Advances in pathogen detection that incorporate artificial intelligence (AI) may capture microbial signals under challenging environmental conditions that traditional methods miss. This systematic review evaluates the application, performance, and methodological characteristics of AI-enabled imaging for pathogen detection, including its impact on speed, accuracy, and modeling under stress conditions. Studies were systematically identified from five electronic databases using search terms related to AI, pathogen, detection, and imaging. Inclusion criteria, defined using the Population, Intervention, Comparators, Outcome, Study design (PICOS) framework, focused on microscopy-based pathogen detection enhanced by AI. Data extraction followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines and captured biological sample preparation, imaging modalities, AI-enabled data analyses, comparator methods, and performance metrics. Of 2736 citations retrieved, 120 were reviewed in full and 28 studies met the inclusion criteria. These represented more than 40 pathogens, most commonly Salmonella spp. and Escherichia coli. Only three studies explicitly evaluated signals from stress or inactivated states. Comparator methods (e.g., culture-based or molecular assays) were infrequently reported, limiting benchmarking against established workflows. Reporting inconsistencies in laboratory protocols and computational pipeline further complicated reproducibility and precluded meta-analysis. Overall, this review offers a comprehensive overview of current AI-enabled imaging approaches from both biological and computational perspectives and highlights the need for standardized benchmarks and reporting practices to support reproducible, transferable pathogen detection.
Papa et al. (Wed,) studied this question.