Artificial Intelligence-based Quantification of Pleural Plaque Volume and Association With Lung Function in Asbestos-exposed Patients

Our plain-English summary of Artificial Intelligence-based Quantification of Pleural Plaque Volume and Association With Lung Function in Asbestos-exposed Patients, by K.B.W. Groot Lipman, T.N. Boellaard, C.J. de Gooijer, et al (Journal of Thoracic Imaging, 2024).

The study

• Authors: K.B.W. Groot Lipman, T.N. Boellaard, C.J. de Gooijer, N. Bogveradze, E.K. Hong, F. Landolfi, F. Castagnoli, N. Vakhidova, I. Smesseim, F. van der Heijden, R.G.H. Beets-Tan, R. Wittenberg, Z. Bodalal, J.A. Burgers, S. Trebeschi
• Journal: Journal of Thoracic Imaging
• Published: 2024;39(3):165-172
• Cited by: 7 articles, as counted by NIH iCite
• Read it: PubMed · Publisher · Free full text

Abstract

Purpose: Pleural plaques (PPs) are morphologic manifestations of long-term asbestos exposure. The relationship between PP and lung function is not well understood, whereas the time-consuming nature of PP delineation to obtain volume impedes research. To automate the laborious task of delineation, we aimed to develop automatic artificial intelligence (AI)-driven segmentation of PP. Moreover, we aimed to explore the relationship between pleural plaque volume (PPV) and pulmonary function tests.

Materials And Methods: Radiologists manually delineated PPs retrospectively in computed tomography (CT) images of patients with occupational exposure to asbestos (May 2014 to November 2019). We trained an AI model with a no-new-UNet architecture. The Dice Similarity Coefficient quantified the overlap between AI and radiologists. The Spearman correlation coefficient ( r ) was used for the correlation between PPV and pulmonary function test metrics. When recorded, these were vital capacity (VC), forced vital capacity (FVC), and diffusing capacity for carbon monoxide (DLCO).

Results: We trained the AI system on 422 CT scans in 5 folds, each time with a different fold (n = 84 to 85) as a test set. On these independent test sets combined, the correlation between the predicted volumes and the ground truth was r = 0.90, and the median overlap was 0.71 Dice Similarity Coefficient. We found weak to moderate correlations with PPV for VC (n = 80, r = -0.40) and FVC (n = 82, r = -0.38), but no correlation for DLCO (n = 84, r = -0.09). When the cohort was split on the median PPV, we observed statistically significantly lower VC ( P = 0.001) and FVC ( P = 0.04) values for the higher PPV patients, but not for DLCO ( P = 0.19).

Conclusion: We successfully developed an AI algorithm to automatically segment PP in CT images to enable fast volume extraction. Moreover, we have observed that PPV is associated with loss in VC and FVC.

Copyright © 2023 The Author(s). Published by Wolters Kluwer Health, Inc.

In plain English

This retrospective study used 523 applications for asbestosis government support (May 2014 to November 2019) from people with occupational asbestos exposure. Applicants were almost all men, median age 75. Excluding 74 without a CT scan and 27 without pleural plaques left 422 CT scans. Radiologists outlined the plaques by hand to train an artificial intelligence (AI) program. Across test sets, the AI's outlines overlapped the radiologists' with a median Dice score of 0.71 (CI 0.70 to 0.73), and the volumes correlated at r = 0.90 (CI 0.88 to 0.92). Among patients without diffuse fibrosis tested within a year of the scan (80 to 84 per test), larger plaque volume was moderately correlated with lower vital capacity (VC; r = −0.40, CI −0.54 to −0.22) and forced vital capacity (FVC; r = −0.38, CI −0.52 to −0.21), but not with diffusing capacity (DLCO; r = −0.09, CI −0.25 to 0.08). The authors concluded that higher plaque volumes were associated with decreased FVC and VC but not DLCO. Asbestos exposure, body mass index and smoking were unknown and could not be adjusted for.

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: Retrospective analysis: development and testing of an AI segmentation model, plus a correlation analysis of plaque volume against lung function tests
• Who was studied: 523 applications for asbestosis government support, May 2014 to November 2019. Median age 75 years; almost exclusively male (2 females). After excluding applications with no CT scan (74) or no pleural plaques (27), 422 CT scans remained (303 with contrast).
• Exposure: Occupational asbestos exposure. The article gives no fiber types, jobs or doses, and states that asbestos exposure information was unknown.
• Main results:
  • On all test sets combined, the median overlap (Dice Similarity Coefficient) between AI and radiologist outlines was 0.71 (0.70 to 0.73)
  • AI-predicted plaque volume correlated strongly with radiologist-segmented volume: Spearman r = 0.90 (CI 0.88 to 0.92), P less than 0.001
  • The AI's overlap with the radiologists was 0.92 for calcified plaques and 0.62 for the noncalcified part of plaques
  • Of patients without diffuse fibrosis, 106 had a lung function test within a year of the CT scan. Vital capacity was available for 80, forced vital capacity for 82 and DLCO for 84.
  • Radiologist-segmented plaque volume was moderately negatively correlated with vital capacity (r = −0.40, CI −0.54 to −0.22, P = 0.0003)
  • Plaque volume was moderately negatively correlated with forced vital capacity (r = −0.38, CI −0.52 to −0.21, P = 0.0005)
  • Plaque volume was not correlated with diffusing capacity for carbon monoxide (r = −0.09, CI −0.25 to 0.08, P = 0.39)
  • Plaque volume divided by lung volume was moderately negatively correlated with VC (r = −0.45) and FVC (r = −0.42)
  • Split at the median plaque volume, mean VC was 80.3 in the higher-volume group (n = 40) versus 94.6 in the lower-volume group (n = 40), P = 0.001
  • Split at the median, mean FVC was 81.0 in the higher-volume group (n = 41) versus 89.6 in the lower-volume group (n = 41), P = 0.037. This is above the corrected significance level of 0.017 that the authors stated for Table 3.
  • Split at the 75th percentile, mean FVC was 69.0 in the higher-volume group (n = 21) versus 90.6 in the lower-volume group (n = 61), P less than 0.001
  • DLCO showed no statistically significant difference between higher and lower plaque volume groups
• The authors' conclusion: The authors concluded that the AI segmentations were adequate and correlated highly with expert readers, and that higher plaque volumes were significantly associated with decreased FVC and VC but not with DLCO.
• Limitations the authors state:
  • A substantial portion of the radiologists' segmentations had to be revised, which the authors say indicates high interobserver variability
  • Confounders such as asbestos exposure, body mass index and smoking could not be corrected for because that information was unknown
  • CT scans taken at both mid-respiratory and inspiratory volume were included, which may bias lung volume and the plaque-to-lung-volume ratio
  • Lung function was in percent-predicted values, which made comparison with an absolute measure such as plaque volume complex

Topics

The National Library of Medicine indexes this study under subjects our site covers: Pleural plaques and thickening.

Medical terms in this study, explained in our Lexicon: Pleural Plaques, Asbestos Exposure, Pulmonary Function Test (PFT), Occupational Exposure, Diffusing Capacity (DLCO), CT Scan.

Related studies

• Artificial intelligence-based diagnosis of asbestosis: analysis of a database with applicants for asbestosis state aid — K.B.W. Groot Lipman, C.J. de Gooijer, et al. (2023)
• 3-Dimensional Quantification of Composite Pleural Plaque Volume in Patients Exposed to Asbestos Using High-resolution Computed Tomography: A Validation Study — G. Dournes, A. Dubois, et al. (2019)
• Development of rapid and highly accurate method to measure concentration of fibers in atmosphere using artificial intelligence and scanning electron microscopy — Y. Iida, K. Watanabe, et al. (2021)
• Rapid fiber-detection technique by artificial intelligence in phase-contrast microscope images of simulated atmospheric samples — T. Yamamoto, K. Iwasaki, et al. (2024)
• The natural course of lung function decline in asbestos exposed subjects with pleural plaques and asbestosis — M. Barnikel, P.M. Million, et al. (2019)
• A systematic review of the association between pleural plaques and changes in lung function — L. Kopylev, K.Y. Christensen, et al. (2015)

Browse more research in Journal of Thoracic Imaging, or return to medical research.