Portland, Ore. -- Engineers, microbiologists and biomedical scientists at Purdue University have collaborated on a turnkey system for identifying food-fouling microorganisms and pathogens in real-time, before products are shipped to store shelves. The scientists believe the laser-based optical pattern-recognition technology can also be adapted to the real-time identification of infections, pinpointing resistant strains in hospitals, and can assess the toxicity of new drugs or warn of biological weapons on the battlefield.
"Today microorganisms are cultured in a petri dish, sometimes in just a few hours, but then it takes several days to identify the strain using biochemical staining or DNA analysis," said Arun Bhunia, a professor of food microbiology in the Department of Food Science at Purdue.
"Our instrument--a scatterometer--can identify any microorganism in our catalog in real-time, cutting days off the usual turnaround time," said E. Daniel Hirleman, a professor at Purdue with joint appointments in the mechanical- and electrical-engineering departments.
Besides speeding the identification of food pathogens, and thereby eliminating the massive recalls of food products that now take place when contamination is found, the scatterometer could also give scientists an assist in other areas. According to Bhunia and Hirleman, who together originated the idea, the instrument could help in the search for new drugs and in identifying stem cells. Moreover, it could reduce the cost of homeland-security sensors by tenfold, the pair said.
The university, which has applied for several patents on the process, is scouting electronics company partners in order to design and engineer an under-$1,000 biosensor system that could be operated in the field even by untrained personnel. The scatterometer today exists only on the bench using $100,000 worth of scientific test equipment.
The instrument is based on what the Purdue scientists call "Bacteria Rapid Detection Using Optical Scattering Technology" in their description in the current issue of the Journal of Biomedical Optics. It works by shining the beam from a 635-nanometer-wavelength, 1-milliwatt laser diode through a petri dish containing the unidentified pathogen; capturing its image with a digital camera; and then analyzing the optical scatter pattern with custom software.
"I first built the scatterometer to analyze nanoscale-sized contaminants on semiconductor wafers, but the technique seems to work even better at identifying micron-sized microorganisms," Purdue's Hirleman related.
Just by visually inspecting the circular scattergrams, Hirleman and Bhunia could see that the patterns were unique; even different strains of the same bacteria seemed to have individual patterns. To automate the recognition of bacteria strains, the researchers enlisted the help of two other scientists at Purdue: Bartek Rajwa, a microbiologist at the Bindley Bioscience Center in the university's Discovery Park, and J. Paul Robinson, a professor in the Weldon School of Biomedical Engineering.
Whereas conventional pattern-recognition systems try to match up the visual shapes, colors and sizes of individual bacteria, here the scatter pattern emerging from a sample is analyzed by software that can learn to distinguish subtle features that are a part of the pattern for a whole colony. The analysis and recognition software designed by Rajwa and Robinson first classified features using a 120-term Zernike polynomial (named after Dutch physicist Fritz Zernike, who created the polynomials in 1934 to analyze how light-wave patterns are distorted after passing through a medium having complex aberrations).