Study to investigate use of artificial intelligence for identifying bloodborne bacteria

The artificial intelligence technology aims to reduce the time required to correctly identify the bacterial species causing an infection to 60 minutes or less onsite.

Dr. Theodore Randolph of the University of Colorado with the FlowCam Nano from Fluid Imaging Technologies, who will lead research using the instrumentation that aims to automatically determine bacterial species in blood for fast, lifesaving treatment.
Dr. Theodore Randolph of the University of Colorado with the FlowCam Nano from Fluid Imaging Technologies, who will lead research using the instrumentation that aims to automatically determine bacterial species in blood for fast, lifesaving treatment.

Fluid Imaging Technologies (Scarborough, ME), and the University of Colorado Boulder (Boulder, CO) have entered into an exclusive agreement to conduct primary laboratory research aimed at determining whether the university's artificial intelligence (AI) software can detect bloodborne bacteria and identify the species from images collected using the company's FlowCam Nano particle imaging and analysis system.

In the study, entitled "Application of Convolutional Neural Networks and Flow Imaging Microscopy to Analysis of Blood Infections," the researchers will evaluate the 10 strains of bacteria most responsible for the 1.5 million sepsis cases and 250,000 fatalities annually in the U.S. per Centers for Disease Control and Prevention (CDC) data. These cases cost upwards of $6 billion in annual Medicare payments, according to the Centers for Medicare and Medicaid Services. The top 10 most wanted bacterial strains are: 

  • Staphylococcus aureus 
  • Staphylococcus epidermis
  • Staphylococcus haemolyticus
  • Enterococcus faecalis
  • Streptococcus agalactiae
  • Escherichia coli
  • Streptococcus pneumonia
  • Listeria monocytogenes
  • Enterobacter cloacae
  • Enterobacter aerogenase

To be conducted under the direction and supervision of Dr. Theodore Randolph of the Department of Chemical and Biological Engineering and the study's principal investigator, the research study will establish training set data for the 10 strains from representative Nano-Flow Imaging microscopy images, and then apply the university's deep convolutional neural network software to train a computer to identify the microorganisms automatically. Ultimately, the research team hopes to reduce the time required to correctly identify the bacterial species causing an infection from several days in a laboratory to 60 minutes or less onsite via the FlowCam Nano. Once identified, the proper antibiotic may be prescribed for fast, effective treatment.

Featuring optical technology for unprecedented image resolution, the FlowCam Nano flow imaging particle analyzer automatically detects, images, and characterizes micron- and sub-micron-sized particles and microorganisms ranging in size from 300 nm to 10+ µm. In addition to imaging bacterial strains, the imaging particle analyzer has proven effective in imaging red and white blood cells, protein agglomerates, silicon oil droplets, carbon nanotubes, yeast, and a variety of other nanoparticles.

For more information, please visit fluidimaging.com.

More in Biophotonics Techniques