A tale of dogs and sausages.
My first science experiment was to determine the power of Buffy's (the family cocker spaniel) nose. I discovered that he could smell sausage in his sleep.
If you put a piece of sausage in front of his nose, his nose would twitch, twitch some more, start to twitch rapidly, and then he'd suddenly wake up. The sausage quickly disappeared courtesy of Buffy (as did most anything that was even remotely edible), followed by an appreciative tail wag.
This lttle guy could even smell a grape in his sleep. I've always been impressed by that; grapes have no smell at all to me.
Although my nose will never rival that of Buffy's, human noses can still detect thousands of scents. You don't have to smell wine or food for a living to be able to clearly describe a complex combination of smells, e.g. "spa scent" (ginger and melon).
Scientists have made progress towards imparting this ability to electronic noses, artificial devices designed to primitively mimic the sense of smell. However, such devices are in their infancy.
The electronic nose challenge.
To date, electronic noses don't even come close to matching the unique, poetic scent descriptions that come as second nature to most people. The devices can identify at least some of the molecules that are causing a smell, and the concentration of these molecules.
However, this information doesn't necessarily help distinguish between different (complex) smells. Furthermore, if the principles of device operation are different, data from one electronic nose often can't be directly compared to that from another.
Scientists need to develop a standard device, with an improved scent description protocol, if electronic noses are ever going to match what a real nose can do. Rising to this challenge, Kenji Yamamoto (International Medical Center of Japan, Tokyo) and coworkers have pitted their electronic nose and scent characterization protocol against the volatile chemicals emitted by yeast and bacteria.
Standardizing the electronic nose.
The scientists' electronic nose is comprised of 10 sensors, each used to detect dfferent molecules at different concentrations. They pitted the sensors against nine gases, e.g. triethylamine (fishy odor) and buytlaldehyde (sweaty feet odor).
An example illustrating device standardization helps to clarify how the artificial nose works. First, the response of each sensor to one chemical was plotted.
Each sensor responded differently to the gas, i.e. some had a greater response than others. Let's say that sensor number 1, on one side of the device, responds twice as strongly to the gas as sensor 2, on the opposite side of the device.
The device software would therefore plot a line that leans more strongly towards sensor 1. Taking into account all of the ten sensors in the device, the software plots a line that is a specific angle away from baseline (no response).
Considering that there are ten sensors in the electronic nose, this measured angle away from baseline is unique for each gas. Measuring this angle for all nine gases, as a function of gas concentration, is a very effective method for standardizing device response.
There's a little bit more math that goes into the standardization procedure. However, this gives the general idea.
Pitting the electronic nose against yeast and bacteria.
The scientists found that when there were 100 yeast colonies per milliliter in their sample, the increase in the measured angle (relative to baseline) in response to hydrogen sulfide and ammonia was small. Little of these chemicals were emitted.
However, there was a large increase in the measured angle in response to carbon-based acid. Much of it was emitted.
At ten thousand and greater yeast colonies per milliliter, an increase in measured angle was observed in response to all chemicals. In contrast, at even ten million bacteria colonies per milliliter, the measured angle increased only for hydrogen sulfide (only this gas was emitted).
This demonstrates that the scientists' electronic nose detects very small numbers of yeast cells. It also demonstrates that it can distinguish between yeast and bacterial cells.
The volatile chemical composition (smell) was dissimilar enough from the control experiments to detect yeast, but can discriminate them from bacteria only when the bacteria concentration is greater than 100,000 colonies per milliliter. Additionally, measuring only smell intensity eliminated the capacity of the device to discriminate between yeast and bacteria.
Future prospects.
These scientists' electronic nose requires 90 minutes to collect data. This is still a lot slower than Buffy, who can go from sleeping to chowing down on a piece of sausage in a few seconds.
It's still progress though, especially in its unique capacity to clearly identify the molecular components of a complex mixture. It's also much cheaper than the specialized chromatography instrumentation that scientists would otherwise need for scent characterization.
for more information:
Fujioka, K., Arakawa, E., Kita, J.-i., Aoyama, Y., Okuda, T., Manome, Y., & Yamamoto, K. (2009). Combination of Real-Value Smell and Metaphor Expression Aids Yeast Detection PLoS ONE, 4 (11) DOI: 10.1371/journal.pone.0007939