Harnessing deep learning to detect bronchiolitis obliterans syndrome from chest CT
Abstract Background Bronchiolitis Obliterans Syndrome (BOS), a fibrotic airway disease that may develop after lung transplantation, conventionally relies on pulmonary function tests (PFTs) for diagnosis due to limitations of CT imaging. Deep neural networks (DNNs) have not previously been used for B...
Saved in:
| Main Authors: | , , , , , , , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Nature Portfolio
2025-01-01
|
| Series: | Communications Medicine |
| Online Access: | https://doi.org/10.1038/s43856-025-00732-x |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832594488561762304 |
|---|---|
| author | Mateusz Koziński Doruk Oner Jakub Gwizdała Catherine Beigelman-Aubry Pascal Fua Angela Koutsokera Alessio Casutt Argyro Vraka Michele De Palma John-David Aubert Horst Bischof Christophe von Garnier Sahand Jamal Rahi Martin Urschler Nahal Mansouri |
| author_facet | Mateusz Koziński Doruk Oner Jakub Gwizdała Catherine Beigelman-Aubry Pascal Fua Angela Koutsokera Alessio Casutt Argyro Vraka Michele De Palma John-David Aubert Horst Bischof Christophe von Garnier Sahand Jamal Rahi Martin Urschler Nahal Mansouri |
| author_sort | Mateusz Koziński |
| collection | DOAJ |
| description | Abstract Background Bronchiolitis Obliterans Syndrome (BOS), a fibrotic airway disease that may develop after lung transplantation, conventionally relies on pulmonary function tests (PFTs) for diagnosis due to limitations of CT imaging. Deep neural networks (DNNs) have not previously been used for BOS detection. This study aims to train a DNN to detect BOS in CT scans using an approach tailored for low-data scenarios. Methods We trained a DNN to detect BOS in CT scans using a co-training method designed to enhance performance in low-data environments. Our method employs an auxiliary task that makes the DNN more sensitive to disease manifestations and less sensitive to the patient’s anatomical features. The DNN was tasked with predicting the sequence of two CT scans taken from the same BOS patient at least six months apart. We evaluated this approach on CT scans from 75 post-transplant patients, including 26 with BOS, and used a ROC-AUC metric to assess performance. Results We show that our DNN method achieves a ROC-AUC of 0.90 (95% CI: 0.840–0.953) in distinguishing BOS from non-BOS in CT scans. Performance correlates with BOS progression, with ROC-AUC values of 0.88 for stage I, 0.91 for stage II, and 0.94 for stage III BOS. Notably, the DNN shows comparable performance on standard- and high-resolution CT scans. It also demonstrates the ability to predict BOS in at-risk patients (FEV1 between 80% and 90% of best FEV1) with a ROC-AUC of 0.87 (95% CI: 0.735–0.974). Using visual interpretation techniques for DNNs, we reveal sensitivity to hyperlucent/hypoattenuated areas indicative of air-trapping or bronchiectasis. Conclusions Our approach shows potential for improving BOS diagnosis by enabling early detection and management. The ability to detect BOS from standard-resolution scans at any stage of respiration makes this method more accessible than previous approaches. Additionally, our findings highlight that techniques to limit overfitting are crucial for unlocking the potential of DNNs in low-data settings, which could assist clinicians in BOS studies with limited patient data. |
| format | Article |
| id | doaj-art-7b44a2e7c7b043c7888fbd888565e81f |
| institution | Kabale University |
| issn | 2730-664X |
| language | English |
| publishDate | 2025-01-01 |
| publisher | Nature Portfolio |
| record_format | Article |
| series | Communications Medicine |
| spelling | doaj-art-7b44a2e7c7b043c7888fbd888565e81f2025-01-19T12:36:54ZengNature PortfolioCommunications Medicine2730-664X2025-01-015111110.1038/s43856-025-00732-xHarnessing deep learning to detect bronchiolitis obliterans syndrome from chest CTMateusz Koziński0Doruk Oner1Jakub Gwizdała2Catherine Beigelman-Aubry3Pascal Fua4Angela Koutsokera5Alessio Casutt6Argyro Vraka7Michele De Palma8John-David Aubert9Horst Bischof10Christophe von Garnier11Sahand Jamal Rahi12Martin Urschler13Nahal Mansouri14Institute of Computer Graphics and Vision, Technische Universität GrazNeuraVision Research Lab, Department of Computer Engineering, Bilkent UniversityComputer Vision Laboratory, School of Computer and Communication Sciences – IC, École Polytechnique Fédérale de Lausanne (EPFL)Department of Diagnostic and Interventional Radiology, Lausanne University Hospital (CHUV)Computer Vision Laboratory, School of Computer and Communication Sciences – IC, École Polytechnique Fédérale de Lausanne (EPFL)Division of Pulmonary Medicine, Department of Medicine, Lausanne University Hospital (CHUV), University of Lausanne (UNIL)Division of Pulmonary Medicine, Department of Medicine, Lausanne University Hospital (CHUV), University of Lausanne (UNIL)Division of Pulmonary Medicine, Department of Medicine, Lausanne University Hospital (CHUV), University of Lausanne (UNIL)Swiss Institute for Experimental Cancer Research (ISREC), School of Life Sciences, École Polytechnique Fédérale de Lausanne (EPFL)Division of Pulmonary Medicine, Department of Medicine, Lausanne University Hospital (CHUV), University of Lausanne (UNIL)Institute of Computer Graphics and Vision, Technische Universität GrazDivision of Pulmonary Medicine, Department of Medicine, Lausanne University Hospital (CHUV), University of Lausanne (UNIL)Laboratory of the Physics of Biological Systems, Institute of Physics, École Polytechnique Fédérale de Lausanne (EPFL)Institute for Medical Informatics, Statistics and Documentation, Medical University of GrazDivision of Pulmonary Medicine, Department of Medicine, Lausanne University Hospital (CHUV), University of Lausanne (UNIL)Abstract Background Bronchiolitis Obliterans Syndrome (BOS), a fibrotic airway disease that may develop after lung transplantation, conventionally relies on pulmonary function tests (PFTs) for diagnosis due to limitations of CT imaging. Deep neural networks (DNNs) have not previously been used for BOS detection. This study aims to train a DNN to detect BOS in CT scans using an approach tailored for low-data scenarios. Methods We trained a DNN to detect BOS in CT scans using a co-training method designed to enhance performance in low-data environments. Our method employs an auxiliary task that makes the DNN more sensitive to disease manifestations and less sensitive to the patient’s anatomical features. The DNN was tasked with predicting the sequence of two CT scans taken from the same BOS patient at least six months apart. We evaluated this approach on CT scans from 75 post-transplant patients, including 26 with BOS, and used a ROC-AUC metric to assess performance. Results We show that our DNN method achieves a ROC-AUC of 0.90 (95% CI: 0.840–0.953) in distinguishing BOS from non-BOS in CT scans. Performance correlates with BOS progression, with ROC-AUC values of 0.88 for stage I, 0.91 for stage II, and 0.94 for stage III BOS. Notably, the DNN shows comparable performance on standard- and high-resolution CT scans. It also demonstrates the ability to predict BOS in at-risk patients (FEV1 between 80% and 90% of best FEV1) with a ROC-AUC of 0.87 (95% CI: 0.735–0.974). Using visual interpretation techniques for DNNs, we reveal sensitivity to hyperlucent/hypoattenuated areas indicative of air-trapping or bronchiectasis. Conclusions Our approach shows potential for improving BOS diagnosis by enabling early detection and management. The ability to detect BOS from standard-resolution scans at any stage of respiration makes this method more accessible than previous approaches. Additionally, our findings highlight that techniques to limit overfitting are crucial for unlocking the potential of DNNs in low-data settings, which could assist clinicians in BOS studies with limited patient data.https://doi.org/10.1038/s43856-025-00732-x |
| spellingShingle | Mateusz Koziński Doruk Oner Jakub Gwizdała Catherine Beigelman-Aubry Pascal Fua Angela Koutsokera Alessio Casutt Argyro Vraka Michele De Palma John-David Aubert Horst Bischof Christophe von Garnier Sahand Jamal Rahi Martin Urschler Nahal Mansouri Harnessing deep learning to detect bronchiolitis obliterans syndrome from chest CT Communications Medicine |
| title | Harnessing deep learning to detect bronchiolitis obliterans syndrome from chest CT |
| title_full | Harnessing deep learning to detect bronchiolitis obliterans syndrome from chest CT |
| title_fullStr | Harnessing deep learning to detect bronchiolitis obliterans syndrome from chest CT |
| title_full_unstemmed | Harnessing deep learning to detect bronchiolitis obliterans syndrome from chest CT |
| title_short | Harnessing deep learning to detect bronchiolitis obliterans syndrome from chest CT |
| title_sort | harnessing deep learning to detect bronchiolitis obliterans syndrome from chest ct |
| url | https://doi.org/10.1038/s43856-025-00732-x |
| work_keys_str_mv | AT mateuszkozinski harnessingdeeplearningtodetectbronchiolitisobliteranssyndromefromchestct AT dorukoner harnessingdeeplearningtodetectbronchiolitisobliteranssyndromefromchestct AT jakubgwizdała harnessingdeeplearningtodetectbronchiolitisobliteranssyndromefromchestct AT catherinebeigelmanaubry harnessingdeeplearningtodetectbronchiolitisobliteranssyndromefromchestct AT pascalfua harnessingdeeplearningtodetectbronchiolitisobliteranssyndromefromchestct AT angelakoutsokera harnessingdeeplearningtodetectbronchiolitisobliteranssyndromefromchestct AT alessiocasutt harnessingdeeplearningtodetectbronchiolitisobliteranssyndromefromchestct AT argyrovraka harnessingdeeplearningtodetectbronchiolitisobliteranssyndromefromchestct AT micheledepalma harnessingdeeplearningtodetectbronchiolitisobliteranssyndromefromchestct AT johndavidaubert harnessingdeeplearningtodetectbronchiolitisobliteranssyndromefromchestct AT horstbischof harnessingdeeplearningtodetectbronchiolitisobliteranssyndromefromchestct AT christophevongarnier harnessingdeeplearningtodetectbronchiolitisobliteranssyndromefromchestct AT sahandjamalrahi harnessingdeeplearningtodetectbronchiolitisobliteranssyndromefromchestct AT martinurschler harnessingdeeplearningtodetectbronchiolitisobliteranssyndromefromchestct AT nahalmansouri harnessingdeeplearningtodetectbronchiolitisobliteranssyndromefromchestct |