A Deep Learning-Based Radiomic Classifier for Usual Interstitial Pneumonia.
deep learning
idiopathic pulmonary fibrosis
interstitial lung disease
progressive pulmonary fibrosis
radiomic
usual interstitial pneumonia
Journal
Chest
ISSN: 1931-3543
Titre abrégé: Chest
Pays: United States
ID NLM: 0231335
Informations de publication
Date de publication:
Feb 2024
Feb 2024
Historique:
received:
18
07
2023
revised:
09
09
2023
accepted:
05
10
2023
medline:
10
2
2024
pubmed:
17
10
2023
entrez:
16
10
2023
Statut:
ppublish
Résumé
Because chest CT scan has largely supplanted surgical lung biopsy for diagnosing most cases of interstitial lung disease (ILD), tools to standardize CT scan interpretation are urgently needed. Does a deep learning (DL)-based classifier for usual interstitial pneumonia (UIP) derived using CT scan features accurately discriminate radiologist-determined visual UIP? A retrospective cohort study was performed. Chest CT scans acquired in individuals with and without ILD were drawn from a variety of public and private data sources. Using radiologist-determined visual UIP as ground truth, a convolutional neural network was used to learn discrete CT scan features of UIP, with outputs used to predict the likelihood of UIP using a linear support vector machine. Test performance characteristics were assessed in an independent performance cohort and multicenter ILD clinical cohort. Transplant-free survival was compared between UIP classification approaches using the Kaplan-Meier estimator and Cox proportional hazards regression. A total of 2,907 chest CT scans were included in the training (n = 1,934), validation (n = 408), and performance (n = 565) data sets. The prevalence of radiologist-determined visual UIP was 12.4% and 37.1% in the performance and ILD clinical cohorts, respectively. The DL-based UIP classifier predicted visual UIP in the performance cohort with sensitivity and specificity of 93% and 86%, respectively, and in the multicenter ILD clinical cohort with 81% and 77%, respectively. DL-based and visual UIP classification similarly discriminated survival, and outcomes were consistent among cases with positive DL-based UIP classification irrespective of visual classification. A DL-based classifier for UIP demonstrated good test performance across a wide range of UIP prevalence and similarly discriminated survival when compared with radiologist-determined UIP. This automated tool could efficiently screen for UIP in patients undergoing chest CT scan and identify a high-risk phenotype among those with known ILD.
Sections du résumé
BACKGROUND
BACKGROUND
Because chest CT scan has largely supplanted surgical lung biopsy for diagnosing most cases of interstitial lung disease (ILD), tools to standardize CT scan interpretation are urgently needed.
RESEARCH QUESTION
OBJECTIVE
Does a deep learning (DL)-based classifier for usual interstitial pneumonia (UIP) derived using CT scan features accurately discriminate radiologist-determined visual UIP?
STUDY DESIGN AND METHODS
METHODS
A retrospective cohort study was performed. Chest CT scans acquired in individuals with and without ILD were drawn from a variety of public and private data sources. Using radiologist-determined visual UIP as ground truth, a convolutional neural network was used to learn discrete CT scan features of UIP, with outputs used to predict the likelihood of UIP using a linear support vector machine. Test performance characteristics were assessed in an independent performance cohort and multicenter ILD clinical cohort. Transplant-free survival was compared between UIP classification approaches using the Kaplan-Meier estimator and Cox proportional hazards regression.
RESULTS
RESULTS
A total of 2,907 chest CT scans were included in the training (n = 1,934), validation (n = 408), and performance (n = 565) data sets. The prevalence of radiologist-determined visual UIP was 12.4% and 37.1% in the performance and ILD clinical cohorts, respectively. The DL-based UIP classifier predicted visual UIP in the performance cohort with sensitivity and specificity of 93% and 86%, respectively, and in the multicenter ILD clinical cohort with 81% and 77%, respectively. DL-based and visual UIP classification similarly discriminated survival, and outcomes were consistent among cases with positive DL-based UIP classification irrespective of visual classification.
INTERPRETATION
CONCLUSIONS
A DL-based classifier for UIP demonstrated good test performance across a wide range of UIP prevalence and similarly discriminated survival when compared with radiologist-determined UIP. This automated tool could efficiently screen for UIP in patients undergoing chest CT scan and identify a high-risk phenotype among those with known ILD.
Identifiants
pubmed: 37844797
pii: S0012-3692(23)05569-1
doi: 10.1016/j.chest.2023.10.012
pii:
doi:
Types de publication
Multicenter Study
Journal Article
Langues
eng
Sous-ensembles de citation
IM
Pagination
371-380Subventions
Organisme : NHLBI NIH HHS
ID : R01 HL169166
Pays : United States
Organisme : NHLBI NIH HHS
ID : T32 HL007749
Pays : United States
Informations de copyright
Copyright © 2023 American College of Chest Physicians. Published by Elsevier Inc. All rights reserved.
Déclaration de conflit d'intérêts
Financial/Nonfinancial Disclosures The authors have reported to CHEST the following: L. K., K. L., S. Z., and C. H. are employees of Imbio, Inc. H. G., E. S., S. L., and C. S. are paid consultants for Imbio, Inc. J. H. C., A. A., and J. M. O. report consulting fees from Genentech unrelated to this investigation. None declared (L. C., J. V. P., J. M. W., A. W. M., S. G., A. G.).