Skip to main content
medRxiv
  • Home
  • About
  • Submit
  • ALERTS / RSS
Advanced Search

An Accurate and Explainable Deep Learning System Improves Interobserver Agreement in the Interpretation of Chest Radiograph

Hieu H. Pham, Ha Q. Nguyen, Khanh Lam, Linh T. Le, Dung B. Nguyen, Hieu T. Nguyen, Tung T. Le, Thang V. Nguyen, Minh Dao, Van Vu
doi: https://doi.org/10.1101/2021.09.28.21264286
Hieu H. Pham
1Medical Imaging Center, VinBigdata, Hanoi, Vietnam
2College of Engineering & Computer Science, VinUniversity, Hanoi, Vietnam
  • Find this author on Google Scholar
  • Find this author on PubMed
  • Search for this author on this site
  • For correspondence: hieuhuy01@gmail.com
Ha Q. Nguyen
1Medical Imaging Center, VinBigdata, Hanoi, Vietnam
2College of Engineering & Computer Science, VinUniversity, Hanoi, Vietnam
  • Find this author on Google Scholar
  • Find this author on PubMed
  • Search for this author on this site
Khanh Lam
1Medical Imaging Center, VinBigdata, Hanoi, Vietnam
3Hospital 108, Department of Radiology, Hanoi, Vietnam
  • Find this author on Google Scholar
  • Find this author on PubMed
  • Search for this author on this site
Linh T. Le
1Medical Imaging Center, VinBigdata, Hanoi, Vietnam
4Hanoi Medical University Hospital, Center of Radiology, Hanoi, Vietnam
  • Find this author on Google Scholar
  • Find this author on PubMed
  • Search for this author on this site
Dung B. Nguyen
1Medical Imaging Center, VinBigdata, Hanoi, Vietnam
  • Find this author on Google Scholar
  • Find this author on PubMed
  • Search for this author on this site
Hieu T. Nguyen
1Medical Imaging Center, VinBigdata, Hanoi, Vietnam
  • Find this author on Google Scholar
  • Find this author on PubMed
  • Search for this author on this site
Tung T. Le
1Medical Imaging Center, VinBigdata, Hanoi, Vietnam
  • Find this author on Google Scholar
  • Find this author on PubMed
  • Search for this author on this site
Thang V. Nguyen
1Medical Imaging Center, VinBigdata, Hanoi, Vietnam
  • Find this author on Google Scholar
  • Find this author on PubMed
  • Search for this author on this site
Minh Dao
1Medical Imaging Center, VinBigdata, Hanoi, Vietnam
  • Find this author on Google Scholar
  • Find this author on PubMed
  • Search for this author on this site
Van Vu
1Medical Imaging Center, VinBigdata, Hanoi, Vietnam
5Yale University, Department of Mathematics, New Heaven, CT 06520, United States
  • Find this author on Google Scholar
  • Find this author on PubMed
  • Search for this author on this site
  • Abstract
  • Full Text
  • Info/History
  • Metrics
  • Data/Code
  • Preview PDF
Loading

ABSTRACT

Interpretation of chest radiographs (CXR) is a difficult but essential task for detecting thoracic abnormalities. Recent artificial intelligence (AI) algorithms have achieved radiologist-level performance on various medical classification tasks. However, only a few studies addressed the localization of abnormal findings from CXR scans, which is essential in explaining the image-level classification to radiologists. Additionally, the actual impact of AI algorithms on the diagnostic performance of radiologists in clinical practice remains relatively unclear. To bridge these gaps, we developed an explainable deep learning system called VinDr-CXR that can classify a CXR scan into multiple thoracic diseases and, at the same time, localize most types of critical findings on the image. VinDr-CXR was trained on 51,485 CXR scans with radiologist-provided bounding box annotations. It demonstrated a comparable performance to experienced radiologists in classifying 6 common thoracic diseases on a retrospective validation set of 3,000 CXR scans, with a mean area under the receiver operating characteristic curve (AUROC) of 0.967 (95% confidence interval [CI]: 0.958–0.975). The sensitivity, specificity, F1-score, false-positive rate (FPR), and false-negative rate (FNR) of the system at the optimal cutoff value were 0.933 (0.898–0.964), 0.900 (0.887–0.911), 0.631 (0.589–0.672), 0.101 (0.089– 0.114) and 0.067 (0.057–0.102), respectively. For the localization task with 14 types of lesions, our free-response receiver operating characteristic (FROC) analysis showed that the VinDr-CXR achieved a sensitivity of 80.2% at the rate of 1.0 false-positive lesion identified per scan. A prospective study was also conducted to measure the clinical impact of the VinDr-CXR in assisting six experienced radiologists. The results indicated that the proposed system, when used as a diagnosis supporting tool, significantly improved the agreement between radiologists themselves with an increase of 1.5% in mean Fleiss’ Kappa. We also observed that, after the radiologists consulted VinDr-CXR’s suggestions, the agreement between each of them and the system was remarkably increased by 3.3% in mean Co-hen’s Kappa. Altogether, our results highlight the potentials of the proposed deep learning system as an effective assistant to radiologists in clinical practice. Part of the dataset used for developing the VinDr-CXR system has been made publicly available at https://physionet.org/content/vindr-cxr/1.0.0/.

Competing Interest Statement

The authors have declared no competing interest.

Funding Statement

This research was supported by Vingroup Big Data Institute.

Author Declarations

I confirm all relevant ethical guidelines have been followed, and any necessary IRB and/or ethics committee approvals have been obtained.

Yes

The details of the IRB/oversight body that provided approval or exemption for the research described are given below:

This study was approved by the Institutional Review Boards (IRB) of the Hanoi Medical University Hospital and Hospital 108.

All necessary patient/participant consent has been obtained and the appropriate institutional forms have been archived.

Yes

I understand that all clinical trials and any other prospective interventional studies must be registered with an ICMJE-approved registry, such as ClinicalTrials.gov. I confirm that any such study reported in the manuscript has been registered and the trial registration ID is provided (note: if posting a prospective study registered retrospectively, please provide a statement in the trial ID field explaining why the study was not registered in advance).

Yes

I have followed all appropriate research reporting guidelines and uploaded the relevant EQUATOR Network research reporting checklist(s) and other pertinent material as supplementary files, if applicable.

Yes

Data Availability

The dataset used for developing the deep learning system has been made publicly available at https://physionet.org/content/vindr-cxr/1.0.0/

https://physionet.org/content/vindr-cxr/1.0.0/

Copyright 
The copyright holder for this preprint is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.
Back to top
PreviousNext
Posted September 30, 2021.
Download PDF
Data/Code
Email

Thank you for your interest in spreading the word about medRxiv.

NOTE: Your email address is requested solely to identify you as the sender of this article.

Enter multiple addresses on separate lines or separate them with commas.
An Accurate and Explainable Deep Learning System Improves Interobserver Agreement in the Interpretation of Chest Radiograph
(Your Name) has forwarded a page to you from medRxiv
(Your Name) thought you would like to see this page from the medRxiv website.
CAPTCHA
This question is for testing whether or not you are a human visitor and to prevent automated spam submissions.
Share
An Accurate and Explainable Deep Learning System Improves Interobserver Agreement in the Interpretation of Chest Radiograph
Hieu H. Pham, Ha Q. Nguyen, Khanh Lam, Linh T. Le, Dung B. Nguyen, Hieu T. Nguyen, Tung T. Le, Thang V. Nguyen, Minh Dao, Van Vu
medRxiv 2021.09.28.21264286; doi: https://doi.org/10.1101/2021.09.28.21264286
Digg logo Reddit logo Twitter logo Facebook logo Google logo LinkedIn logo Mendeley logo
Citation Tools
An Accurate and Explainable Deep Learning System Improves Interobserver Agreement in the Interpretation of Chest Radiograph
Hieu H. Pham, Ha Q. Nguyen, Khanh Lam, Linh T. Le, Dung B. Nguyen, Hieu T. Nguyen, Tung T. Le, Thang V. Nguyen, Minh Dao, Van Vu
medRxiv 2021.09.28.21264286; doi: https://doi.org/10.1101/2021.09.28.21264286

Citation Manager Formats

  • BibTeX
  • Bookends
  • EasyBib
  • EndNote (tagged)
  • EndNote 8 (xml)
  • Medlars
  • Mendeley
  • Papers
  • RefWorks Tagged
  • Ref Manager
  • RIS
  • Zotero
  • Tweet Widget
  • Facebook Like
  • Google Plus One

Subject Area

  • Health Informatics
Subject Areas
All Articles
  • Addiction Medicine (160)
  • Allergy and Immunology (412)
  • Anesthesia (90)
  • Cardiovascular Medicine (855)
  • Dentistry and Oral Medicine (156)
  • Dermatology (97)
  • Emergency Medicine (247)
  • Endocrinology (including Diabetes Mellitus and Metabolic Disease) (392)
  • Epidemiology (8542)
  • Forensic Medicine (4)
  • Gastroenterology (383)
  • Genetic and Genomic Medicine (1741)
  • Geriatric Medicine (167)
  • Health Economics (371)
  • Health Informatics (1235)
  • Health Policy (618)
  • Health Systems and Quality Improvement (467)
  • Hematology (196)
  • HIV/AIDS (371)
  • Infectious Diseases (except HIV/AIDS) (10274)
  • Intensive Care and Critical Care Medicine (552)
  • Medical Education (192)
  • Medical Ethics (51)
  • Nephrology (210)
  • Neurology (1668)
  • Nursing (97)
  • Nutrition (248)
  • Obstetrics and Gynecology (325)
  • Occupational and Environmental Health (450)
  • Oncology (925)
  • Ophthalmology (263)
  • Orthopedics (100)
  • Otolaryngology (172)
  • Pain Medicine (111)
  • Palliative Medicine (40)
  • Pathology (251)
  • Pediatrics (534)
  • Pharmacology and Therapeutics (246)
  • Primary Care Research (206)
  • Psychiatry and Clinical Psychology (1760)
  • Public and Global Health (3829)
  • Radiology and Imaging (622)
  • Rehabilitation Medicine and Physical Therapy (318)
  • Respiratory Medicine (518)
  • Rheumatology (207)
  • Sexual and Reproductive Health (165)
  • Sports Medicine (156)
  • Surgery (190)
  • Toxicology (36)
  • Transplantation (100)
  • Urology (74)