Articles Vol. 63 No. 5 (2022) 29/09/2022

SCREENING FOR PULMONARY TUBERCULOSIS USING CHEST X-RAY WITH COMPUTER-AIDED DETECTION SYSTEMS: A LITERATURE REVIEW

Dao Huy Cu1,2, Le Thi Hoan3,4, Cung Van Cong1,2
1 National Lung Hospital
2 Bệnh viện Phổi Trung ương
3 Hanoi Medical University
4 Đại học Y Hà Nội
Corresponding author: huycubn119@gmail.com
DOI: 10.52163/yhc.v63i5.440
22 Views
0 Downloads
Keywords
TB screening, Pulmonary tuberculosis, Chest X-ray, Computer-aided detection.
Abstract

Objective: To review the literature on pulmonary TB screening using automatic X-ray film reading
systems.
Subjects and methods: 13 publications from 2 databases MEDLINE and Cochrance in the period
from January 2010 to December 2021 were analyzed.
Results: Our study has shown that there are currently 12 automatic chest X-ray reading systems that
have been applied in screening for pulmonary tuberculosis. The sensitivities of the systems ranged
from 0.70 to 0.95 with specificities of 0.42 to 0.99, respectively. Of these, there are 3 systems that
achieve the sensitivity and specificity recommended by WHO for a TB screening tool (sensitivity
>=90% and specificity >=70%) namely: qXR, CAD4TB and INSIGHT CXR.
Conclusion: Our study has highlighted the overview and value of some AI software applications in
reading chest X-ray automatically supporting the diagnosis of pulmonary tuberculosis.

References
[1]
“Global tuberculosis report 2020.” https://www. Google Scholar
[2]
who.int/publications/i/item/9789240013131 Google Scholar
[3]
(accessed Apr. 16, 2021). Google Scholar
[4]
Nguyen HV, “The second national tuberculosis Google Scholar
[5]
prevalence survey in Vietnam,” PLoS One, vol. Google Scholar
[6]
, no. 4, p. e0232142, 2020, doi: 10.1371/ Google Scholar
[7]
journal.pone.0232142. Google Scholar
[8]
Li B, “Prevalence of pulmonary tuberculosis in Google Scholar
[9]
Tibet Autonomous Region, China, 2014,” Int J Google Scholar
[10]
Tuberc Lung Dis, vol. 23, no. 6, pp. 735–740, Google Scholar
[11]
Jun. 2019, doi: 10.5588/ijtld.18.0614. Google Scholar
[12]
Law I, Floyd K, and African TB Prevalence Google Scholar
[13]
Survey Group, “National tuberculosis prevalence Google Scholar
[14]
surveys in Africa, 2008-2016: an overview of Google Scholar
[15]
results and lessons learned,” Trop Med Int Health, Google Scholar
[16]
vol. 25, no. 11, pp. 1308–1327, Nov. 2020, doi: Google Scholar
[17]
1111/tmi.13485. Google Scholar
[18]
Migambi P, “Prevalence of tuberculosis in Google Scholar
[19]
Rwanda: Results of the first nationwide survey in Google Scholar
[20]
yielded important lessons for TB control,” Google Scholar
[21]
PLoS One, vol. 15, no. 4, p. e0231372, 2020, doi: Google Scholar
[22]
1371/journal.pone.0231372. Google Scholar
[23]
“World Health Organization: What is DOTS?: Google Scholar
[24]
a guide... - Google Scholar.” https://scholar.google.com/scholar_lookup?title=What Google Scholar
[25]
+is+DOTS?+A+Guide+to+Understanding+the+WHO%E2%80%94Recommended+TB+- Google Scholar
[26]
Control.+Strategy+Known+as+DOTS&publication_year=1999& (accessed Mar. 21, 2022). Google Scholar
[27]
Qin ZZ, “Tuberculosis detection from chest Google Scholar
[28]
x-rays for triaging in a high tuberculosis-burden Google Scholar
[29]
setting: an evaluation of five artificial intelligence Google Scholar
[30]
algorithms,” The Lancet. Digital health, vol. 3, Google Scholar
[31]
no. 9, pp. e543–e554, Sep. 2021, doi: 10.1016/ Google Scholar
[32]
s2589-7500(21)00116-3. Google Scholar
[33]
Codlin AJ, “Independent evaluation of 12 Google Scholar
[34]
artificial intelligence solutions for the detection Google Scholar
[35]
of tuberculosis,” Scientific reports, vol. 11, no. 1, Google Scholar
[36]
p. 23895, Dec. 2021, doi: 10.1038/s41598-021- Google Scholar
[37]
-0. Google Scholar
[38]
Nash M, “Deep learning, computer-aided Google Scholar
[39]
radiography reading for tuberculosis: a diagnostic Google Scholar
[40]
accuracy study from a tertiary hospital in India,” Google Scholar
[41]
Scientific reports, vol. 10, no. 1, p. 210, Jan. 2020, Google Scholar
[42]
doi: 10.1038/s41598-019-56589-3. Google Scholar
[43]
Habib SS, “Evaluation of computer aided Google Scholar
[44]
detection of tuberculosis on chest radiography Google Scholar
[45]
among people with diabetes in Karachi Pakistan,” Google Scholar
[46]
Scientific reports, vol. 10, no. 1, p. 6276, Apr. Google Scholar
[47]
, doi: 10.1038/s41598-020-63084-7. Google Scholar
[48]
Murphy K, “Computer aided detection of Google Scholar
[49]
tuberculosis on chest radiographs: An evaluation Google Scholar
[50]
of the CAD4TB v6 system,” Scientific reports, Google Scholar
[51]
vol. 10, no. 1, p. 5492, Mar. 2020, doi: 10.1038/ Google Scholar
[52]
s41598-020-62148-y. Google Scholar
[53]
Khan FA, “Chest x-ray analysis with deep Google Scholar
[54]
learning-based software as a triage test for Google Scholar
[55]
pulmonary tuberculosis: a prospective study Google Scholar
[56]
of diagnostic accuracy for culture-confirmed Google Scholar
[57]
disease,” The Lancet. Digital health, vol. 2, no. Google Scholar
[58]
, pp. e573–e581, Nov. 2020, doi: 10.1016/ Google Scholar
[59]
s2589-7500(20)30221-1. Google Scholar
[60]
Qin ZZ, “Using artificial intelligence to read chest Google Scholar
[61]
radiographs for tuberculosis detection: A multisite evaluation of the diagnostic accuracy of three Google Scholar
[62]
deep learning systems,” Scientific reports, vol. 9, Google Scholar
[63]
no. 1, p. 15000, Oct. 2019, doi: 10.1038/s41598- Google Scholar
[64]
-51503-3. Google Scholar
[65]
Zaidi SMA, “Evaluation of the diagnostic Google Scholar
[66]
accuracy of Computer-Aided Detection of Google Scholar
[67]
tuberculosis on Chest radiography among private Google Scholar
[68]
sector patients in Pakistan,” Scientific reports, Google Scholar
[69]
vol. 8, no. 1, p. 12339, Aug. 2018, doi: 10.1038/ Google Scholar
[70]
s41598-018-30810-1. Google Scholar
[71]
Melendez J, “Accuracy of an automated system Google Scholar
[72]
for tuberculosis detection on chest radiographs in Google Scholar
[73]
high-risk screening,” Int J Tuberc Lung Dis, vol.22, no. 5, pp. 567–571, May 2018, doi: 10.5588/ Google Scholar
[74]
ijtld.17.0492. Google Scholar
[75]
Rahman MT, “An evaluation of automated Google Scholar
[76]
chest radiography reading software for Google Scholar
[77]
tuberculosis screening among public- and Google Scholar
[78]
private-sector patients,” The European Google Scholar
[79]
respiratory journal, vol. 49, no. 5, May 2017, doi: Google Scholar
[80]
1183/13993003.02159-2016. Google Scholar
[81]
Philipsen RH, “Automated chest-radiography as Google Scholar
[82]
a triage for Xpert testing in resource-constrained Google Scholar
[83]
settings: a prospective study of diagnostic Google Scholar
[84]
accuracy and costs,” Scientific reports, vol. 5, p. Google Scholar
[85]
, Jul. 2015, doi: 10.1038/srep12215. Google Scholar
[86]
Muyoyeta M, “The sensitivity and specificity Google Scholar
[87]
of using a computer aided diagnosis program Google Scholar
[88]
for automatically scoring chest X-rays of Google Scholar
[89]
presumptive TB patients compared with Xpert Google Scholar
[90]
MTB/RIF in Lusaka Zambia,” PLoS One, vol. Google Scholar
[91]
, no. 4, p. e93757, 2014, doi: 10.1371/journal. Google Scholar
[92]
pone.0093757. Google Scholar
[93]
Breuninger M, “Diagnostic accuracy of computeraided detection of pulmonary tuberculosis in chest Google Scholar
[94]
radiographs: a validation study from sub-Saharan Google Scholar
[95]
Africa,” PLoS One, vol. 9, no. 9, p. e106381, Google Scholar
[96]
, doi: 10.1371/journal.pone.0106381. Google Scholar