Electronic Nose Detects Pneumonia in Critically Ill Patients

Electronic Nose Detects Pneumonia in Critically Ill Patients

(Philadelphia, PA) According to a team of researchers from University of Pennsylvania School of Medicine, an electronic nose – a relatively new version of a sensor previously used in the food, wine and perfume industries – can quickly and accurately diagnose pneumonia in critically ill, mechanically ventilated patients. The results will be presented at the CHEST 2002 Annual Meeting Tuesday, November 5th in San Diego.

“We wanted to further explore using the e-nose after the exciting results of an initial study we conducted back in 1997 with only 20 patients,” said C. William Hanson, III, MD, professor of Anesthesia, Surgery and Internal Medicine, and lead author of the study. When it comes to lower pulmonary infections, especially in critically ill patients, time is of the essence for disease control. “Rather than waiting two to three days for the results of a bacterial culture or relying on chest X-rays which aren’t always accurate, the e-nose can give us a head start toward a proper diagnoses. We could avoid over-prescribing powerful antibiotics which are usually given to patients while we’re waiting for their test results, even though we don’t know if they actually need them,” adds Hanson.

In the current study, 415 mechanically ventilated, critical care patients were screened for the presence of ventilator associated pneumonia (VAP) using a clinical pneumonia score (CPIS). Patients with high CPIS scores were enrolled in the study as well as control patients who had no evidence of pneumonia. An exhaled breath sample was taken from each patient directly from the expiratory limb of the ventilator circuit into an electronic nose made by Cyrano Sciences, Pasadena, CA. This differs from the original study where breath samples were collected in plastic bags from the ventilators of intubated intensive care patients and then fed into a different electronic nose.

The e-nose contains an array of sensors consisting of carbon-black/polymer composites. The patient’s exhaled breath gas was passed over these sensors which interact with volatile molecules to produce unique patterns that are displayed in two-dimensional “maps,” or dot patterns on a computer screen. The results were analyzed using pattern recognition algorithms and assessed for a correlation between the actual CPIS scores and the one predicted by the nose. Hanson and his colleagues found that the nose made clear distinctions between the patients who were infected and those who were not.

“The data show good correlation between the actual scores and those predicted by the data from the e-nose sensor,” said Hanson. “Furthermore, this study suggests that the commercial electronic nose, as is, would be reasonably successful in predicting ventilator associated pneumonia. It would be even more suited to the task if the sensor array could be customized.” Preliminary data also suggests that the e-nose may be able to distinguish between pneumonias caused by different bacterial infections.

Cyrano Sciences, Inc., donated a “Cyranose” electronic nose for use in this study.