Demonstrating the Diagnostic Power of an Electronic Nose: Pilot Study on Exhaled Air Samples

The investigators don't know yet how the nose and the brain decode the smells. Scientific studies in neuroscience have shown that people who have tumors may have changes in the smell of secretions. Dogs are extremely efficient at detecting these changes, even before imaging studies. A review of the recent literature shows the different work done on the diagnosis of dogs on human pathologies, especially oncology. It is now known that the smell of exhaled gases is representative of the intestinal biotope and that a large number of pathologies are related to the type of microbial populations that inhabit the intestines.

Copying the olfactory organs could thus be of major interest for the early diagnosis of pathologies. More and more works are interested in the diagnostic power of electronic noses. From a technical point of view, these are nano-sensors that mimic the olfactory receptors from the breath gas of the subjects. They analyze the molecules present and compare them with a database to establish a diagnosis according to a probabilistic algorithm.

The use of exhaled air for the diagnosis of cancerous pathologies has already been the subject of scientific work. A classification using the SVM (support vector machine) method using data from 320 sensors made it possible to differentiate patients with lung cancer from controls in 98.8% of cases. The differential diagnosis of obstructive bronchopneumopathy was also very well done in this same study. Another study shows equally encouraging results, highlighting sensitivities and specificities above 80%.