Developing an electronic nose for trouble

Lung cancer, ulcers and even schizophrenia may soon be diagnosed early by a machine that smells a patient's breath. Steve Connor finds out how George Dodd has a nose for smells. "Rub your scalp like this," he says before bringing his fingertips to his nose, "and smell ... you'll find it has a lovely warm and agreeable odour." Dr Dodd, a smell researcher for more than 25 years, believes the body's odoriferous offerings have a deep significance. His ambition is to build an electronic nose that can provide doctors with an early diagnosis of diseases ranging from lung cancer to schizophrenia. He calls it his intelligent breathalyser.

Dodd's vision of the future hospital is one where the patient walks through a battery of sophisticated scanners and automatic diagnostic machines before being seen by a doctor. "You'll probably spend a minute in something that looks like a telephone booth which gives you your own individual odour signature." The breathalyser sniffs the patient's mouth for the molecular by-products of disease, its electronic nose quivering whenever it detects a smell that could indicate something wrong.

The scalp-rubbing exercise was Dr Dodd's demonstration that we are often unaware of the quite normal, healthy and occasionally pleasing smells emanating from our bodies. One of his professional sidelines has been developing new perfumes, and he has researched the chemistry of scents. "Two of the molecules from that smell of the scalp turn out to be very important in perfumery and have been used intuitively for thousands of years. Exotic oils from India which are used in some of the most expensive perfumes in the world have those molecules," he says.

But it is the smell of what goes wrong with the body that interests him most. For much of his quarter- century of olfactory research, Dr Dodd was based at the University of Warwick. Last autumn he moved to a new base in Scotland where he intends to establish a new "institute for body odour imaging" at the Highland Psychiatric Research Group at Craig Dunain Hospital in Inverness. He hopes to develop his intelligent breathalyser with the help of John Barker, professor of microelectronics at Glasgow University.

Electronic noses already exist in the commercial world where they are used for simple, routine analysis. Since more than 95 per cent of the sense of taste is in fact due to the sensation of smell, such detectors have the potential for replacing human taste panels where skilled professionals are employed to sample the quality of a product.

Cigarette companies now have electronic noses for sniffing tobacco; the food industry uses them to determine the ripeness of fruit and the maturity of cheese, and beverage manufacturers have them for anal-ysing anything from instant coffee to beer and blended whiskies.

These commercial electronic dev-ices, however, are not particularly good at deciphering the range of odours a human nose can detect, but they are far better at picking out extremely low concentrations of the molecular scents they are "trained" to identify. Electronic noses used by instant coffee manufacturers, for instance, can detect the one molecule among 670 in the smell of coffee that gives it its unique odour. There are five of these molecules for every million, million (a British billion) other co ffee molecules. The molecule is present in such low concentrations that it proved impossible to identify until the electronic nose came along.

Understanding the human sense of smell was critical to designing an artificial nose, Dr Dodd says. "Smell is special and different from all other senses. The bone at the top of your nose is drilled like a colander with millions of incredibly fine holes, and the brain cells or neurons actually hang through into your nose, waving like anemones on a coral reef. Those neurons are the receptors. Smell is more than just intimately connected with the brain; it is the brain."

There are about five million of these sensory "anemones" on either side of the nasal cavity. They are the primary receptors which make contact with the volatile molecules that waft by. As each odour molecule strikes a sensory cell, it causes electrical discharges in the neurons which are relayed to a second set of nerve cells. "You have millions of these primary sniffer cells and on average about 20,000 of these `talk' to the secondary type of cell in the hierarchy," Dr Dodd says. "All these primary sniffer cells are generalist cells. No specialist cells appear to be present. They have the capacity to sniff almost anything. This means that, in engineering terms, it's an `ensemble' problem. You need a vast computer to interrogate the array o f neurons and process the information."

Dr Dodd's electronic nose is based on a similar, if far simpler, principle. The sensory "cells" are in fact hair-thin strands of plastic polymers which act as semiconductors. Each time a molecule from a certain smell touches the polymer, it alters the electrical conductance of the material, triggering an electrical signal. The typical odorants which can do this are small "lipophilic" molecules (which can dissolve in fat) such as limonene, one of the active ingredients in the smell of lemon; thymol, an ingredient of the smell of thyme; and octadienone, the molecule responsible for giving rancid butter its off-flavour.

Any smelling device, whether it is one of nature's noses or a man-made electrical sniffer, has to cope with the problem of permanent damage to the delicate sensing machinery. "The common event in any type of sensor is a molecule coming in, sticking to a surface and causing some kind of electrical signal," Dr Dodd says.

The trauma of being bombarded by molecules damages living tissue and the delicate plastic polymers of electronic noses. "Olfactory neurons last about a month," says Dodd. "The supposition is that they are damaged by the environment and have to be replaced. They are oxidised by the air being breathed in all the time."

The same element of built-in redundancy has to be applied to electronic noses. "You have to take on- board the whole problem of poisoning. These machines are going to earn a living in factories, sniffing beer or wine for instance, and we know they are going to be damaged. We are not going to prevent that. Our strategy is to build massive redundancy into the next generation of noses. We will probably end up with a situation where only 1 per cent of the sensors are working at any one time, the other 99 per cent being cleaned."

An alternative to cleaning is to discard any cells that become damaged. Ten years ago, smell researchers deemed this was so expensive and wasteful they did not consider it a serious option. But, as sensing dev-ices become smaller and cheaper, there is now a belief that an electronic nose can be built where just a small fraction of its sensory cells will be employed at any one time, ready to be replaced by an army of others should any of them fail. In fact the sensory strands of electroactive polymers Dr Dodd uses as smell sensors can now be just one molecule thick. They are, he says, the "smallest wires anyone can imagine making".

As the electronics of artificial noses become smaller, cheaper and more sensitive, Dr Dodd's ambitions for an intelligent breathalyser be-come more realistic. But what sort of diseases is he hoping to diagnose?

There are some medical conditions that doctors are already trained to spot with the help of breath analysis, he says. "If you are in the beginnings of a diabetic coma, the body tries to divert energy metabolism and as a result produces acetone on the breath. All doctors and nurses are trained to sniff the breath of an unconscious person to detect the smell of acetone. But you only get it if somebody is in a diabetic coma or is about to enter a diabetic coma." With a machine that can detect acetone at much lower levels than the human nose, perhaps doctors will get an earlier warning of diabetic coma, he says.

Lung cancer is another condition he hopes to investigate as a possible candidate for breath analysis. It has already been shown that cancer patients have greater amounts of certain chemicals on their breath, he says. The same is true of tooth infection. "Breath analysis can indicate emission of volatile chemicals at the very early stages of the invasion of a tooth root by bacteria, long before a patient is aware of any swelling or discomfort. That means it could alert you to early trouble long before itdevelops into an abscess."

Other candidates for the intelligent breathalyser include liver disease, kidney problems, pancreatitis and duodenal ulcers. Detecting ovulation by monitoring the volatile chemicals on the breath of women has already been used to follow the menstrual cycle, Dr Dodd says. Such research would prove invaluable for in vitro fertilisation and may also lead to a more accurate "natural" method of birth control. Dr Dodd and his Glasgow colleague, Professor Barker, have already won a contract to develop such a system to help detect ovulation in cattle, a critical time for artificial insemination.

Even schizophrenia, Dr Dodd says, may prove amenable to breath analysis. Research published last year demonstrated, for example, that schizophrenic patients have significantly higher levels of hydrocarbons - gases such as pentane - and carbon disulphide on their breath, possibly as a result of an imbalance in metabolism which affects nerve tissue. If his hunch proves right, a quick and simple diagnosis of schizophrenia would reduce the time and effort currently spent on trying to diagnose new arrivals at mental hospitals.

Should Dr Dodd's ambitions become a practical reality, then an intelligent breathalyser for hospitals will transform the diagnosis of disease. It could well make a science out of the art of sniffing. !