Pattern Classification Techniques for Lung Cancer Diagnosis by an Electronic Nose - Computational Intelligence in Healthcare: Advanced Methodologies - page 382

electronic nose, combined with the appropriate computational intelli-
gence methodologies, is a promising alternative to current lung cancer
diagnostic techniques: not only the instrument is completely non inva-
sive, but the obtained predictive errors are lower than those achieved by
present diagnostic methods, and the cost of the analysis, both in money,
time and resources, is lower. The introduction of this cutting edge tech-
nology will lead to very important social and business effects: its low
price and small dimensions allow a large scale distribution, giving the
opportunity to perform non invasive, cheap, quick, and massive early
diagnosis and screening.

1 Introduction

In the last decades, attention on electronic noses has considerably increased and
research in the analysis of olfactory signal has become very lively; this is mainly
due to the wide variety of problems that this innovative technology is potentially
able to solve. Electronic noses have been successfully applied to a vast range of
applications, from cosmetic productions to food and beverages manufacturing,
from chemical engineering to environmental monitoring, passing through mines
and explosives detection as well as medical diagnosis. The latter application
plays a crucial role in social environment and health care monitoring, provid-
ing diagnostic clues, guide to laboratory evaluation and affecting the choice of
immediate therapy.
One of the diseases with the highest mortality rate is lung cancer, which causes
3000 deaths each day in the world and is the leading cause of cancer death
among both men and women. The diagnosis in an advanced stage represents
the main cause of the therapeutic failure: the surviving rate after five years of
treatment, if lung cancer is diagnosed in stage IV, is around 2%; this percentage
increases to more than 50% if the cancer is discovered in its earliest stage. These
considerations highlight the relevance of performing massive accurate diagnosis
and, in particular, early diagnosis. Chest x-rays screening have proven ineffective,
and spiral Computed Tomography (CT) trials are on-going. However, imaging
tools are expensive and their accuracy is limited by false positive findings. Less
invasive and more accurate techniques are necessary to identify lung cancer in
its early stages.
A possible solution to this task can be found in the fundamental principle of
clinical chemistry, according to which every pathology changes people chemical
composition, modifying the concentration of some chemicals in the human body.
This happens also in lung cancer: it has been demonstrated that the presence of
lung cancer alters the percentage of some volatile organic compounds (VOCs) in
the human breath [8,9]. These VOCs can thus be considered as lung cancer mark-
ers; the analysis of the olfactory signal of patients' breaths and the recognition
of these VOCs in them, through the appropriate multivariate pattern analysis
algorithms, can be the tools key to detect lung cancer.