Dataset Information

Evaluation of chest X-ray with automated interpretation algorithms for mass tuberculosis screening in prisons: a cross-sectional study.

ABSTRACT:

Background

The World Health Organization (WHO) recommends systematic tuberculosis (TB) screening in prisons. Evidence is lacking for accurate and scalable screening approaches in this setting. We aimed to assess the accuracy of artificial intelligence-based chest x-ray interpretation algorithms for TB screening in prisons.

Methods

We performed prospective TB screening in three male prisons in Brazil from October 2017 to December 2019. We administered a standardized questionnaire, performed a chest x-ray in a mobile unit, and collected sputum for confirmatory testing using Xpert MTB/RIF and culture. We evaluated x-ray images using three algorithms (CAD4TB version 6, Lunit version 3.1.0.0 and qXR version 3) and compared their accuracy. We utilized multivariable logistic regression to assess the effect of demographic and clinical characteristics on algorithm accuracy. Finally, we investigated the relationship between abnormality scores and Xpert semi-quantitative results.

Findings

Among 2075 incarcerated individuals, 259 (12.5%) had confirmed TB. All three algorithms performed similarly overall with area under the receiver operating characteristic curve (AUC) of 0.88-0.91. At 90% sensitivity, only LunitTB and qXR met the WHO Target Product Profile requirements for a triage test, with specificity of 84% and 74%, respectively. All algorithms had variable performance by age, prior TB, smoking, and presence of TB symptoms. LunitTB was the most robust to this heterogeneity but nonetheless failed to meet the TPP for individuals with previous TB. Abnormality scores of all three algorithms were significantly correlated with sputum bacillary load.

Interpretation

Automated x-ray interpretation algorithms can be an effective triage tool for TB screening in prisons. However, their specificity is insufficient in individuals with previous TB.

Funding

This study was supported by the US National Institutes of Health (grant numbers R01 AI130058 and R01 AI149620) and the State Secretary of Health of Mato Grosso do Sul.

SUBMITTER: Soares TR

PROVIDER: S-EPMC9904090 | biostudies-literature | 2023 Jan

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Evaluation of chest X-ray with automated interpretation algorithms for mass tuberculosis screening in prisons: a cross-sectional study.

Soares Thiego Ramon TR Oliveira Roberto Dias de RD Liu Yiran E YE Santos Andrea da Silva ADS Santos Paulo Cesar Pereira Dos PCPD Monte Luma Ravena Soares LRS Oliveira Lissandra Maia de LM Park Chang Min CM Hwang Eui Jin EJ Andrews Jason R JR Croda Julio J

Lancet regional health. Americas 20221104

<h4>Background</h4>The World Health Organization (WHO) recommends systematic tuberculosis (TB) screening in prisons. Evidence is lacking for accurate and scalable screening approaches in this setting. We aimed to assess the accuracy of artificial intelligence-based chest x-ray interpretation algorithms for TB screening in prisons.<h4>Methods</h4>We performed prospective TB screening in three male prisons in Brazil from October 2017 to December 2019. We administered a standardized questionnaire, ...[more]

PMID: 36776567

Similar Datasets

Project description:BackgroundFamily medicine physicians may encounter a wide variety of conditions, including acute and urgent cases. Considering the limited access to diagnostic investigations in primary care practice, chest X-ray remains the imaging modality of choice. The current study assessed the competency of family medicine residents in the interpretation of chest X-rays for emergency conditions and to compare it with that of diagnostic radiology residents, general practitioners, and medical interns.MethodsAn online survey was distributed to 600 physicians, including family medicine residents, medical interns, general practitioners, and diagnostic radiology residents. The study included some background information such as gender, years in practice, training type, interest in pulmonary medicine and diagnostic radiology, and having adequate training on the interpretation of chest X-rays. The survey had 10 chest X-ray cases with brief clinical information. Participants were asked to choose the most likely diagnosis and to rate their degree of confidence in the interpretation of the chest X-ray for each case.ResultsThe survey was completed by 205 physicians (response rate = 34.2%). The overall diagnostic accuracy was 63.1% with a significant difference between family medicine and radiology residents (58.0% vs. 90.5%; P < 0.001). The COVID-19 pneumonia (85.4%) and pneumoperitoneum (80.5%) cases had the highest diagnostic accuracy scores. There was a significant correlation between the diagnostic confidence and accuracy (rs = 0.39; P < 0.001). Multivariable regression analysis revealed that being diagnostic radiology residents (odds ratio [OR]: 13.0; 95% confidence interval [CI]: 2.5-67.7) and having higher diagnostic confidence (OR: 2.2; 95% CI: 1.3-3.8) were the only independent predictors of achieving high diagnostic accuracy.ConclusionThe competency of family medicine residents in the interpretation of chest X-ray for emergency conditions was far from optimal. The introduction of radiology training courses on emergency conditions seems imperative. Alternatively, the use of tele-radiology in primary healthcare centers should be considered.

Project description:BackgroundCommunity active case finding (ACF) for tuberculosis was widely implemented in Europe and North America between 1940 and 1970, when incidence was comparable to many present-day high-burden countries. Using an interrupted time series analysis, we analysed the effect of the 1957 Glasgow mass chest X-ray campaign to inform contemporary approaches to screening.Methods and findingsCase notifications for 1950 to 1963 were extracted from public health records and linked to demographic data. We fitted Bayesian multilevel regression models to estimate annual relative case notification rates (CNRs) during and after a mass screening intervention implemented over 5 weeks in 1957 compared to the counterfactual scenario where the intervention had not occurred. We additionally estimated case detection ratios and incidence. From 11 March 1957 to 12 April 1957, 714,915 people (622,349 of 819,301 [76.0%] resident adults ≥15 years) were screened with miniature chest X-ray; 2,369 (0.4%) were diagnosed with tuberculosis. Pre-intervention (1950 to 1956), pulmonary CNRs were declining at 2.3% per year from a CNR of 222/100,000 in 1950. With the intervention in 1957, there was a doubling in the pulmonary CNR (RR: 1.95, 95% uncertainty interval [UI] [1.81, 2.11]) and 35% decline in the year after (RR: 0.65, 95% UI [0.59, 0.71]). Post-intervention (1958 to 1963) annual rates of decline (5.4% per year) were greater (RR: 0.77, 95% UI [0.69, 0.85]), and there were an estimated 4,599 (95% UI [3,641, 5,683]) pulmonary case notifications averted due to the intervention. Effects were consistent across all city wards and notifications declined in young children (0 to 5 years) with the intervention. Limitations include the lack of data in historical reports on microbiological testing for tuberculosis, and uncertainty in contributory effects of other contemporaneous interventions including slum clearances, introduction of BCG vaccination programmes, and the ending of postwar food rationing.ConclusionsA single, rapid round of mass screening with chest X-ray (probably the largest ever conducted) likely resulted in a major and sustained reduction in tuberculosis case notifications. Synthesis of evidence from other historical tuberculosis screening programmes is needed to confirm findings from Glasgow and to provide insights into ongoing efforts to successfully implement ACF interventions in today's high tuberculosis burden countries and with new screening tools and technologies.

Dataset Information

Evaluation of chest X-ray with automated interpretation algorithms for mass tuberculosis screening in prisons: a cross-sectional study.

Background

Methods

Findings

Interpretation

Funding

Publications

Evaluation of chest X-ray with automated interpretation algorithms for mass tuberculosis screening in prisons: a cross-sectional study.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets