Abstract
In the paper, the authors analyze the preliminary results of testing a classical gas sensing instrument - the electronic nose (a metal oxide transistor sensor of chemical substances) in a hospital where patients with different lung diseases are treated. To reveal the correlation between the amplitudes of the sensor's responses and the patients' diagnoses, different statistical analysis methods have been used. It is shown that the lung cancer can easily be discriminated from other lung diseases if short breath sampling and analysis time (less than 1 min) is used in the test. Volatiles obtained from a breath sample of a patient with lung cancer give the major contribution to the responses of different e-nose sensors, so in these cases highly precise identification could be achieved.