What are The Comparative Diagnostic Accuracies of X-Ray, CT, and Ultrasound in Terms of Sensitivity and Specificity for Detecting Pulmonary Tuberculosis in Adult Patients with Suspected Infection ? : A Systematic Review

Abstract

Introduction: Pulmonary tuberculosis (TB) poses a significant global health challenge, necessitating timely and accurate diagnosis. While chest X-ray (CXR) is widely used, its diagnostic accuracy varies. Advanced modalities like computed tomography (CT) and ultrasound offer potential improvements but have their own limitations regarding cost, accessibility, and operator dependency. This systematic review aims to compare the diagnostic accuracies (sensitivity and specificity) of X-ray, CT, and ultrasound for detecting pulmonary TB in adults with suspected infection. Methods: This systematic review followed PRISMA 2020 guidelines. Studies were included if they assessed X-ray, CT, or ultrasound in adult patients (≥18 years) with suspected pulmonary TB, reported sensitivity and/or specificity, and used microbiological confirmation as the reference standard. Data on study design, participant characteristics, imaging modalities, and diagnostic accuracy metrics were extracted. Searches were conducted on PubMed, Semantic Scholar, Sagepub, and Google Scholar. Results: Twenty-three studies were included. CT demonstrated the highest diagnostic accuracy, with sensitivity and specificity frequently exceeding 93% (e.g., sensitivity 93.8%-100%, specificity 94.2%-97.9% with deep learning). CXR showed high sensitivity (often >85%, up to 98.6%) but widely variable specificity (23%-99.9%). AI-enhanced CXR reported sensitivity between 85.3%-94.4% and specificity of 37.1%-69.2%. Ultrasound sensitivity for specific findings (e.g., subpleural nodules) ranged from 72.5%-100%, but specificity was generally lower and more variable (e.g., 66.7%). Diagnostic performance was influenced by HIV status and prior TB. Discussion: CT offers superior accuracy but is limited by cost and accessibility. CXR remains a primary screening tool due to accessibility, though its variable specificity, especially with AI, is a concern. Ultrasound shows promise as an adjunctive, radiation-free tool in resource-limited settings, but operator dependency and lack of standardization are challenges. The heterogeneity in study designs and risk of bias in many included studies temper the generalizability of findings. Conclusion: CT provides the highest diagnostic accuracy for pulmonary TB in adults. CXR is the most accessible screening modality, with AI potentially enhancing sensitivity but requiring specificity improvements. Ultrasound is a useful adjunct in specific contexts. Optimal diagnostic strategies should integrate these modalities based on clinical context, resource availability, and patient characteristics.