Highly sensitive quantitative phase microscopy and deep learning complement whole genome sequencing for rapid detection of infection and antimicrobial resistance

The current state-of-the-art infection and antimicrobial resistance diagnostics (AMR) is based mainly on culture-based methods with a detection time of 48-96 hours. Slow diagnoses lead to adverse patient outcomes that directly correlate with the time taken to administer optimal antimicrobials. Mortality risk doubles with a 24-hour delay in providing appropriate antibiotics in cases of bacteremia. Therefore, it is essential to develop novel methods that can promptly and accurately diagnose microbial infections at both species and strain levels in clinical settings. Here, we demonstrate that the complimentary use of label-free optical assay with whole-genome sequencing (WGS) can enable high-speed culture-free diagnosis of infection and AMR. Our assay is based on microscopy methods exploiting label-free, highly sensitive quantitative phase microscopy (QPM) followed by deep convolutional neural networks (DCNNs) based classification. We benchmarked our proposed workflow on 21 clinical isolates from four WHO priority pathogens (Escherichia coli, Staphylococcus aureus, Klebsiella pneumoniae, and Acinetobacter baumannii) that were antibiotic susceptibility testing (AST) phenotyped, and their antimicrobial resistance (AMR) profile was determined by WGS. The proposed optical assay was in good agreement with the WGS characterization. Highly accurate classification based on the gram staining (100% for gram-negative and 83.4% for gram-positive), species (98.6%), and resistant/susceptible type (96.4%), as well as at the individual strain level (100% accurate in predicting 19 out of the 21 strains). These results demonstrate the potential of the QPM assay as a rapid and first-stage tool for species, presence, and absence of AMR, and strain-level classification, which WGS can follow up for confirmation of the pathogen ID and the characterization of the AMR profile and susceptibility antibiotic. Taken together, all this information is of high clinical importance. Such a workflow could potentially facilitate efficient antimicrobial stewardship and prevent the spread of AMR.