We've updated our Privacy Policy to make it clearer how we use your personal data. We use cookies to provide you with a better experience. You can read our Cookie Policy here.


Using Artificial Intelligence To Identify Bloodborne Bacteria

Using Artificial Intelligence To Identify Bloodborne Bacteria content piece image
Dr. Ted Randolph with the FlowCam Nano
Listen with
Register for free to listen to this article
Thank you. Listen to this article using the player above.

Want to listen to this article for FREE?

Complete the form below to unlock access to ALL audio articles.

Read time: 1 minute

Fluid Imaging Technologies and the University of Colorado Boulder have recently entered an exclusive research agreement to determine whether the University’s artificial intelligence software can identify bloodborne bacteria.

The collaboration will establish training set data from microscopy images for the top ten most wanted bacterial strains causing blood infections, and then train a computer to identify the bacteria automatically in tandem with Fluid Imaging Technologies’ FlowCam oil immersion flow imaging microscopes.

We spoke with Kent Peterson, CEO, Fluid Imaging Technologies, to learn more about the collaboration, the technologies involved, and how they could impact patients.

Anna MacDonald (AM): Can you give us a little background to how the collaboration between the University of Colorado and Fluid Imaging Technologies has come about?

Kent Peterson (KP): Dr. Ted Randolph had been using the FlowCam instrumentation for years. He collaborated with an AI software expert to test biopharma formulations using FlowCam images. Later, when we introduced the patented FlowCam Nano, Dr. Randolph and I talked about using it to try to identify bacteria in tandem with the AI software. If so, it could alter the course of sepsis treatment in hospitals and save many lives by speeding up the process of applying the appropriate antibiotic.

AM: What benefits do you see the use of AI to identify blood borne bacteria having over traditional methods?

KP: The traditional method of identifying bacteria is to culture the bacteria long enough to obtain a large enough sample (CFU’s) to conduct a series of well plate assays to see which antibiotic is effective on which well plate. This series of assays determines which antibiotic to use on the patient. This process takes between two to three days. With bacteria imaged with the FlowCam Nano and identified with this AI software, the process of determining the appropriate antibiotic is anticipated to take about an hour. This is a material improvement and especially critical for treating newborns and elderly patients.

AM: Can you tell us more about the nano-flow imaging technology that will be used?

KP: FIT earned a patent on oil immersion flow microscopy and applied the technology in the FlowCam Nano. The concept uses high magnification objectives, collimated blue LED light source and proprietary arrangement to apply and retain oil in a vertical configuration. This permits nanoscale particles and microorganisms to be detected and imaged with unprecedented clarity.

Ruairi MacKenzie (RM): How much data will be required to train the neural networks?

KP: CU has been targeting 500,000 individual images for each bacteria species used. The plan calls for testing up to the top twenty species of bacteria commonly found in a neonatal hospital environment.

RM: How will researchers ultimately be able to input images to be analyzed by the CNN if it comes to market?

KP: For a neonatal case, one (1) drop of blood will be treated and run through the FlowCam Nano (about twenty minutes per run). The images will feed directly into the CNN software and matched against the training sets (500,000 per file). The matching process is expected to take under a minute of computer run time.

Kent Peterson was speaking to Anna MacDonald and Ruairi MacKenzie, Science Writers for Technology Networks.