A machine that has a nose to detect lung cancer

Breathe, Scout. There are a dozen Franco-Belgian researchers and companies who have been working for some years now on the Pathacov project, a machine for detecting bronchopulmonary cancer using volatile organic compounds found in the breath. If not yet fully developed, this electronic nose promises to save precious time in the early detection of this pathology.

Everyone has had the bitter experience of guessing what a person might have eaten or drunk just by smelling their breath, especially when it comes to garlic or alcohol. But diagnosing a disease in your interlocutor in this way is another matter, although the principle remains the same: detect and analyze the volatile organic compounds (VOCs) specific to such foods, such drinks or such diseases. The exercise is not impossible, but not with a human nose. On the other hand, dogs are already having success, especially for breast cancer through the KDOG project under the Institut Curie. “I can’t see myself setting up a kennel in the CHU or putting a dog in every GP office,” jokes Professor Arnaud Scherpereel, head of the thoracic pulmonary oncology department at Lille University Hospital.

A machine that acts as a biological nose

Hence the idea of ​​developing a machine capable of doing the dog’s work in detecting CoVs specific to lung cancer. “The CoVs in the organs change when they become cancerous and thus have their own molecular signature”, explains Professor Régis Matran, pulmonologist at Lille University Hospital. “We collected samples from 500 sick or healthy patients to build a data set that is a real gold mine,” he continues. This dataset will be used to teach the machine’s artificial intelligence to detect and identify fingerprints of a sick person from a healthy person.

This “sorting” will be the task of several miniaturized gas sensors designed by Doctor Driss Lahem from the company Materia Nova. The module with the sensors is integrated into the “electronic nose”. “As a whole, the machine works exactly like a biological nose, whose receptors transmit information via neurons to the brain”, explains Justin Martin, PhD student from the University of Liège, who developed the first prototype. The impression, generated by the “nose” algorithm of Covs, is sent to the attending physician, who, by including the patient’s data, will be able to arrive at an interpretation.

Scientists still lack enough data to make their “nose” work optimally, admits Justin Martin. A deficiency that was soon to be filled by clinical tests on sick patients. The machine itself also needs to be streamlined to easily integrate practitioners. “The aim is to make it a non-invasive, accessible and inexpensive diagnostic tool upstream of breast scanner screening”, insists Professor Scherpereel. There is a lot at stake in terms of public health. Detected early, lung cancer can be treated curatively, giving a 90% life expectancy at 5 years, compared to 20% over the same period for cancer that has entered the metastatic stage.

Leave a Comment