Artificial nose sniffs surgical smoke to identify brain tumours

Finnish researchers have developed an artificial nose that analyses the smoke from electrosurgery in real-time, detecting different types of cancerous brain tissue.

artificial nose
Flue gas created by an electric knife is fed directly into the measurement system (Credit: Antti Roine)

Electrosurgery devices such as electric knives or diathermy blades use electrical current to vaporise tissue, enabling surgeons to operate with limited blood loss. Vaporising biological tissue produces smoke that contains signatures of the underlying cells. Using differential mobility spectrometry (DMS), the team from Finland’s University of Tampere was able to analyse the smoke from cancerous tissue almost instantaneously, detecting the digital fingerprint of various forms of brain tumour. The research is published in the Journal of Neurosurgery.

"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," said Tampere University researcher Ilkka Haapala.

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

Prevailing practice for analysing tumours during surgery involves small samples of tissue being passed to a pathologist, who examines the samples with a microscope and reports back to the operating theatre by phone. This process is time consuming and the feedback to the surgeon is not immediate.

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

The Tampere study used its artificial nose to analyse 694 tissue samples collected from 28 brain tumours and control specimens. The system's classification accuracy was 83 per cent when all the samples were analysed, with accuracy improving under more restricted circumstances. When comparing low malignancy tumours (gliomas) to control samples, the classification accuracy of the system was 94 per cent, reaching 97 cent sensitivity and 90 per cent specificity.