Electronic nose device for early diagnosis of gastric cancer

The current standard method for gastric cancer diagnosis is based on endoscopy, followed by biopsy in the case of detection of anomalies. This technique needs to be realised by a qualified doctor, it is invasive, unpleasant, and can lead to complications. Because of these disadvantages, some patients can refuse to do it even if they are at the category of high risk. Moreover, it is expensive and for this reason, it is not available for general population screening in most the developing countries (such as Colombia). The accuracy of endoscopy is around 95%. The results of this test are obtained in several days.

The system developed is portable, inexpensive, it causes no danger or unpleasant to the patients, provides quick results (within minutes), and it is easy to use. Therefore it can be applied even in non-specialist settings (e.g., by the family doctor). The accuracy obtained in the first trial was > 94%. This test is very suitable for general population screening, for selecting the subjects at high risk of suffering from gastric cancer, and send only these patients for further investigations by endoscopy.

Specifications

The device is formed by a matrix of resistive chemical gas sensors based on gold nanoparticles functionalised with different organic compounds, placed inside a sealed test chamber. For realising the test, the patient breathes in a BioVOC breath sampler system until he/she completely empties his/her lungs. From the BioVOC the breath is transferred to the test chamber by means of a plunger, which is pushed slowly (10 sec). The sensors react through physo-chemical interactions with the compounds from the breath, which produces variations in the electronic properties of the sensor. These changes depend on the interactions between the sensing material of the sensors and the compounds from the breath. In our case, during device operation we applied a constant voltage of 8V to the sensors and we acquired the current through the sensors. We analysed sensors response, from which we extracted the most important features, which were used to train a pattern recognition model. Sensors responses to unknown samples are then projected onto the model, which produces an estimation (ill / not ill).

Current development status

Working prototypes

Applications

The invention can be applied for the general screening of the high risk population of suffering from gastric cancer. It is very simple to use, and can be applied even in non-specialist settings (e.g., by family doctor). It have great potential to be introduced as a routine test at the annual medical checking.