Biomedical Engineering Reference
In-Depth Information
for example, are sensitive not only to blood glucose but also to urea and other elements in
the blood. Fuzzy logic can be used to help compensate for the limitations of sensors.
Fuzzy logic is being used in a variety of biomedical engineering applications. Closed
loop drug delivery systems, which are used to automatically administer drugs to patients,
have been developed by using fuzzy logic. In particular, fuzzy logic may prove valuable in
the development of drug delivery systems for anesthetic administration, since it is difficult
to precisely measure the amount of anesthetic that should be delivered to an individual
patient by using conventional computing methods. Fuzzy logic is also being used to
develop improved neuroprosthetics for paraplegics. Neuroprosthetics for locomotion use
sensors controlled by fuzzy logic systems to electrically stimulate necessary leg muscles
and will ideally enable the paraplegic patient to walk.
EXAMPLE PROBLEM 11.28
A fuzzy system is used to categorize people by heart rates. The system is used to help determine
which patients have normal resting heart rates, bradycardia, or tachycardia. Bradycardia is a cardiac
arrhythmia in which the resting heart rate is less than 60 beats per minute, while tachycardia is
defined as a cardiac arrhythmia in which the resting heart rate is greater than 100 beats per minute.
A normal heart rate is considered to be in the range of 70-80 beats per minute. What are three
linguistic variables that might be used to describe the resting heart rates of the individuals?
Solution
A variety of linguistic variables may be used. The names are important only in that they offer a
good description of the categories and problem. Slow, normal, and fast might be used. Another
possibility is simply bradycardia, normal, and tachycardia.
11.9.2 Artificial Neural Networks
Artificial neural networks (ANN) are the theoretical counterpart of real biological neural
networks. The human brain is a highly sophisticated biological neural network, consisting
of billions of brain cells (i.e., neurons) that are highly interconnected among one another.
Such a highly interconnected architecture of neurons allows for immense computational
power, typically far beyond our most sophisticated computers. The brains of humans,
mammals, and even simple invertebrate organism (e.g., a fly) can easily learn from experi-
ence, recognize relevant sensory signals (e.g., sounds and images), and react to changes in
the organism's environment. Artificial neuronal networks are designed to mimic and
attempt to replicate the function of real brains.
ANNs are simpler than biological neural networks. A sophisticated ANN contains only a
few thousand neurons with several hundred connections. Although simpler than biological
neural networks, the aim of ANNs is to build computer systems that have learning,
generalized processing, and adaptive capabilities resembling those seen in real brains. Arti-
ficial neural networks can learn to recognize certain inputs and to produce a particular
output for a given input. Therefore, artificial neural networks are commonly used for
pattern detection and classification of biosignals.
