Large-Field Single-Cell Morphodynamic Scattering Imaging Enables Rapid, Heterogeneity-Resolved Phenotypic Antimicrobial Susceptibility Testing

Abstract Timely antimicrobial susceptibility testing (AST) is essential for the management of bloodstream infections, yet conventional phenotypic AST relies primarily on bulk bacterial growth, requiring prolonged incubation and masking clinically relevant population heterogeneity. Here, we present a morphology-based AST approach that quantifies time-resolved responses of individual bacteria as an alternative phenotypic measure of antimicrobial susceptibility. Large-field, label-free scattering imaging enables parallel monitoring of thousands of bacteria, while deep learning-assisted segmentation provides quantitative extraction of integrated scattering intensity and image-derived geometric features during β-lactam antibiotic exposure. This high-throughput single-cell analysis preserves population heterogeneity and supports measurements across clinically relevant samples. Distinct antibiotic-dependent morphodynamic trajectories emerged within 60 min, enabling rapid susceptibility determination while retaining information beyond binary growth inhibition. These single-cell response profiles further enabled preliminary minimum inhibitory concentration determination, evaluation of responses to antibiotic combinations, and quantification of resistant subpopulations in a defined susceptible-resistant mixture. Direct analysis of ten clinical positive blood culture samples tested against five antibiotics achieved 96.7% categorical agreement with the clinical reference method, with an estimated post-positivity workflow time of approximately 95 min. These findings establish time-resolved single-cell morphology as an informative phenotypic measure for rapid AST that preserves drug-dependent response patterns and bacterial population heterogeneity.