Accuracy of an automated system for tuberculosis detection on chest radiographs in high-risk screening

J. Melendez*, L. Hogeweg, C. I. Sánchez, R. H.H.M. Philipsen, R. W. Aldridge, A. C. Hayward, Ibrahim Abubakar, B. Van Ginneken, Alistair Story

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

11 Citations (Scopus)

Abstract

S E T T I N G: Tuberculosis (TB) screening programmes can be optimised by reducing the number of chest radiographs (CXRs) requiring interpretation by human experts. O B J E C T I V E: To evaluate the performance of computerised detection software in triaging CXRs in a high-throughput digital mobile TB screening programme. D E S I G N: A retrospective evaluation of the software was performed on a database of 38 961 postero-anterior CXRs from unique individuals seen between 2005 and 2010, 87 of whom were diagnosed with TB. The software generated a TB likelihood score for each CXR. This score was compared with a reference standard for notified active pulmonary TB using receiver operating characteristic (ROC) curve and localisation ROC (LROC) curve analyses. R E S U L T S: On ROC curve analysis, software specificity was 55.71% (95%CI 55.21-56.20) and negative predictive value was 99.98% (95%CI 99.95-99.99), at a sensitivity of 95%. The area under the ROC curve was 0.90 (95%CI 0.86-0.93). Results of the LROC curve analysis were similar. C O N C L U S I O N: The software could identify more than half of the normal images in a TB screening setting while maintaining high sensitivity, and may therefore be used for triage.

Original languageEnglish
Pages (from-to)567-571
Number of pages5
JournalInternational Journal of Tuberculosis and Lung Disease
Volume22
Issue number5
DOIs
Publication statusPublished - 1 May 2018
Externally publishedYes

Bibliographical note

Funding Information:
LH was supported by the European & Developing Countries Clinical Trials Partnership (‘Evaluation of multiple novel and emerging technologies for TB diagnosis, in smear-negative and HIV-infected persons, in high burden countries’ [the TB-NEAT] project); RWA by the Wellcome Trust, London [206602/Z/17/Z]; ACH (2009–2014) by the National Institute for Health Research (NIHR), London, UK; and IA by an NIHR Senior Research Fellowship.

Funding Information:
The authors thank J Knight and D Taubman, the two reporting radiographers on the mobile X-ray unit in London, UK, who collected all of the chest X-rays, and S Thanabalasingham who quality assures the Find&Treat screening service. LH was supported by the European & Developing Countries Clinical Trials Partnership ('Evaluation of multiple novel and emerging technologies for TB diagnosis, in smear-negative and HIV-infected persons, in high burden countries' [the TB-NEAT] project); RWA by the Wellcome Trust, London [206602/Z/17/Z]; ACH (2009-2014) by the National Institute for Health Research (NIHR), London, UK; and IA by an NIHR Senior Research Fellowship. Conflict of interest: JM and RHHMP are employees of Thirona, Nijmegen, The Netherlands, which develops CAD4TB software. BvG is co-founder and shareholder of Thirona. The remaining authors report no conflicts. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Publisher Copyright:
© 2018 Melendez et al.

Keywords

  • Chest radiography
  • Computer-aided detection
  • Computerised image analysis
  • TB

Fingerprint

Dive into the research topics of 'Accuracy of an automated system for tuberculosis detection on chest radiographs in high-risk screening'. Together they form a unique fingerprint.

Cite this