Abstract:

Active case finding remains a cornerstone of global tuberculosis (TB) control, and chest radiography (CXR) is the primary imaging tool recommended for screening asymptomatic populations. In China, a dual-track case-finding system combining passive case finding among symptomatic individuals and communitybased active screening has been established. Building on this framework, expert consensus recommends the deployment of chest imaging combined with artificial intelligence-based computer-aided detection (AI-CAD) at primary care facilities, targeting highrisk groups to improve case-finding efficiency. However, existing studies have largely focused on validating the diagnostic performance of AI models, while critical gaps remain in the development of stratified screening standards tailored to China’s lowtomoderate TB burden setting, standardized endtoend workflows, systematic cost-effectiveness evaluations, and comprehensive quality assurance systems. Based on current World Health Organization (WHO) recommendations and national TB control guidelines, this commentary first clarifies the policy role of CXR in TB screening and proposes a stratified framework defining appropriate populations for AIassisted screening and tiered requirements for imaging equipment deployment. We then focus on the “mobile digital radiography (DR)-based screening combined with AI” model in primary care settings and analyze five major implementation challenges: population applicability, workflow optimization, costeffectiveness, model generalizability, and quality assurance. To address these challenges, we propose actionable strategies across five dimensions: (1)establishing standardized multisource imaging databases; (2)developing stratified and stepwise screening strategies and algorithms; (3)constructing integrated, end-to-end quality assurance mechanisms; (4)exploring multidisease integrated screening approaches; (5)conducting localized, realworld cohort evaluations. These efforts aim to provide systematic evidencebased support for establishing a closedloop TB prevention and control system integrating communitybased initial screening, AIdriven risk stratification, etiological confirmation, and longitudinal followup management.

Key words: Tuberculosis, pulmonary, Tomography, X-ray computed, Artificial intelligence, Active screening, Costs and cost analysis