Student or educator? Sign up with your school email and get 6 months of Pro free.
CoChat

The DOI matches its Crossref record; 49 of the 49 references it cites were found in a registry or an index.

Computer-aided interpretation of chest radiography reveals the spectrum of tuberculosis in rural South Africa

Open access
Fehr, Jana, Konigorski, Stefan, Olivier, Stephen, Gunda, Resign, Surujdeen, Ashmika, Gareta, Dickman, Smit, Theresa, Baisley, Kathy, Moodley, Sashen, Moosa, Yumna, Hanekom, Willem, Koole, Olivier, Ndung’u, Thumbi, Pillay, Deenan, Grant, Alison D., Siedner, Mark J., Lippert, Christoph, Wong, Emily B., the Vukuzazi Team, Ramnanan, Anand, Mkhwanazi, Anele, Rapulana, Antony, Singh, Anupa, Govender, Ashentha, Zungu, Ayanda, Mfolo, Boitsholo, Magwaza, Bongani, Ndlovu, Bongumenzi, Mavimbela, Clive, Criticos, Costa, Munatsi, Day, Kalyan, Dilip, Mlambo, Doctar, Mfeka, Fezeka, Mabetlela, Freddy, Ording-Jespersen, Gregory, Keal, Hannah, Dlamini, Hlengiwe, Khathi, Hlengiwe, Chonco, Hlobisile, Gumede, Hlobisile, Khumalo, Hlolisile, Ngubane, Hloniphile, Shen, Hollis, Kambonde, Hosea, Mpofana, Innocentia, Kwinda, Jabu, Dreyer, Jaco, Cousins, Jade, Kalideen, Jaikrishna, Seeley, Janet, Chetty, Kandaseelan, Brien, Kayleen, Nyamande, Kennedy, Moropane, Kgaugelo, Malomane, Khabonina, Khan, Khadija, Buthelezi, Khanyisani, Perumal, Kimeshree, Herbst, Kobus, Mthembu, Lindani, Pillay, Logan, Dlamini, Mandisi, Zikhali, Mandlakayise, Mbuyisa, Mbali, Mofokeng, Mbuti, Sibiya, Melusi, Dube, Mlungisi, Suleman, Mosa, Steto, Mpumelelo, Buthelezi, Mzamo, Padayachi, Nagavelli, Gqaleni, Nceba, Mhlongo, Ngcebo, Ntshakala, Nokukhanya, Majozi, Nomathamsanqa, Zondi, Nombuyiselo, Luthuli, Nomfundo, Ngema, Nomfundo, Buthelezi, Nompilo, Mfeka, Nonceba, Khuluse, Nondumiso, Mabaso, Nondumiso, Zitha, Nondumiso, Mfekayi, Nonhlanhla, Mzimela, Nonhlanhla, Mbonambi, Nozipho, Mkhwanazi, Ntombiyenhlanhla, Ntombela, Ntombiyenkosi, Ramkalawon, Pamela, Tshivase, Pfarelo, Mkhwanazi, Phakamani, Mathews, Philippa, Mthethwa, Phumelele, Ngcobo, Phumla, Jackpersad, Ramesh, Zondo, Raynold, Singh, Rochelle, Myeni, Rose, Bucibo, Sanah, Mthembu, Sandile, Harilall, Sashin, Makhari, Senamile, Mchunu, Seneme, Mkhwanazi, Senzeni, Gumbi, Sibahle, Nene, Siboniso, Mhlongo, Sibusiso, Mkhwanazi, Sibusiso, Nsibande, Sibusiso, Ntshangase, Simphiwe, Dlamini, Siphephelo, Ngcobo, Sithembile, Nsibande, Siyabonga, Nxumalo, Siyabonga, Ndlela, Sizwe, Mthombeni, Skhumbuzo, Zulu, Smangaliso, Mthembu, Sphiwe Clement, Ntuli, Sphiwe, Ntimbane, Talente, Zondi, Thabile, Khoza, Thandeka, Nkosi, Thengokwakhe, Bhengu, Thokozani, Simelane, Thokozani, Modise, Tshwaraganang, Madolo, Tumi, Vellem, Velile, Mthembu, Welcome Petros, Mkhize, Xolani, Mbatha, Zamashandu, Buthelezi, Zinhle, Mthembu, Zinhle, Sikhosana, Zizile · 2021 · npj Digital Medicine

Abstract

Computer-aided digital chest radiograph interpretation (CAD) can facilitate high-throughput screening for tuberculosis (TB), but its use in population-based active case-finding programs has been limited. In an HIV-endemic area in rural South Africa, we used a CAD algorithm (CAD4TBv5) to interpret digital chest x-rays (CXR) as part of a mobile health screening effort. Participants with TB symptoms or CAD4TBv5 score above the triaging threshold were referred for microbiological sputum assessment. During an initial pilot phase, a low CAD4TBv5 triaging threshold of 25 was selected to maximize TB case finding. We report the performance of CAD4TBv5 in screening 9,914 participants, 99 (1.0%) of whom were found to have microbiologically proven TB. CAD4TBv5 was able to identify TB cases at the same sensitivity but lower specificity as a blinded radiologist, whereas the next generation of the algorithm (CAD4TBv6) achieved comparable sensitivity and specificity to the radiologist. The CXRs of people with microbiologically confirmed TB spanned a range of lung field abnormality, including 19 (19.2%) cases deemed normal by the radiologist. HIV serostatus did not impact CAD4TB’s performance. Notably, 78.8% of the TB cases identified during this population-based survey were asymptomatic and therefore triaged for sputum collection on the basis of CAD4TBv5 score alone. While CAD4TBv6 has the potential to replace radiologists for triaging CXRs in TB prevalence surveys, population-specific piloting is necessary to set the appropriate triaging thresholds. Further work on image analysis strategies is needed to identify radiologically subtle active TB.

Overview

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1038/s41746-021-00471-y

Cites

49 of 49 references verified0 citation errors · 0 not checkable

Statistics

Figures

Fig. 1 TB screening in the first year of Vukuzazi. A total of 10,320 participants were enrolled, of which 9914 underwent chest radiography. Participants were triaged for sputum collection if they had symptoms or if their CAD4TBv5 score was equal to or above the triaging threshold. For triaged participants who could spontaneously produce a sputum sample (n = 4976), the sputum was tested with Xpert MTB/RIF Ultra and liquid culture. Independently, the radiologist indicated whether chest radiographs were normal or abnormal and if abnormal were diagnostic of active TB. If the radiologist indicated abnormal lung fields among participants who were not triaged in the camp, the sputum was collected during a home visit. Participants were grouped into definite (microbiologically confirmed), probable (radiologically diagnosed), and no evidence of TB. A more stringent subgroup of definite TB excluded participants whose only microbiological evidence of TB was an Xpert Ultra “trace” result.
Fig. 1 TB screening in the first year of Vukuzazi. A total of 10,320 participants were enrolled, of which 9914 underwent chest radiography. Participants were triaged for sputum collection if they had symptoms or if their CAD4TBv5 score was equal to or above the triaging threshold. For triaged participants who could spontaneously produce a sputum sample (n = 4976), the sputum was tested with Xpert MTB/RIF Ultra and liquid culture. Independently, the radiologist indicated whether chest radiographs were normal or abnormal and if abnormal were diagnostic of active TB. If the radiologist indicated abnormal lung fields among participants who were not triaged in the camp, the sputum was collected during a home visit. Participants were grouped into definite (microbiologically confirmed), probable (radiologically diagnosed), and no evidence of TB. A more stringent subgroup of definite TB excluded participants whose only microbiological evidence of TB was an Xpert Ultra “trace” result.CC BY · PubMed Central
Fig. 2 Radiographic spectrum of two participants with microbiologically confirmed active TB. Digital CXR images and the corresponding CAD4TBv5 heatmaps (blue colors mark normal lung fields and red marks TB-related abnormalities contributing to CAD4TB score). Participant 1 (a, b) was 62 years old, HIV-positive on ART, and had a CAD4TB v5 score of 95 which the radiologist assessed as abnormal and diagnostic of TB. Participant 2 (c, d) was 31 years old, HIV-negative, and had a CAD4TB v5 score of 39, which the radiologist assessed as normal. Both participants had positive liquid culture and XpertUltra (greater than trace) results, no TB symptoms, and no history of previous TB treatment.
Fig. 2 Radiographic spectrum of two participants with microbiologically confirmed active TB. Digital CXR images and the corresponding CAD4TBv5 heatmaps (blue colors mark normal lung fields and red marks TB-related abnormalities contributing to CAD4TB score). Participant 1 (a, b) was 62 years old, HIV-positive on ART, and had a CAD4TB v5 score of 95 which the radiologist assessed as abnormal and diagnostic of TB. Participant 2 (c, d) was 31 years old, HIV-negative, and had a CAD4TB v5 score of 39, which the radiologist assessed as normal. Both participants had positive liquid culture and XpertUltra (greater than trace) results, no TB symptoms, and no history of previous TB treatment.CC BY · PubMed Central
Fig. 3 CAD4TB scores differ by TB but not HIV status. CAD4TB v5 (a) and v6 (b) distributions across TB diagnostic groups: definite TB (microbiologically proven n = 99), definite TB, trace excluded (a more stringent group that excludes those whose only microbiological evidence of TB is an Xpert “trace” results, n = 75), probable TB (radiologically diagnosed, n = 172), and no evidence of TB (n = 9643). Scores in each group were stratified by HIV status (definite TB HIV-positive: n = 41, HIV-negative: n = 57; definite TB, trace excluded HIV-positive: n = 34, HIV-negative: n = 40; probable TB HIV-positive: n = 76, HIV-negative: n = 96; no TB HIV-positive: n = 2837, HIV-negative: 6755). Horizontal lines mark the median. P-values from comparisons between HIV-positive vs. HIV-negative in each diagnostic group were for definite TB: 0.26 (v5), 0.26 (v6), definite TB, trace excluded: 0.70 (v5), 0.95 (v6), probable TB: 0.88 (v5), 0.06 (v6) no TB 2.9x10-5 (v5), 0.70 (v6).
Fig. 3 CAD4TB scores differ by TB but not HIV status. CAD4TB v5 (a) and v6 (b) distributions across TB diagnostic groups: definite TB (microbiologically proven n = 99), definite TB, trace excluded (a more stringent group that excludes those whose only microbiological evidence of TB is an Xpert “trace” results, n = 75), probable TB (radiologically diagnosed, n = 172), and no evidence of TB (n = 9643). Scores in each group were stratified by HIV status (definite TB HIV-positive: n = 41, HIV-negative: n = 57; definite TB, trace excluded HIV-positive: n = 34, HIV-negative: n = 40; probable TB HIV-positive: n = 76, HIV-negative: n = 96; no TB HIV-positive: n = 2837, HIV-negative: 6755). Horizontal lines mark the median. P-values from comparisons between HIV-positive vs. HIV-negative in each diagnostic group were for definite TB: 0.26 (v5), 0.26 (v6), definite TB, trace excluded: 0.70 (v5), 0.95 (v6), probable TB: 0.88 (v5), 0.06 (v6) no TB 2.9x10-5 (v5), 0.70 (v6).CC BY · PubMed Central
Fig. 4 Performance of CAD4TB v5 and v6. Receiver-operating curve of CAD4TBv5 (a) and v6 (b) triaging thresholds for each TB diagnostic group: definite TB (microbiologically confirmed, positive: n = 99, negative: n = 4877), definite TB, trace excluded (a more stringent group that excludes those whose only microbiological evidence of TB is an Xpert “trace” results, positive: n = 75, negative: 4901), and probable TB (radiologically diagnosed, positive: n = 172, negative: n = 9742). The radiologist’s sensitivity and specificity of detecting definite TB (+) and definite TB trace excluded (x) is marked. The (c) percentage of participants who would have required sputum testing at each CAD4TB triaging threshold (n = 9914) and (d) number of definite TB cases that would have been missed at each CAD4TB triaging threshold (n = 99). In (c) and (d) the horizontal dashed line indicates the radiologist’s performance and vertical lines indicate the CAD4TBv5 triaging thresholds used in the pilot (60) and main study phase (25).
Fig. 4 Performance of CAD4TB v5 and v6. Receiver-operating curve of CAD4TBv5 (a) and v6 (b) triaging thresholds for each TB diagnostic group: definite TB (microbiologically confirmed, positive: n = 99, negative: n = 4877), definite TB, trace excluded (a more stringent group that excludes those whose only microbiological evidence of TB is an Xpert “trace” results, positive: n = 75, negative: 4901), and probable TB (radiologically diagnosed, positive: n = 172, negative: n = 9742). The radiologist’s sensitivity and specificity of detecting definite TB (+) and definite TB trace excluded (x) is marked. The (c) percentage of participants who would have required sputum testing at each CAD4TB triaging threshold (n = 9914) and (d) number of definite TB cases that would have been missed at each CAD4TB triaging threshold (n = 99). In (c) and (d) the horizontal dashed line indicates the radiologist’s performance and vertical lines indicate the CAD4TBv5 triaging thresholds used in the pilot (60) and main study phase (25).CC BY · PubMed Central

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Show all 48 references

Cited by

See all 82

Metadata

Authors
Fehr, J., Konigorski, S., Olivier, S., Gunda, R., Surujdeen, A., Gareta, D., Smit, T., Baisley, K., Moodley, S., Moosa, Y., Hanekom, W., Koole, O., Ndung’u, T., Pillay, D., Grant, A. D., Siedner, M. J., Lippert, C., Wong, E. B., the Vukuzazi Team, Ramnanan, A., Mkhwanazi, A., Rapulana, A., Singh, A., Govender, A., Zungu, A., Mfolo, B., Magwaza, B., Ndlovu, B., Mavimbela, C., Criticos, C., Munatsi, D., Kalyan, D., Mlambo, D., Mfeka, F., Mabetlela, F., Ording-Jespersen, G., Keal, H., Dlamini, H., Khathi, H., Chonco, H., Gumede, H., Khumalo, H., Ngubane, H., Shen, H., Kambonde, H., Mpofana, I., Kwinda, J., Dreyer, J., Cousins, J., Kalideen, J., Seeley, J., Chetty, K., Brien, K., Nyamande, K., Moropane, K., Malomane, K., Khan, K., Buthelezi, K., Perumal, K., Herbst, K., Mthembu, L., Pillay, L., Dlamini, M., Zikhali, M., Mbuyisa, M., Mofokeng, M., Sibiya, M., Dube, M., Suleman, M., Steto, M., Buthelezi, M., Padayachi, N., Gqaleni, N., Mhlongo, N., Ntshakala, N., Majozi, N., Zondi, N., Luthuli, N., Ngema, N., Buthelezi, N., Mfeka, N., Khuluse, N., Mabaso, N., Zitha, N., Mfekayi, N., Mzimela, N., Mbonambi, N., Mkhwanazi, N., Ntombela, N., Ramkalawon, P., Tshivase, P., Mkhwanazi, P., Mathews, P., Mthethwa, P., Ngcobo, P., Jackpersad, R., Zondo, R., Singh, R., Myeni, R., Bucibo, S., Mthembu, S., Harilall, S., Makhari, S., Mchunu, S., Mkhwanazi, S., Gumbi, S., Nene, S., Mhlongo, S., Mkhwanazi, S., Nsibande, S., Ntshangase, S., Dlamini, S., Ngcobo, S., Nsibande, S., Nxumalo, S., Ndlela, S., Mthombeni, S., Zulu, S., Mthembu, S. C., Ntuli, S., Ntimbane, T., Zondi, T., Khoza, T., Nkosi, T., Bhengu, T., Simelane, T., Modise, T., Madolo, T., Vellem, V., Mthembu, W. P., Mkhize, X., Mbatha, Z., Buthelezi, Z., Mthembu, Z., & Sikhosana, Z.
Affiliations
  • Hasso Plattner Institute (DE)
  • Africa Health Research Institute (ZA)
  • London School of Hygiene & Tropical Medicine (GB)
  • Harvard University (US)
  • University of KwaZulu-Natal (ZA)
Corresponding author
Wong, Emily B.
Venue
npj Digital Medicine
Published
2021-07-02
Volume
4
Issue
1
Pages
106-106
ISSN
2398-6352 (ISSN-L)
OpenAlex
W3173691875
Cited by
74
References
48
Citations by year
2026: 17 · 2025: 11 · 2024: 17 · 2023: 18 · 2022: 7 · 2021: 4
FWCI
3.64
Type
journal article
Language
en
Open access
Licence
Locations
  • npj Digital Medicine (publishedVersion, cc-by)
  • medRxiv (acceptedVersion, cc-by-nd)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • LSHTM Research Online (London School of Hygiene and Tropical Medicine) (acceptedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
  • UCL Discovery (University College London) (submittedVersion)
Funders
  • Bill and Melinda Gates Foundation (OPP1175182)
  • Wellcome Trust (201433/Z/16/Z, DEL-15-006, 107752/Z/15/Z, 107752/Z/15/, 201433, /Z/15/Z, 107752)
  • Government of the United Kingdom
  • New Partnership for Africa's Development (DEL-15-006, 107752/Z/15/Z)
  • African Academy of Sciences (DEL-15- 006, 107752/Z/15/Z)
  • National Institutes of Health (75N93019C00070, DEL-15-006)
  • Medical Research Council (MR/R010161/1)
  • Alliance for Accelerating Excellence in Science in Africa (DEL-15-006)
  • National Institute of Allergy and Infectious Diseases (75N93019C00070, 107752/Z/15/Z, K08AI118538)
  • Fogarty International Center (R21TW011687)
Topics
  • Health Sciences › Medicine › Infectious Diseases › Tuberculosis Research and Epidemiology
  • Health Sciences › Medicine › Radiology, Nuclear Medicine and Imaging › COVID-19 diagnosis using AI
  • Physical Sciences › Computer Science › Artificial Intelligence › AI in cancer detection
Keywords
tuberculosis, computer-aided diagnosis, chest radiography, active case finding, mobile health screening, HIV, asymptomatic tuberculosis, CAD4TB, rural South Africa

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: Computer-aided interpretation of chest radiography reveals the spectrum of tuberculosis in rural South Africa. https://app.cochat.ai/papers/10.1038/s41746-021-00471-y
The paper
Fehr, J., Konigorski, S., Olivier, S., Gunda, R., Surujdeen, A., Gareta, D., Smit, T., Baisley, K., Moodley, S., Moosa, Y., Hanekom, W., Koole, O., Ndung’u, T., Pillay, D., Grant, A. D., Siedner, M. J., Lippert, C., Wong, E. B., the Vukuzazi Team, . . . Sikhosana, Z. (2021). Computer-aided interpretation of chest radiography reveals the spectrum of tuberculosis in rural South Africa. npj Digital Medicine. https://doi.org/10.1038/s41746-021-00471-y

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.