Artificial Intelligence in Pleural Diseases: Current Applications and Next Steps

Our plain-English summary of Artificial Intelligence in Pleural Diseases: Current Applications and Next Steps, by F. Karataş, Ö. Dikensoy (Thoracic Research and Practice, 2026).

The study

• Authors: F. Karataş, Ö. Dikensoy
• Journal: Thoracic Research and Practice
• Published: 2026;27(1):57-67
• Cited by: 3 articles, as counted by NIH iCite
• Read it: PubMed · Publisher · Free full text

Abstract

Pleural diseases pose a significant burden on healthcare systems due to diagnostic challenges and high costs. Artificial intelligence (AI) has the potential to provide faster, more accurate, and more reliable results in the diagnosis of these diseases. This review evaluates the current status of AI technologies in the diagnosis of pleural effusion (PE), malignant PE, tuberculosis pleurisy (TP), pneumothorax, and malignant pleural mesothelioma (MPM). Deep learning algorithms developed for radiological diagnosis provide high sensitivity and specificity in determining the presence and severity of PE. AI models that integrate clinical parameters such as chest computed tomography (CT), positron emission tomography (PET)-CT, and tumour markers in distinguishing between benign and malignant effusions have significantly improved diagnostic accuracy (area under the curve: >0.90). In cytological diagnosis, computer-assisted systems such as Aitrox have demonstrated performance comparable to that of expert cytopathologists in diagnosing malignant effusions. In the diagnosis of TP, AI models outperform conventional diagnostic methods, particularly when combined with laboratory parameters such as adenosine deaminase. Food and Drug Administration-approved AI models are effectively used for the rapid diagnosis of pneumothorax and for emergency interventions. In MPM diagnosis, AI models using PET-CT images and three-dimensional segmentation offer significant advantages in prognostic evaluation and treatment response monitoring. However, large-scale, multi-centre studies are needed to standardise and generalise AI models. In light of these developments, AI may fundamentally change the diagnostic management of pleural diseases.

Copyright© 2026 The Author(s). Published by Galenos Publishing House on behalf of Turkish Thoracic Society.

In plain English

This article was a review, not a new study. The authors searched PubMed, without date limits, for English-language human studies on artificial intelligence (AI) systems in pleural diseases. These covered pleural effusion, malignant pleural effusion, tuberculous pleurisy, pneumothorax and malignant pleural mesothelioma. They did not say how many studies they included, and all results described came from other researchers' work. The authors noted that mesothelioma is associated with asbestos exposure in 80% of cases. In one cited mesothelioma study, AI reading PET-CT images alone did worse than expert interpretation. AI that combined PET-CT with SUVmax, gender and age did better than both. Another cited study, of 188 patients with undiagnosed pleural effusion, reported a model using biomarkers with an area under the curve of 0.97 and 91% accuracy for mesothelioma. The authors said most studies were small, single-centre and retrospective. They called for large multicentre studies and concluded that AI may redefine diagnostic approaches in pleural disease.

We wrote this plain-English summary from the full article with the help of AI, and a second AI model checked each sentence against the article. It is not the authors' wording; the abstract and the article are their own account.

What the study did and found

• Design: Narrative review of published studies, based on a PubMed search with no date restriction
• Who was studied: Published English-language human studies of AI systems in pleural diseases (pleural effusion, malignant pleural effusion, tuberculous pleurisy, pneumothorax, malignant pleural mesothelioma). The number of included studies is not stated.
• Exposure: The review did not study asbestos exposure. It reviewed AI tools applied to chest X-ray, CT, PET-CT, lung ultrasound, cytology and laboratory data. Asbestos appears only as background: the authors state that malignant pleural mesothelioma is associated with asbestos exposure in 80% of cases.
• Main results:
  • In a cited study (Kitajima et al.) using a 3D deep convolutional neural network to tell mesothelioma from benign pleural disease, four approaches had AUC values of 82.5%, 85.4%, 88.1% and 89.6. The approaches were PET-CT with AI, radiologist interpretation, the SUVmax method, and AI combining PET-CT, SUVmax, gender and age.
  • In that cited study, AI using PET-CT alone performed worse than expert interpretation and SUVmax-based methods. AI combining PET-CT, SUVmax, gender and age performed better than both.
  • In a cited study of 188 patients with undiagnosed pleural effusion, six models were tested using the biomarkers SMRP, CEA and CYFRA21-1. The logistic regression model reached an AUC of 0.97 and an accuracy of 91% for mesothelioma.
  • In a cited study (Er and Tanrikulu) involving 324 mesothelioma cases, a newly developed AI system had an accuracy rate of 97.7%.
  • In a cited study (Karapınar Şentürk and Çekiç) of five AI models for diagnosing mesothelioma, accuracies ranged from 80% to 100% and sensitivities from 50% to 100%. The review cautions that most samples in that study came from non-diseased groups, so real-world rates may differ.
  • In a cited study comparing four AI models for the differential diagnosis of mesothelioma from cytology and tumor markers, the logic learning machine had the highest accuracy, 77.5%.
  • For pneumothorax, the review reports that FDA-approved AI tools had sensitivity from 84.3% to 94.6% and specificity from 87.9% to 95.1%.
  • In a cited cytology study, the Aitrox AI model, used to classify benign and malignant lung cancer cells in pleural effusion cytology, had 91.67% accuracy, 87.5% sensitivity, 94.4% specificity and an AUC of 0.9526.
• The authors' conclusion: The authors concluded that AI tools offer non-invasive, rapid and objective clinical support and may redefine diagnosis of pleural disease. They added that large-scale, multicentre studies with standardized protocols are needed before routine use.
• Limitations the authors state:
  • Most AI studies in pleural diseases used relatively small, single-centre cohorts, which limits their external validity and generalizability.
  • Most existing models were validated only retrospectively.
  • Data imbalance and selection bias may lead to overfitting.
  • The 'black-box' nature of deep learning limits clinicians' ability to understand how outputs are generated.

Topics

Medical terms in this study, explained in our Lexicon: Pleural Effusion, Pleurisy, Pneumothorax, Pleural Mesothelioma, PET-CT.

Related studies

• Predicting Mesothelioma Using Artificial Intelligence: A Scoping Review of Common Models and Applications — M. Ram, M.R. Afrash, et al. (2025)
• Application of artificial intelligence in the analysis of asbestos fibers — R. Lee, D. Van Orden, et al. (2025)
• Personalized Oncolytic Therapy: The Next Step Toward the Successful Clinical Application of Vaccine-Strain Measles Viruses for Cancer Therapy? — T. Peikert (2020)
• Artificial Intelligence Techniques to Predict the Airway Disorders Illness: A Systematic Review — A. Koul, R.K. Bawa, et al. (2023)
• Gender Differences in the Use of ChatGPT as Generative Artificial Intelligence for Clinical Research and Decision-Making in Occupational Medicine — P. Mashburn, F.A. Weuthen, et al. (2025)
• Artificial Intelligence-based Quantification of Pleural Plaque Volume and Association With Lung Function in Asbestos-exposed Patients — K.B.W. Groot Lipman, T.N. Boellaard, et al. (2024)

Browse return to medical research.