52. doi: 10.15537/smj.2020.3.24917. PubMed PMID: 32114596; PubMed Central PMCID: PMC7841555.
  6. Çallı E, Sogancioglu E, van Ginneken B, van Leeuwen KG, Murphy K. Deep learning for chest X-ray analysis: A survey. Med Image Anal. 2021;72:102125. doi: 10.1016/j.media.2021.102125. PubMed PMID: 34171622.
  7. Thian YL, Ng D, Hallinan J, Jagmohan P, Sia SY, Tan CH, et al. Deep Learning Systems for Pneumothorax Detection on Chest Radiographs: A Multicenter External Validation Study. Radiol Artif Intell. 2021;3(4):e200190. doi: 10.1148/ryai.2021200190. PubMed PMID: 34350409; PubMed Central PMCID: PMC8328109.
  8. Lee KL, Graham CA, Yeung JH, Ahuja AT, Rainer TH. Occult pneumothorax in Chinese patients with significant blunt chest trauma: incidence and management. Injury. 2010;41(5):492-4. doi: 10.1016/j.injury.2009.12.017. PubMed PMID: 20097342.
  9. Tian Y, Wang J, Yang W, Wang J, Qian D. Deep multi-instance transfer learning for pneumothorax classification in chest X-ray images. Med Phys. 2022;49(1):231-43. doi: 10.1002/mp.15328. PubMed PMID: 34802144.
  10. Li X, Thrall JH, Digumarthy SR, Kalra MK, Pandharipande PV, Zhang B, et al. Deep learning-enabled system for rapid pneumothorax screening on chest CT. Eur J Radiol. 2019;120:108692. doi: 10.1016/j.ejrad.2019.108692. PubMed PMID: 31585302.
  11. Wang Q, Liu Q, Luo G, Liu Z, Huang J, Zhou Y, et al. Automated segmentation and diagnosis of pneumothorax on chest X-rays with fully convolutional multi-scale ScSE-DenseNet: a retrospective study. BMC Medical Informatics and Decision Making. 2020;20(14):317. doi: 10.1186/s12911-020-01325-5. PubMed PMID: 33323117; PubMed Central PMCID: PMC7739478.
  12. Jensen L, Meyer C. Reducing errors in portable chest radiography. Applied Radiology. 2015;44(4):7-15.
  13. Salam A, Andono PN, Purwanto, Soeleman MA, Sidiq M, Alzami F, et al. NCT-CXR: Enhancing Pulmonary Abnormality Segmentation on Chest X-Rays Using Improved Coordinate Geometric Transformations. J Imaging. 2025;11(6):186. doi: 10.3390/jimaging11060186. PubMed PMID: 40558785; PubMed Central PMCID: PMC12194474.
  14. Chassagnon G, Vakalopoulou M, Paragios N, Revel MP. Artificial intelligence applications for thoracic imaging. Eur J Radiol. 2020;123:108774. doi: 10.1016/j.ejrad.2019.108774. PubMed PMID: 31841881.
  15. Do S, Song KD, Chung JW. Basics of Deep Learning: A Radiologist's Guide to Understanding Published Radiology Articles on Deep Learning. Korean J Radiol. 2020;21(1):33-41. doi: 10.3348/kjr.2019.0312. PubMed PMID: 31920027; PubMed Central PMCID: PMC6960318.
  16. Xie Y, Zhu B, Jiang Y, Zhao B, Yu H. Diagnosis of pneumonia from chest X-ray images using YOLO deep learning. Front Neurorobot. 2025;19:1576438. doi: 10.3389/fnbot.2025.1576438. PubMed PMID: 40370634; PubMed Central PMCID: PMC12077197.
  17. Hao S, Li X, Peng W, Fan Z, Ji Z, Ganchev I. YOLO-CXR: A novel detection network for locating multiple small lesions in chest X-ray images. IEEE Access. 2024;12:156003-19. doi: 10.1109/ACCESS.2024.3482102.
  18. Kitamura G, Deible C. Retraining an open-source pneumothorax detecting machine learning algorithm for improved performance to medical images. Clin Imaging. 2020;61:15-9. doi: 10.1016/j.clinimag.2020.01.008. PubMed PMID: 31954346; PubMed Central PMCID: PMC7085967.
  19. Hwang EJ, Hong JH, Lee KH, Kim JI, Nam JG, Kim DS, et al. Deep learning algorithm for surveillance of pneumothorax after lung biopsy: a multicenter diagnostic cohort study. Eur Radiol. 2020;30(7):3660-71. doi: 10.1007/s00330-020-06771-3. PubMed PMID: 32162001.
  20. D'Souza R N, Huang PY, Yeh FC. Structural Analysis and Optimization of Convolutional Neural Networks with a Small Sample Size. Sci Rep. 2020;10(1):834. doi: 10.1038/s41598-020-57866-2. PubMed PMID: 31965034; PubMed Central PMCID: PMC6972775.
  21. Sabottke CF, Spieler BM. The Effect of Image Resolution on Deep Learning in Radiography. Radiol Artif Intell. 2020;2(1):e190015. doi: 10.1148/ryai.2019190015. PubMed PMID: 33937810; Pubmed Central PMCID: PMC8017385.
  22. Suhaimi MS, Zainuddin K, Ghazali MD, Marzukhi F, Samad AM, Majid Z, et al., editors. Comparison of One-Stage and Two-Stage Strategies of Machine Learning Model for Rock Art Object Detection. 2023 IEEE 13th International Conference on System Engineering and Technology (ICSET); 2023 Oct 2; Shah Alam, Selangor, Malaysia. USA: IEEE Xplore; 2023. P. 215-20. doi: 10.1109/ICSET59111.2023.10295089.
  23. Hoechter DJ, Speck E, Siegl D, Laven H, Zwissler B, Kammerer T. Tension Pneumothorax During One-Lung Ventilation - An Underestimated Complication? J Cardiothorac Vasc Anesth. 2018;32(3):1398-402. doi: 10.1053/j.jvca.2017.07.022. PubMed PMID: 29361455.
  24. Zarogoulidis P, Kioumis I, Pitsiou G, Porpodis K, Lampaki S, Papaiwannou A, et al. Pneumothorax: from definition to diagnosis and treatment. J Thorac Dis. 2014;6(Suppl 4):S372-6. doi: 10.3978/j.issn.2072-1439.2014.09.24. PubMed PMID: 25337391; PubMed Central PMCID: PMC4203989.
  25. Feng S, Liu Q, Patel A, Bazai SU, Jin CK, Kim JS, et al. Automated pneumothorax triaging in chest X-rays in the New Zealand population using deep-learning algorithms. J Med Imaging Radiat Oncol. 2022;66(8):1035-43. doi: 10.1111/1754-9485.13393. PubMed PMID: 35224858.
  26. Rudie JD, Duda J, Duong MT, Chen P-H, Xie L, Kurtz R, et al. Brain mri deep learning and bayesian inference system augments radiology resident performance. J Digit Imaging. 2021;34(4):1049-58. doi: 10.1007/s10278-021-00470-1. PubMed PMID: 34131794; PubMed Central PMCID: PMC8455800.
  27. Cheng CT, Chen CC, Fu CY, Chaou CH, Wu YT, Hsu CP, Chang CC, Chung IF, Hsieh CH, Hsieh MJ, Liao CH. Artificial intelligence-based education assists medical students' interpretation of hip fracture. Insights Imaging. 2020;11(1):119. doi: 10.1186/s13244-020-00932-0. PubMed PMID: 33226480; PubMed Central PMCID: PMC7683624.
  28. Wang M, Sun Z, Jia M, Wang Y, Wang H, Zhu X, et al. Intelligent virtual case learning system based on real medical records and natural language processing. BMC Med Inform Decis Mak. 2022;22(1):60. doi: 10.1186/s12911-022-01797-7.
  29. Blumstein G, Zukotynski B, Cevallos N, Ishmael C, Zoller S, Burke Z, et al. Randomized trial of a virtual reality tool to teach surgical technique for tibial shaft fracture intramedullary nailing. Journal of surgical education. 2020;77(4):969-77. doi: 10.1016/j.jsurg.2020.01.002. PubMed PMID: 32035854; PubMed Central PMCID: PMC7351249.
  30. Park S, Lee SM, Kim N, Choe J, Cho Y, Do KH, Seo JB. Application of deep learning-based computer-aided detection system: detecting pneumothorax on chest radiograph after biopsy. Eur Radiol. 2019;29(10):5341-8. doi: 10.1007/s00330-019-06130-x. PubMed PMID: 30915557.
  31. Tseng WC, Wang YC, Chen WC, Lin KP. Development of an AI model for pneumothorax imaging: Dataset and model optimization strategies for real-world deployment. Eur J Radiol Open. 2025;14:100664. doi: 10.1016/j.ejro.2025.100664. PubMed PMID: 40547323; PubMed Central PMCID: PMC12179720.
  32. Kim D, Lee JH, Kim SW, Hong JM, Kim SJ, Song M, Choi JM, Lee SY, Yoon H, Yoo JY. Quantitative Measurement of Pneumothorax Using Artificial Intelligence Management Model and Clinical Application. Diagnostics (Basel). 2022;12(8). doi: 10.3390/diagnostics12081823. PubMed PMID: 36010174; PubMed Central PMCID: PMC9406694.
Interdisciplinary Journal of Virtual Learning in Medical Sciences
Volume 16, Issue 3 - Serial Number 62
September 2025
Pages 302-313

Files

History

  • Receive Date: 25 March 2025
  • Revise Date: 29 August 2025
  • Accept Date: 30 August 2025
  • Publish Date: 01 September 2025

Share

How to cite

Statistics