Author(s): Feng Guo, Elise Bauer, Kimball Henstrom, Caroline Torgersen, Hannah Thrupp, Isaac Zabriskie, Alex Martinez, Keaton Fuller, Clint Flinders
Mentor(s): Vern Hart
Institution UVU
The current standard for UTI diagnosis involves culturing bacteria acquired from urine samples to determine the number and type of colony forming units present in the fluid. However, this process requires several days, and general antibiotics are often prescribed before a formal identification of the specific bacteria type, which contributes to increasing antibiotic resistance and can produce unwanted side effects. We present a new device for rapid UTI detection that utilizes scattered laser light to identify bacterial species, without requiring a cultured sample. In this process, light from a visible diode laser (635 nm) is projected through the fluid as it is peristaltically pumped through a circuit of plastic tubing. A special glass viewing window with a flat surface and a narrow (<0.1 mm) flow channel provides a thin layer of bacteria to reduce multi-scattering effects while maintaining a high flow volume for fast data collection. The resulting speckle patterns are then collected by a CMOS detector and analyzed by a convolutional neural network (CNN). The structure of the proposed device will be demonstrated and detection results will be presented for bacteria samples in saline solution to determine morphology of the bacteria in the sample.