Optical imaging catches bad bugs

Lest Americans believe chemical and biological warfare and terrorism are byproducts only of fanatical third-world dictatorships, consider this. In 1995, a man was arrested after crossing the border between the USA and Canada with guns, 20,000 rounds of ammunition, $89,000, and a large quantity of ricin--an extremely toxic poison extracted from castor beans. That same year, an Ohio man illegally obtained samples of yersinia pestis--the organism responsible for plague.

Nov 1st, 1998

Optical imaging catches bad bugs

Paula m. noaker

Lest Americans believe chemical and biological warfare and terrorism are byproducts only of fanatical third-world dictatorships, consider this. In 1995, a man was arrested after crossing the border between the USA and Canada with guns, 20,000 rounds of ammunition, $89,000, and a large quantity of ricin--an extremely toxic poison extracted from castor beans. That same year, an Ohio man illegally obtained samples of yersinia pestis--the organism responsible for plague.

It is thus no surprise that chemical and biological-weapons detection is becoming big business. According to Eric Croddy, a defense-industry analyst at Frost & Sullivan (New York, NY), the USA will spend $254 million on development of such technology this year. A large share of the research is focusing on development of technologies to detect and characterize biological hazards in the field at the point of contamination.

According to Barbara Seiders, a member of the pathogen detection team at Pacific Northwest National Laboratory (PNNL: Richland, WA), visual inspection of aerosol particles may serve as a first-cut screening leading to other, even more sophisticated analytical methods. In line with this, lab researchers are adapting decades of optical automatic target recognition R&D to biological and chemical weapons detection. Techniques are under review that correlate the images or scattering signatures of suspect bugs to stored baseline images and signatures via optical Fourier transform methods. Each unique image produces a unique 2-D Fourier transform. There is a clear correlation between both the angular orientation of features in an image and its transform, as well as between the radial distance of transform features and the fineness of detail in the image.

The sequence of images shown here illustrates the feasibility of such a process. At the left is a raw gray-scale image acquired with a phase-contrast microscope at 1000X that includes two E. Coli bacteria and one Erwinia herbicola bacterium. The background noise is typical of the medium used to grow the bacteria. The center image is the 2-D Fourier transform of the raw image, which is representative of the Fourier transform formed in the optical correlator. The "cloud" of diffuse structure in the center is mostly due to background. The darker areas of the structure relate to the various discrete features in the raw image. For example, the somewhat diffuse, but elongated structure at a clock angle of ~1:30 indicates the E. Coli below and to the right of center in the raw image. The image at the right is the binary image input to a 256 ¥ 256 binary optical correlator. The two E. Coli bacteria were correctly identified--there was high correlation with the E. Coli filter in the baseline data set. Neither the Erwinia herbicola bacterium nor the background noise correlated above a preset threshold--even in cases where clusters of dark pixels approximated the shape of the E. Coli reference.

The demonstration illustrates that there may be enough information present in the Fourier transforms for rapid, automated screening of bug suspects. According to Seiders, the method may indeed allow screening out benign groups of particles with high confidence and "fingering" possible suspect particles that can then be analyzed by methods capable of more-subtle identification and differentiation.

More in Optics