Artificial nose can 'sniff out' cancerous tissues

Scientists have developed an artificial nose that may help doctors identify cancerous tissue during surgery, and enable them to remove tumours with more precision.

Electrosurgical resection using devices such as an electric knife or diathermy blade is currently a widely used technique in neurosurgery.

When tissue is burned, tissue molecules are dispersed in the form of surgical smoke.

Researchers at Tampere University in Finland developed a method in where surgical smoke is fed into a new type of measuring system that can identify malignant tissue and distinguish it from healthy tissue.

"In current clinical practice, frozen section analysis is the gold standard for intraoperative tumour identification. In that method, a small sample of the tumour is given to a pathologist during surgery," said Ilkka Haapala from Tampere University.

The pathologist undertakes a microscopic analysis of the sample and phones the operating theatre to report the results.

"Our new method offers both a promising way to identify malignant tissue in real time and the ability to study several samples from different points of the tumour," Haapala said.

"The specific advantage of the equipment is that it can be connected to the instrumentation already present in neurosurgical operating theatres," Haapala said.

The technology, described in the Journal of Neurosurgery, is based on differential mobility spectrometry (DMS), wherein flue gas ions are fed into an electric field.

The distribution of ions in the electric field is tissue-specific, and the tissue can be identified on the basis of the resulting "odour fingerprint."

The study analysed 694 tissue samples collected from 28 brain tumours and control specimens.

The equipment used was developed specifically for the study. It consists of a machine learning system, which analyses the flue gas with DMS technology, and an electric knife, which is used to produce the flue gas from the tissues.

The system's classification accuracy was 83 per cent when all the samples were analysed. The accuracy improved in more restricted settings.

When comparing low malignancy tumours to control samples, the classification accuracy of the system was 94 per cent, reaching 97 per cent sensitivity and 90 per cent specificity.