Led by the nose

In a lab overflowing with circuit boards and bits of wire, electrical engineer Pamela Abshire holds a 5-centimeter-long, rectangle-shaped device between her thumb and index finger. From the bottom of the device, dozens of tiny copper-colored teeth jut out, while up top, a tiny round, clear plastic container covers a bright yellow square with a tick-sized silicon chip at its center. Abshire is hoping the clear plastic container will soon house odor-sensing mammalian cells that can detect drugs, explosives, or bodies buried under rubble.

Abshire and her colleagues have spent the last few years pondering how to employ the keen smell of a dog's nose without having to actually employ a dog. Dogs can sniff out anything, but each one costs more than $100,000 to train, and they quickly get bored and tired.

When interviewing for a position at the University of Maryland's Institute for Systems Research in College Park in 2002, Abshire began trading ideas on how to mimic biological sensing abilities with mechanical engineer Elisabeth Smela. Smela had been working on ways to combine cells and chips since 1998, and Abshire had studied blowfly photoreceptors as a grad student, which gave her a good feel for the sensory system. She got the job. Soon, they brought aerospace engineer Benjamin Shapiro into the discussion. "We started talking together, and instantly we all had some sense that 'that's cool, we should do it!'" Shapiro says.

By 2004, they were working on an approach that combined olfactory neurons and computer chips. Once the device is complete, the neurons, which are currently harvested from rats, will sense smells, and transmit signals to semiconductor chips like those in cell phones or Palm Pilots. Abshire is building the electronic components, while Smela and Shapiro are devising ways to place the neurons precisely in position on the chips. The team has already developed the packaging, the computer chips, and neuron placement methods, and have presented their progress at several IEEE conferences. They hope to have a very simple proof-of-concept prototype with one live neuron connected to a chip in a year's time. But it could be many years before the technology is viable.

Part of the problem is the complexity of the signaling process itself. When an odorant—whether it's a molecule that makes up the scent of fresh chocolate cake or explosives—binds to the receptors on the surface of an olfactory neuron, it sets off a cascade inside the cell, amplifying the signal and triggering the cell membrane to depolarize. But most smells are a combination of many different odorant molecules in different proportions, so the brain has to process the signals it gets from dozens of neurons to pinpoint a specific aroma. To address that challenge, each device will likely have to be able to process signals from many neurons in order to detect smells accurately.

Though the task is large, it's not completely out of left field, says Tim Pearce, a bioengineer at the University of Leicester in England who studies pheromone receptors in moths and is working on his own nose-on-a-chip. Directly detecting the electrical signals from the neurons is potentially "a low-risk approach," because technology for detecting these signals from slices of brain tissue are already on the market, he says.

Before they can get a working prototype, however, they'll have to troubleshoot several problems. They rely on immortalized neuronal cell lines—a set of cloned cells drawn from a single rat olfactory neuron—and it's not clear how long (or if) the cells could stay alive on a functioning chip. Even if the cells survive for a reasonable length of time (the group hopes a month), the team still needs to experiment with neuron spacing, and determine whether the olfactory neurons must form networks with each other to function. Despite the many problems, the group doesn't seem daunted. "What I like about us is the 'no fear' attitude," Shapiro says.

Although I agree with the above two comments, I think this is a very innovative idea you are working on. This will be something new and exciting for young scientists and it will bring out the extent of your intelligence and potential. Good luck.

While this technique may work as postulated, it sounds not only expensive, but cumbersome. But if viable, new designs may be arrived at stepwise, until a device exists that can detect multiple smells, and at a molecular level.

The smells would be collected in a database. It ould be invaluable for various kinds of investigative sciences, as well as forensic police work.

The ultimate method of detection may differ significantly from the current project, where animal smell detection is being mimicked.

If you think training a dog is very expensive, figure out how much this experience for training a single neuron will cost. The prospect may be of immense value, but it will cost much more than a dozen of dogs and take much more time than training the whole bunch of them. If you need to sniff out something soon, I respectfully suggest that you're better off with the dog.