A Comprehensive Systematic Review of AI-based Tuberculosis Detection using Imaging and Clinical Data

Despite being a major cause of death in developing countries, tuberculosis (TB) continues to rank among the leading infectious diseases, claiming over 1.3 million lives yearly. Traditional diagnostics have been constrained by limitations due to the lack of radiological expertise and advanced technology in developing nations. Recent studies suggest that the utilization of artificial intelligence (AI) technologies such as deep learning (DL) and hybrid intelligent systems offers a revolutionary solution to the problem of TB detection based on chest X-rays, CT scans, and patient information. This study provides a comprehensive review of recent publications on AI-based TB diagnosis through systematic PRISMA methodology. In total, fifteen relevant papers from the Scopus database during the period of January 2024 to March 2025 were selected and critiqued. These included various machine learning algorithms such as DL models, transformers, explainable AI (XAI) methodologies, federated learning paradigms, and multimodal source data processing. Accuracy rates of 95.7%-99.4% were reported for DL model-based TB detection on chest images. XAI applications were found to improve clinicians’ diagnoses with significant effect. Critical knowledge gaps identified include almost zero use of fuzzy logic, inadequate multimodal source data analysis, poor cross-population testing, and insufficient assessment of practical deployment.