Your nose is often the first thing to alert you to the fact that something is wrong, whether it's sniffing out a fire or knowing whether food is unsafe to eat. Now, scientists are drawing on the complicated neuroscience behind our olfactory abilities to build sophisticated smelling machines.
Electronic noses, or e-noses, are devices that are designed to ‘smell’ odours by detecting chemicals in the air. They consist of an array of sensors which react to Volatile Organic Compounds (VOCs) – chemicals which are gases at room temperature.
VOCs settling on a sensor’s surface trigger a physical change, which is recorded electronically. Each sensor in the array will respond differently to each VOC, creating a response pattern often called a fingerprint. An integrated computing system compares the resulting pattern to a 'smell database' in order to report on what it has found.
Compared to the real thing, however, current e-noses are still generally quite slow. The human nose is a very sophisticated tool and the neurochemistry of animals means that they can swiftly sniff out threats, locate food or find a mate using their sense of smell.
Neuromorphic computing
‘The olfactory system has evolved over millions of years to process chemical information,’ said Dr Michael Schmuker, a Marie Skłodowska-Curie research fellow at the University of Sussex, UK. Dr Schmuker is leading the BIOMACHINELEARNING project, which aims to use knowledge about how our brain works to improve machines, an approach known as neuromorphic computing.
“‘I’ve no doubt that this cheap, point-of-care technology will get into the doctor’s office for diagnosing and monitoring patients.’
He is using hardware developed by the Human Brain Project, a large research project that is building a computer simulation of the human brain, in order to design an electronic system that mimics how smells are processed in people. ‘By using a neuromorphic approach, we can tap an enormous resource of neuroscience knowledge about the sense of smell to improve signal processing,’ he said.
The emerging technology of this bio-inspired computing could massively improve the accuracy of e-noses. Employing brain-like neuronal networks would make e-noses faster, more accurate, and more power-efficient.
Although the project only started in September 2014, data crunching has begun in earnest, and the team have high hopes for what they might achieve.
‘Brain-like computing has the potential to outperform conventional computing when dealing with very complex, multi-dimensional signals, such as chemicals in the air,’ said Dr Schmuker. ‘Our project is about pushing the boundaries and finding out what’s possible with this new technology.’
Detecting disease
It is hoped that the new generation of e-noses will be a useful tool in detecting disease, especially in diagnosing conditions where current tests are expensive, lacking in accuracy, or unpopular with patients.
There are about 3 000 molecules in exhaled breath, which vary by disease state. An e-nose doesn’t identify these individual molecules, but recognizes mixtures of them in a pattern, similar to the way our own noses do. This could help not only diagnose whether someone is suffering from a particular condition, but also which strain of the condition they have.
The U-BIOPRED project aims to identify the biological characteristics of different types of severe asthma in order to better treat people who suffer from the condition. Early trials have shown that their e-nose could diagnose patients and pinpoint asthma subtypes in order to predict the effectiveness of therapy – all in real time and at point of care.
The project is funded under the Innovative Medicines Initiative, a public-private partnership between the EU and the European Federation of Pharmaceutical Industries and Associations, which aims to speed up the development of better and safer medicines for patients.
‘The next stage is collecting all these data into a 'breath cloud' that can be shared by doctors around the world,’ said Professor Peter Sterk, a clinical physiologist at the University of Amsterdam, who leads U-BIOPRED. ‘This will allow exchange of data, so each e-nose does not need to encounter every disease before being able to recognise it.’
The need to accurately account for environmental changes – correcting for recent meals, for example, or for patients who smoke – is an ongoing challenge. Yet, Prof. Sterk said: ‘I’ve no doubt that this cheap, point-of-care technology will get into the doctor’s office for diagnosing and monitoring patients.’
Similarly, the goal of the BreathDX project is to develop a sensor technique for detecting respiratory infections at the bedside. This could be of incredible value, given that such infections are the primary cause of death in intensive care units.
Researchers on the EU-funded project are working together to develop a simple breath test that can help diagnose pneumonia in people on ventilators to help them breathe.
Spotting which patients need antibiotics early on would significantly reduce the impact of the illness. It would also prevent antibiotic resistance by only medicating those patients with the infection.
E-noses are increasingly able to recognise and interpret a range of gases to report meaningful information. The next few years could see an e-nose explosion as sensor technology rapidly improves and becomes more affordable.
Recommended for you
