."Automated Interpretation of Blood Culture Gram Stains using a Deep Convolutional Neural Network" was published online today in the Journal of Clinical Microbiology. Congratulations to co-first authors: postdoctoral fellow, KP Smith, and medical microbiology fellow, Anthony Kang.
The article describes use of artificial intelligence in combination with a Metafer (MetaSystems) automated microscope to automatically interpret blood culture Gram stains without human intervention. Link to Abstract. Link to final accepted manuscript file.
2 Comments
Smith KP, Richmond DL, Brennan-Krohn T, Elliott HL, Kirby JE. Development of MAST: a Microscopy-Based Antimicrobial Susceptibility Testing Platform. SLAS Technology, In press.
Use of inkjet printing, advanced imaging, and machine learning to achieve reference standard, microdilution antimicrobial susceptibility testing readout in two hours with off-the-shelf supplies. For any antimicrobial at will. Update: Now Published On-Line in SLAS Technology Website!
.On acceptance of her manuscript by the Journal of Antimicrobial Chemotherapy:
Brennan-Krohn T, Truelson KA, Smith KP, Kirby JE. Screening for synergistic activity of antimicrobial combinations against carbapenem-resistant Enterobacteriaceae using inkjet printer-based technology. J Antimicrobial Chemotherapy. 2017 July. Link to abstract. Link to Journal Full Text. The manuscript highlights the remarkable activity of apramycin against multidrug-resistant, extensively drug resistant and pandrug-resistant Acinetobacter and Pseudomonas. Both organisms are significant multidrug-resistance threats. Importantly, frank resistance to apramycin was observed in < 2% of isolates. Apramycin is an aminocyclitol-based aminoglycoside that is currently approved for veterinary use. The activity in these groups of bacteria was especially notable in light of the high level of resistance of the same strain set to aminoglycosides (amikacin, gentamicin, tobramycin) approved for human use. This study complements our prior study, also published in DMID, demonstrating activity of apramycin against a high proportion of carbapenem-resistant Enterobacteriaceae strains.
Published online today in Infection and Immunity: "Promotion and Rescue of Intracellular Brucella neotomae Replication During Co-Infection With Legionella pneumophila." Yoon-Suk created a versatile bioreporter toolkit to enable analysis of the fate of individual pathogens in polymicrobial infections. He then validated a type IV secretion system-dependent Brucella model using the wood rat pathogen, Brucella neotomae. In contrast to wild type organisms, Brucella T4SS mutants were completely defective in their ability to growth inside of macrophages. Fascinatingly co-infection with Legionella pneumophila, another T4SS-dependent pathogen, was able to rescue intracellular growth of the T4SS-mutant Brucella and also stimulated growth of wild type Brucella organisms! This was a one way rescue: Legionella could rescue Brucella, but wild type Brucellla could not rescue T4SS-defective Legionella.
In the Journal of Clinical Microbiology, we describe an unusual Candida pseudo-outbreak that based on patterning on culture plates mimicked true infection. Specifically, colonies were not randomly distributed on plates, but demonstrated in a graded pattern typical of true infections with decreasing number of colonies on subsequent quandrants of streak plates. The culprit was eventually identified, as a contaminated Anoxomat chamber that was used to hold anaerobic plates prior to use in cultures. The Anoxomat rapidly purges and instills a hydrogen/nitrogen gas mixture into a closed jar in order to rapidly establish anaerobiosis. We believe that the Anoxomat overcame the natural swan neck barrier of Petri dishes allowing Candida guilliermondii containing aerosols to randomly contaminate anaerobic plates. C. guilliermondii does not grow anaerobically obscuring this contamination. However, randomly contaminated anaerobic plates were subsequently the source of contamination for the primary quandrants of aerobic media during specimen planting leading to simulation of true infection by microbiological criteria. Pseudo-outbreak strains were proven clonal by a combination of pulsed-field gel electropheresis and optical restriction mapping. The study represented a combined effort of the clinical microbiology laboratory and the infection control/hospital epidemiology divisions at BIDMC; Dan Diekema at the University of Iowa (PFGE analysis and expertise); and OpGen (optical mapping).
See: www.pnas.org/content/early/2017/01/10/1616248114.abstract. This is part of our ongoing collaboration with Ashlee Earl's group at the Broad Institute, Cambridge, MA
Co-First Authors, KP Smith and Thea Brennan-Krohn, from the Kirby Research Laboratory and Susan Weir from the BIDMC Clinical Microbiology Laboratory collaboratively investigated the ability of commonly used clinical methods to support new "susceptible dose-dependent." MIC breakpoints newly introduced by the Clinical Laboratoryand Standards Institute for the antibiotic cefepime. The idea behind these SDD breakpoints was to offer clinicians the chance to treat otherwise poorly susceptible pathogens by increasing the cefepime dose in a manner tied to the isolate's MIC. To perform the study, we enriched for strains that should have borderline cefepime susceptibility based on a ceftriaxone resistant phenotype. Surprisingly, three commercial methods (Vitek 2, disk diffusion, and a manual microscan panel) performed poorly with only 40-60% categorical agreement with the broth microdilution reference standard. In contrast, the Digital Dispensing Method, (DDM) previously described by our laboratory was stastically equivalent to the reference method, and therefore was the only method capable of supporting susceptible dose dependent therapeutic rescue. The manuscript, "Improved Accuracy of Cefepime Susceptibility Testing for ESBL-producing Enterobacteriaceae using an On-Demand Digital Dispensing Method" can be found on the Journal of Clinical Microbiology website.
Lucius, Yoon-Suk and I had fun filming this article in our tissue culture area and at the ICCB-Longwood High Throughput Screening Facility. Here is the link: "High Throughput, Real-time, Dual-readout Testing of Intracellular Antimicrobial Activity and Eukaryotic Cell Cytotoxicity." Legionella pneumophila was received an Academy Award for best supporting role.
In an article posted online today in Future Microbiology titled "How inkjet printing technology can defeat multidrug-resistant pathogens", postdoctoral fellow, KP Smith, and I discuss the potential uses of inkjet printing digital dispensing technology for addressing the antimicrobial susceptibility testing gap. |
Kirby Lab Blog Categories
All
Archives
January 2024
|