Chest X-ray Analysis With Deep Learning-Based Software as a Triage Test for Pulmonary Tuberculosis: An Individual Patient Data Meta-Analysis of Diagnostic Accuracy

Abstract

Background Automated radiologic analysis using computer-aided detection software (CAD) could facilitate chest X-ray (CXR) use in tuberculosis diagnosis. There is little to no evidence on the accuracy of commercially-available deep learning-based CAD in different populations, including patients with smear-negative tuberculosis and people living with HIV (PLWH). Methods We collected CXRs and individual patient data (IPD) from studies evaluating CAD in patients self-referring for tuberculosis symptoms with culture or nucleic acid amplification testing as the reference. We re-analyzed CXRs with three CAD (CAD4TB version (v) 6, Lunit v3.1.0.0, and qXR v2). We estimated sensitivity and specificity within each study and pooled using IPD meta-analysis. We used multivariable meta-regression to identify characteristics modifying accuracy. Results We included CXRs and IPD of 3727/3967 participants from 4/7 eligible studies. 17% (621/3727) were PLWH. 17% (645/3727) had microbiologically-confirmed tuberculosis. Despite using the same threshold score for classifying CXR in every study, sensitivity and specificity varied from study to study. The software had similar unadjusted accuracy (at 90% pooled sensitivity, pooled specificities were: CAD4TBv6, 56.9% [95%CI:51.7-61.9]; Lunit, 54.1% [44.6-63.3]; qXRv2, 60.5% [51.7-68.6]). Adjusted absolute differences in pooled sensitivity between PLWH and HIV-uninfected participants was: CAD4TBv6, -13.4% [-21.1, -6.9]; Lunit, +2.2% [-3.6, +6.3]; qXRv2: -13.4% [-21.5, -6.6]); between smear-negative and smear-positive tuberculosis was: CAD4TBv6, -12.3% [-19.5, -6.1]; Lunit, -17.2% [-24.6, -10.5]; qXRv2, -16.6% [-24.4, -9.9]. Accuracy was similar to human readers. Conclusions For CAD CXR analysis to be implemented as a high-sensitivity tuberculosis rule-out test, users will need threshold scores identified from their own patient populations, and stratified by HIV- and smear-status.

Clinical predictors of pulmonary tuberculosis among South African adults with HIV

“…However, it might be possible to train a more discriminatory model if a larger number of subclinical TB cases were available. 16 , 50 Chest radiography with computer-aided detection has reasonable performance as a rule-out test for symptomatic individuals with presumptive TB 9 , 10 and presents a feasible and affordable tool which is already purposed for active case-finding, however sensitivity for subclinical TB has yet to be ascertained.…”

“…Chest radiography has been used successfully in prevalence surveys as a screening tool to guide confirmatory sputum microbiological testing and for diagnosis of subclinical TB, 3 , 4 , 5 , 6 and computer-aided detection offers promise for affordable mass screening. 9 , 10 Rapid point-of-care blood-based tools, such as C-reactive protein (CRP) 11 and host-blood transcriptomics, 12 are also currently being investigated.…”

Evaluation of chest X-Ray with automated interpretation algorithms for mass tuberculosis screening in prisons

“…Recent studies have evaluated x-ray interpretation algorithms among individuals presenting to clinics with TB symptoms, finding variable results. An individual participant meta-analysis found that none of the systems investigated met the WHO TPP criteria for triage, with specificities ranging from 54-61% at 90% sensitivity (14). By contrast, a study in Bangladesh found that the qXR system achieved 74% specificity at the same threshold, and that all algorithms outperformed interpretation by radiologists (15).…”

Child Contact Case Management—A Major Policy-Practice Gap in High-Burden Countries

“…While current CXR technologies and assays for TB infection have considerable limitations when it comes to implementation in resource limited settings, innovative solutions such as digital and portable CXR, used in combination with computer aided CXR interpretation (CAD) have the potential to improve access to CXR [21][22][23][24][25]. CAD has been recommended by the WHO as an alternative to human reader interpretation of CXR screening for TB in people aged 15 years and above [18].…”