Biomedical Engineering Reference
In-Depth Information
maximum likelihood estimator. Neural networks have also been used for processing signals
from microbial sensors (28,29). Lobanov et al. (28) describe how it is possible to use sensors
based upon whole microbial cells to selectively measure ethanol and glucose in mixtures.
One active area of pattern recognition and identification research in recent years is the
development of “electronic noses” for qualitative classification of various kinds of chemical
stimuli (30); Figure 5.2. A comprehensive overview of electronic nose design are provided by
Göpel et al. (31) and Ziegler et al. (32). The role of neural networks and genetic algorithms in
enhancing the performance of electronic noses is discussed in the paper by Kermani et al. (33).
The two main components of an electronic nose are the sensing system and the automated
pattern recognition system (30). The combination of broadly tuned biosensors coupled with
sophisticated information processing algorithms enables the electronic nose to be a powerful
instrument for odor and toxin analysis. The sensing system may be an array of biochemical
sensors where each individual sensor measures a different property of the sensed analyte, or
it can be single sensor that produces an array of measurements for each sample. The desired
odorant (or volatile chemical compound) presented to the sensor array produces a signature
Class 1
Multidimensional data
exploration
and
visualization techniques
Automated molecule
identification or visual
display
Class 2
Feature
extraction and
pattern
recognition
Artificial neural networks
•
Backpropagation network
•
Radial basis function (RBF)
networks
•
Self-organizing feature
maps (SOFM)
Output
Sensor signal
processing and
calibration
Input
Chemical sensor
array
Odorant
Measurement chamber
FIGURE 5.2
Flow diagram showing the components of an electronic nose that utilizes artificial neural networks for perform-
ing sensor calibration, pattern classification, and multidimensional data visualization.
