New microscope technique could speed bacteria identification
08 June 2015
Holographic microscope coupled with machine-learning software could provide an inexpensive, point-of-care solution for hospitals and contamination screening of food.
A new way of rapidly identifying bacteria, which requires a slight modification to a simple microscope, may change the way doctors approach treatment for patients who develop potentially deadly infections, as well as helping the food industry screen against contamination with harmful pathogens.
Described in the journal Optics Express, the new approach, developed by researchers at the Korea Advanced Institute of Science and Technology (KAIST) in Daejeon, South Korea, involves scattering laser light off individual bacteria under the microscope, creating holographic images of them, and then using a Fourier transform and computer software to analyse the images and identify them by comparing them to other, known bacteria. The software uses a machine-learning algorithm similar to that used by security cameras for automated face recognition.
If the approach proves to be effective in controlled clinical trials, it could lead to a powerful new way of routinely and almost immediately identifying dangerous bacteria at the bedside - much faster than the days it normally takes to culture a colony in the laboratory from a patient's blood, which is still the gold standard in the health care industry for making a definitive diagnosis.
Also routinely used today is a newer method for rapidly identifying bacteria based on a DNA-analysis technique called quantitative polymerase chain reaction (qPCR), but it still may take hours to return a result, and it requires expensive sample preparation that is often cost prohibitive for routine use in rural or resource poor settings around the world.
The challenge of meeting clinical needs in the developing world was one of the motivations behind the work, according to the KAIST team. They wanted to find an inexpensive method that could surpass the speed of qPCR.
"Employing laser holographic techniques, we achieved rapid and label-free identification of bacterial species at the single bacterium level with a single-shot measurement," says physicist YongKeun Park, who led the KAIST team. "This means the present method can be utilised as a pre-screening test for point-of-care bacterial diagnosis for various applications including medicine and food hygiene."
"We have also developed a compact portable device [a quantitative phase imaging unit] to convert a simple existing microscope to a holographic one, in order to measure light scattering patterns of individual bacteria," says Park.
The new quantitative phase imaging (QPI) technology promises to deliver better point-of-care diagnostics by reducing the time it takes to specifically identify bacteria, which may guide treatment, allowing doctors to prescribe the best drugs available to treat an infection and improving outcomes for people with hospital-acquired infections, though the effectiveness of the approach remains to be proven in future clinical trials.
In their initial experiments, Park and his colleagues showed as a proof of principle that they could identify bacteria with high accuracy. They examined four different bacterial species (Listeria monocytogenes, Escherichia coli, Lactobacillus casei, and Bacillus subtilis). The first three are all pathogens known to infect humans through the food chain or via hospital-acquired infections. The fourth is a harmless bacteria used in laboratory research but of great interest because it is closely related to the deadly Bacillus anthracis, which is the base for Anthrax.
Under a microscope, all four of these rod-like bacteria look nearly identical. They would be virtually impossible to distinguish. The KAIST team, however, sorted them using the QPI process with an accuracy greater than 94 percent.
The technique employs a Fourier Transform which allows the distinct, fingerprint-like light scattering pattern for any given bacterial cell to be defined. Further analysis uses a conventional approach to statistical classification or machine learning - a sorting strategy based on pattern similarities. Park believes this is the first time anyone has applied machine learning to Fourier Transform light scattering data.
They are now looking to extend their initial work to see if they can distinguish between several types of bacterial subgroups, to identify the most drug resistant or virulent strains from the innocuous ones.