Automated interstitial lung abnormalities detection at CT: external validation and potential recognition of traction bronchiectasis/bronchiolectasis

Yusei Nakamura; Taiki Fukuda; Kota Aoyagi; Masami Kawagishi; Yuki Ko; Noriaki Wada; Takuya Hino; Tomoyuki Hida; Meike W Vernooij; Daniel Bos; Daan W Loth; Masahiro Ozaki; Akihiro Koga; Heida Bjarnadottir; Valborg Gudmundsdottir; Gunnar Gudmundsson; Vilmundur Gudnason; Mizuki Nishino; David C Christiani; Gary M Hunninghake; Kousei Ishigami; Hiroto Hatabu

doi:10.1007/s11604-025-01917-z

Automated interstitial lung abnormalities detection at CT: external validation and potential recognition of traction bronchiectasis/bronchiolectasis

Yusei Nakamura^*
, Taiki Fukuda
, Kota Aoyagi
, Masami Kawagishi
, Yuki Ko
, Noriaki Wada
, Takuya Hino
, Tomoyuki Hida
, Meike W Vernooij
, Daniel Bos
, Daan W Loth
, Masahiro Ozaki
, Akihiro Koga
, Heida Bjarnadottir
, Valborg Gudmundsdottir
, Gunnar Gudmundsson
, Vilmundur Gudnason
, Mizuki Nishino
, David C Christiani
, Gary M Hunninghake

Kousei Ishigami, Hiroto Hatabu

^*Corresponding author for this work

Brigham and Women’s Hospital
Canon Medical Systems Corporation
Canon Inc.
Kyushu University
Amphia Hospital
Institute for Advanced Diagnosis for Rare Diseases & Conditions K.K.
University of Iceland
Massachusetts General Hospital and Harvard Medical School

Research output: Contribution to journal › Article › Academic › peer-review

5 Downloads (Pure)

Abstract

PURPOSE:

An artificial intelligence (AI) system for detecting interstitial lung abnormalities (ILA) was previously developed but requires external validation. This study aimed to examine the robustness across different populations and investigate associations between the system outputs and traction bronchiectasis/bronchiolectasis severity patterns.

MATERIALS AND METHODS:

CT scans from population-based samples of the Rotterdam Study (2018-2019) and the Age Gene/Environment Susceptibility Reykjavik (AGES-Reykjavik) Study (baseline CT: 2002-2006, follow-up CT: 2007-2011) were used in this secondary analysis of the two cohorts. The AI system calculated ILA probability score (AI score) in the range from 0 to 1. Three experienced readers evaluated independently all CT scans for ILA, and two chest radiologists assessed traction bronchiectasis/bronchiolectasis using the 4-scale traction bronchiectasis/bronchiolectasis index (TBI) for severity by consensus. Receiver operating characteristic (ROC) analysis and Kruskal-Wallis test were used for statistical analysis.

RESULTS:

The system analyzed 932 CT scans of the Rotterdam Study (mean participant age, 79.6 years ± 4.3 (SD), 482 women) and 5242 CT scans of the AGES-Reykjavik Study (mean participant age, 76.4 years ± 5.6, 3032 women), and achieved area under the ROC curve of 0.841 (95% CI 0.804, 0.879) and 0.823 (95% CI 0.798, 0.847), respectively. AI scores correlated with readers' certainty, decreasing from unanimous ILA cases to No-ILA cases. Higher baseline AI scores correlated with greater severity of traction bronchiectasis/bronchiolectasis (TBI-3: 0.931 [IQR, 0.911-0.932], TBI-2: 0.738 [IQR, 0.406-0.880], TBI-1: 0.537 [IQR, 0.317-0.761], TBI-0: 0.250 [IQR, 0.136-0.455]).

CONCLUSION:

The system demonstrated robust ILA detection performance across different populations, with AI scores showing associations with traction bronchiectasis/bronchiolectasis severity.

Original language	English
Pages (from-to)	660-672
Number of pages	13
Journal	Japanese Journal of Radiology
Volume	44
Issue number	4
Early online date	11 Dec 2025
DOIs	https://doi.org/10.1007/s11604-025-01917-z
Publication status	Published - Apr 2026

Bibliographical note

Publisher Copyright:
© The Author(s) 2025.

Access to Document

10.1007/s11604-025-01917-zLicence: CC BY

Automated interstitial lung abnormalities detection at CTFinal published version, 2.02 MBLicence: CC BY

Cite this

Nakamura, Y., Fukuda, T., Aoyagi, K., Kawagishi, M., Ko, Y., Wada, N., Hino, T., Hida, T., Vernooij, M. W., Bos, D., Loth, D. W., Ozaki, M., Koga, A., Bjarnadottir, H., Gudmundsdottir, V., Gudmundsson, G., Gudnason, V., Nishino, M., Christiani, D. C., ... Hatabu, H. (2026). Automated interstitial lung abnormalities detection at CT: external validation and potential recognition of traction bronchiectasis/bronchiolectasis. Japanese Journal of Radiology, 44(4), 660-672. https://doi.org/10.1007/s11604-025-01917-z

@article{4a4fa88c871c43f987936cca8015ed60,

title = "Automated interstitial lung abnormalities detection at CT: external validation and potential recognition of traction bronchiectasis/bronchiolectasis",

abstract = "PURPOSE: An artificial intelligence (AI) system for detecting interstitial lung abnormalities (ILA) was previously developed but requires external validation. This study aimed to examine the robustness across different populations and investigate associations between the system outputs and traction bronchiectasis/bronchiolectasis severity patterns.MATERIALS AND METHODS: CT scans from population-based samples of the Rotterdam Study (2018-2019) and the Age Gene/Environment Susceptibility Reykjavik (AGES-Reykjavik) Study (baseline CT: 2002-2006, follow-up CT: 2007-2011) were used in this secondary analysis of the two cohorts. The AI system calculated ILA probability score (AI score) in the range from 0 to 1. Three experienced readers evaluated independently all CT scans for ILA, and two chest radiologists assessed traction bronchiectasis/bronchiolectasis using the 4-scale traction bronchiectasis/bronchiolectasis index (TBI) for severity by consensus. Receiver operating characteristic (ROC) analysis and Kruskal-Wallis test were used for statistical analysis.RESULTS: The system analyzed 932 CT scans of the Rotterdam Study (mean participant age, 79.6 years ± 4.3 (SD), 482 women) and 5242 CT scans of the AGES-Reykjavik Study (mean participant age, 76.4 years ± 5.6, 3032 women), and achieved area under the ROC curve of 0.841 (95\% CI 0.804, 0.879) and 0.823 (95\% CI 0.798, 0.847), respectively. AI scores correlated with readers' certainty, decreasing from unanimous ILA cases to No-ILA cases. Higher baseline AI scores correlated with greater severity of traction bronchiectasis/bronchiolectasis (TBI-3: 0.931 [IQR, 0.911-0.932], TBI-2: 0.738 [IQR, 0.406-0.880], TBI-1: 0.537 [IQR, 0.317-0.761], TBI-0: 0.250 [IQR, 0.136-0.455]).CONCLUSION: The system demonstrated robust ILA detection performance across different populations, with AI scores showing associations with traction bronchiectasis/bronchiolectasis severity.",

author = "Yusei Nakamura and Taiki Fukuda and Kota Aoyagi and Masami Kawagishi and Yuki Ko and Noriaki Wada and Takuya Hino and Tomoyuki Hida and Vernooij, \{Meike W\} and Daniel Bos and Loth, \{Daan W\} and Masahiro Ozaki and Akihiro Koga and Heida Bjarnadottir and Valborg Gudmundsdottir and Gunnar Gudmundsson and Vilmundur Gudnason and Mizuki Nishino and Christiani, \{David C\} and Hunninghake, \{Gary M\} and Kousei Ishigami and Hiroto Hatabu",

note = "Publisher Copyright: {\textcopyright} The Author(s) 2025.",

year = "2026",

month = apr,

doi = "10.1007/s11604-025-01917-z",

language = "English",

volume = "44",

pages = "660--672",

journal = "Japanese Journal of Radiology",

issn = "1867-1071",

publisher = "Springer",

number = "4",

}

Nakamura, Y, Fukuda, T, Aoyagi, K, Kawagishi, M, Ko, Y, Wada, N, Hino, T, Hida, T, Vernooij, MW , Bos, D, Loth, DW, Ozaki, M, Koga, A, Bjarnadottir, H, Gudmundsdottir, V, Gudmundsson, G, Gudnason, V, Nishino, M, Christiani, DC, Hunninghake, GM, Ishigami, K & Hatabu, H 2026, 'Automated interstitial lung abnormalities detection at CT: external validation and potential recognition of traction bronchiectasis/bronchiolectasis', Japanese Journal of Radiology, vol. 44, no. 4, pp. 660-672. https://doi.org/10.1007/s11604-025-01917-z

TY - JOUR

T1 - Automated interstitial lung abnormalities detection at CT

T2 - external validation and potential recognition of traction bronchiectasis/bronchiolectasis

AU - Nakamura, Yusei

AU - Fukuda, Taiki

AU - Aoyagi, Kota

AU - Kawagishi, Masami

AU - Ko, Yuki

AU - Wada, Noriaki

AU - Hino, Takuya

AU - Hida, Tomoyuki

AU - Vernooij, Meike W

AU - Bos, Daniel

AU - Loth, Daan W

AU - Ozaki, Masahiro

AU - Koga, Akihiro

AU - Bjarnadottir, Heida

AU - Gudmundsdottir, Valborg

AU - Gudmundsson, Gunnar

AU - Gudnason, Vilmundur

AU - Nishino, Mizuki

AU - Christiani, David C

AU - Hunninghake, Gary M

AU - Ishigami, Kousei

AU - Hatabu, Hiroto

PY - 2026/4

Y1 - 2026/4

N2 - PURPOSE: An artificial intelligence (AI) system for detecting interstitial lung abnormalities (ILA) was previously developed but requires external validation. This study aimed to examine the robustness across different populations and investigate associations between the system outputs and traction bronchiectasis/bronchiolectasis severity patterns.MATERIALS AND METHODS: CT scans from population-based samples of the Rotterdam Study (2018-2019) and the Age Gene/Environment Susceptibility Reykjavik (AGES-Reykjavik) Study (baseline CT: 2002-2006, follow-up CT: 2007-2011) were used in this secondary analysis of the two cohorts. The AI system calculated ILA probability score (AI score) in the range from 0 to 1. Three experienced readers evaluated independently all CT scans for ILA, and two chest radiologists assessed traction bronchiectasis/bronchiolectasis using the 4-scale traction bronchiectasis/bronchiolectasis index (TBI) for severity by consensus. Receiver operating characteristic (ROC) analysis and Kruskal-Wallis test were used for statistical analysis.RESULTS: The system analyzed 932 CT scans of the Rotterdam Study (mean participant age, 79.6 years ± 4.3 (SD), 482 women) and 5242 CT scans of the AGES-Reykjavik Study (mean participant age, 76.4 years ± 5.6, 3032 women), and achieved area under the ROC curve of 0.841 (95% CI 0.804, 0.879) and 0.823 (95% CI 0.798, 0.847), respectively. AI scores correlated with readers' certainty, decreasing from unanimous ILA cases to No-ILA cases. Higher baseline AI scores correlated with greater severity of traction bronchiectasis/bronchiolectasis (TBI-3: 0.931 [IQR, 0.911-0.932], TBI-2: 0.738 [IQR, 0.406-0.880], TBI-1: 0.537 [IQR, 0.317-0.761], TBI-0: 0.250 [IQR, 0.136-0.455]).CONCLUSION: The system demonstrated robust ILA detection performance across different populations, with AI scores showing associations with traction bronchiectasis/bronchiolectasis severity.

AB - PURPOSE: An artificial intelligence (AI) system for detecting interstitial lung abnormalities (ILA) was previously developed but requires external validation. This study aimed to examine the robustness across different populations and investigate associations between the system outputs and traction bronchiectasis/bronchiolectasis severity patterns.MATERIALS AND METHODS: CT scans from population-based samples of the Rotterdam Study (2018-2019) and the Age Gene/Environment Susceptibility Reykjavik (AGES-Reykjavik) Study (baseline CT: 2002-2006, follow-up CT: 2007-2011) were used in this secondary analysis of the two cohorts. The AI system calculated ILA probability score (AI score) in the range from 0 to 1. Three experienced readers evaluated independently all CT scans for ILA, and two chest radiologists assessed traction bronchiectasis/bronchiolectasis using the 4-scale traction bronchiectasis/bronchiolectasis index (TBI) for severity by consensus. Receiver operating characteristic (ROC) analysis and Kruskal-Wallis test were used for statistical analysis.RESULTS: The system analyzed 932 CT scans of the Rotterdam Study (mean participant age, 79.6 years ± 4.3 (SD), 482 women) and 5242 CT scans of the AGES-Reykjavik Study (mean participant age, 76.4 years ± 5.6, 3032 women), and achieved area under the ROC curve of 0.841 (95% CI 0.804, 0.879) and 0.823 (95% CI 0.798, 0.847), respectively. AI scores correlated with readers' certainty, decreasing from unanimous ILA cases to No-ILA cases. Higher baseline AI scores correlated with greater severity of traction bronchiectasis/bronchiolectasis (TBI-3: 0.931 [IQR, 0.911-0.932], TBI-2: 0.738 [IQR, 0.406-0.880], TBI-1: 0.537 [IQR, 0.317-0.761], TBI-0: 0.250 [IQR, 0.136-0.455]).CONCLUSION: The system demonstrated robust ILA detection performance across different populations, with AI scores showing associations with traction bronchiectasis/bronchiolectasis severity.

UR - https://www.scopus.com/pages/publications/105024659579

U2 - 10.1007/s11604-025-01917-z

DO - 10.1007/s11604-025-01917-z

M3 - Article

C2 - 41372552

SN - 1867-1071

VL - 44

SP - 660

EP - 672

JO - Japanese Journal of Radiology

JF - Japanese Journal of Radiology

IS - 4

ER -

Automated interstitial lung abnormalities detection at CT: external validation and potential recognition of traction bronchiectasis/bronchiolectasis

Abstract

Bibliographical note

Access to Document

Other files and links

Fingerprint

Cite this