News for Nose: Machines Use Odor to Diagnose Disease

Why do machines that can smell the roses seem so exotic? Maybe it’s that the sense of smell seems characteristically human, intimately tied to the way we think and remember. (The association of smell and higher brain functions may not be that farfetched considering that mammals may have followed their noses down the path of evolution.  As a 2011 Nature news article put it, “an improved ability to suss out scents in our 200-million-year old ancestors may have laid the groundwork for the bulging brains of humans and all other mammals.”)

But a rising tide of research clearly shows that breath analysis by artificial noses is very close to being a fast, noninvasive, and practical everyday diagnostic tool—as papers presented at the American Chemical Society National Meeting  (ACS, 8-12 Sept. in Indianapolis, Ind.) and the European Respiratory Society Annual Congress (ERS, 7-11 Sept. in Barcelona) underscore.

Odor has been a key diagnostic criterion as long as there have been people. Even today, parents and doctors know that there’s a “nose” in diagnosis. The breath of a child with strep smells metallic. Diabetics’ urine smells sweet, but their breath can smell of acetone and rotten apples. Infectious diseases like cholera, diphtheria, smallpox, pneumonia, tuberculosis, typhoid fever, and yellow fever all produce characteristic odors; so do many types of cancer. The sweat of a baby with phenylketonuria will smell like locker-room towels; other metabolic diseases produce scents redolent of sweaty feet, or maple syrup, or hops, or cabbage, or rotten eggs. Even a disorder like schizophrenia can be associated with a musty smell (a symptom noted in the 1870s and reconfirmed by gas chromatography-mass spectrometry (GC-MS) in the past decade).

Cyranose 320, SensigentAt the ERS meeting in Barcelona, a group from the Pauls Stradins Clinical University Hospital and the University of Latvia, both in Riga, presented five papers on diagnosing respiratory diseases by artificial-nose analysis of exhaled breath. One of the studies used a commercial electronic nose (the Cyranose 320, a spinoff of a NASA project) to screen the exhaled breath of 475 patients, of whom 252 had been diagnosed with lung cancer and 223 with non-cancer lung diseases. The breath test properly identified 96 percent of the patients with cancer. Among non-cancer patients, the false-positive rate was about 9 percent. A companion study went old school, with GC-MS profiling of patients with a variety of respiratory conditions: 31 with lung cancer, 19 with congestive obstructive pulmonary disorder (COPD), 11 with pneumonia, and 10 healthy volunteers. The researchers examined the exhalate for levels of seven volatile organic compounds, and found clear diagnostic signatures: Levels of four compounds—methanol, ethanol, decane, and dodecane—were markedly lower among patients with cancer than in any of the other groups. Patients with pneumonia and COPD showed elevated levels of p-xylole and decane, while COPD patients had lower methanol levels than pneumonia patients. Exhaled breath, in short, should be a quick, useful tool in diagnosing lung disorders.

At ACS in Indianapolis, James R. Carey and collaborators from the National University of Kaohsiung in Taiwan described an infectious-disease-sensing device that is, if not exactly an artificial nose, certainly a color-coded biotech tongue. It’s a disposable, palm-sized plastic bottle, filled with a microbial growth medium. On the side of the bottle, separated from the growth medium by a permeable membrane, is an array comprising 36 dots of different chemoresponsive dyes. A diagnostician injects a small sample of the patient’s blood into the bottle and allows any infectious organisms to multiply. 

As the bacteria or fungi grow, they release characteristic combinations of chemical byproducts. These metabolites pass through the membrane and react with the dye dots, changing their colors. The dots’ changes are tracked over time on a flatbed color scanner and plotted out to reveal the organism’s unique signature. The process takes 24 hours—a third of the time of conventional cultures, a significant advantage when starting the right treatment fast can make a big clinical difference.

These are just the latest in an accelerating trend. In recent months, IEEE Spectrum has noted reports on surgical knives that can tell by smell whether they are cutting through healthy or cancerous tissue, and robots that can recoil at the smell of a person’s breath, smell the difference between apples and pears, and detect explosives. Other recent reports include new carbon-nanotube-based Raman spectroscopy sensors “almost as sensitive as a dog’s nose” that can detect chemical vapors at the level of a few hundred femtomoles per liter; advances in nanosensor technology for detecting pathogens and chemical contaminants in the food supply; and using GC-MS to identify melanomas from volatile organic compounds wafting into the air.

But this is just the tip of the iceberg: According to Google Scholar, so far this year, some 1200 journal papers have discussed the "electronic nose," and 31 employed the Cyranose 320 in particular (The more tightly curated PubMed tallies are 59 overall and 3 using the Cyranose, respectively.) The applications have included methods for diagnosing colorectal cancers, adenomas, pulmonary aspergillosis, sleep apnea, gastro-esophageal reflux disease (GERD), malignant mesothelioma, liver function, gastric cancer, and intestinal bacterial overgrowth…among many other conditions. 

Images: Hospital of the University of Pennsylvania; Sensigent

Advertisement

Tech Talk

IEEE Spectrum’s general technology blog, featuring news, analysis, and opinions about engineering, consumer electronics, and technology and society, from the editorial staff and freelance contributors.

Newsletter Sign Up

Sign up for the Tech Alert newsletter and receive ground-breaking technology and science news from IEEE Spectrum every Thursday.

Advertisement