Seminar Report

NEURAL NETWORKS
“Investigating the similarities b/w human & artificial neurons”
SWARNANDHRA COLLEGE OF ENGINEERING
AND
TECHNOLOGY
NARSAPUR
WEST GODAVARI-534275
Authors:
Name: VIJAY BHUSHAN KUMAR
IT (3/4),
e-mail: [email protected]
Name: K.SAGAR CHANDRA
IT (3/4),
e-mail: [email protected]
1. Introduction
This paper discusses the recent trends in
user requirements.
BIOMEDICAL ENGINEERING. The rapid

The observation consistency.
progress

Objective and uniform
of
biomedical
instrumentation
allows the production of systems and
components
with
a
low
cost-
to-
measurements;

performance ratio.
provision;
The paper introduces new trends in sensor

technology encompassing the principles of
Fuzzy
Logic
Controllers
and
Laser
operating principles, detailed setup and


in
the
field
of


Continuous measurement of the
atmosphere (each minute up-to-
their applications in:
date observations);
Fuzzy Logic Control of Inspired
Concentration
Higher density of observations
available in real time;
biomedical instrumentation which includes
Oxygen
More frequent special
observations;
overview of the recent advances and new
trends
Better timeliness and data
availability;
measuring various parameters by different
methods. In this paper we will give an
Higher accuracy and quality of
data;
Displacement Sensors. It discusses the
development
High frequency of data
in

Ventilated
Data from Automatic systems
can be integrated more
Newborn Infants
effectively with the data from
other systems;
2. Need to Automate Biomedical

Lesser cost per data piece.
Instrumentation Systems

Automatic systems can work in
3. Latest Trends in Biomedical
different modes depending on
Instrumentation :
Fuzzy Logic Control of Inspired Oxygen
unit (NICU) of Children's Hospital, Boston,
Concentration in Ventilated Newborn
MA.
Infants
3.1.1. Introduction to Fuzzy Logic
3.1 Fuzzy Logic Control of Inspired
Oxygen Concentration in Ventilated
Newborn Infants
Wide range of opportunities is open
in the field of digitally controlled systems by
microcomputer
development.
Digitally
measured information is rapidly processed.
The
control
of
to
Complex systems are easily controlled. The
mechanically ventilated newborn infants is a
diversity of controlling software allows
time intensive process that must balance
more creativity to experts. New approach in
adequate tissue oxygenation against possible
controlling was introduced, using the fuzzy
toxic
exposure.
regulators. The expert, familiar with the
Investigation in computer assisted control of
certain process, easily can implement the
mechanical
solution, by simply writing and changing the
effects
although
of
are
is
very
increasing,
few
code.
Flexibility
in
control
implemented a fuzzy controller for the
software, simply changing the code. New
adjustment of inspired oxygen concentration
methods of control are introduced. In many
(FIO2)
The
different conditions, these prove to be very
by
efficient. A specific application: neural
neonatologists, and operates in real-time. A
networks and fuzzy logic use on system
clinical trial of this controller is currently
control.
ventilated
utilizes
rules
We
program
realization and modifications are easy using
in
infants.
studies
have
controller
newborn
delivery
oxygen
ventilation
there
involving
oxygen
newborns.
produced
taking place in the neonatal intensive care
3.1.2. FLC Designing phases
Fuzzy Logic Controllers (FLC) are
desired value and maintains it, until a
an example of such kind of controlling.
demand for another value appears, given by
Convenient for the systems that are hardly
the user. It is desirable to describe the
modeled.
system mathematically, to simulate the
The fuzzy set of a concept is defined
process using a microcontroller and then to
by a distribution function of the degree of
implement the program. The advantages of
belief (DoB) in a qualitative parameter (the
such systems are to insure stability,
concept), over a range of variation in a
precision and speed.
quantitative or less-qualitative parameter
3.1.3. What is Fuzzy Logic?
(the scale). The concept may be determined
by different scaling parameters and each
parameter on its own is not necessarily
unique.
These are the phases of designing the Fuzzy
regulators:
In the classical theory of groups, we say that
a certain element belongs or does not belong
to the appropriate group.
1. Process analysis
2. Rule determination, by an expert
The experiment - water temperature
3. Simulation of the Fuzzy regulator.
regulation. The water can be hot or cold.
If we do not receive the desired results, we
have to repeat the steps.
The realization of FLC in a digital
Such approach is simplification of real
situation. In this example, the graduations of
temperature are neglected. This approach is
control system consists in source code
thus not precise. In everyday life, we can
writing (we used the C language). Based on
hear terms like, little bit colder, little bit
given temperature, the regulator reaches the
warmer, cold, hot, very hot, etc. One
solution is to introduce terms like : Very
logic controllers are applicable in the
Negative, Little Negative, Around Zero,
problems of temperature regulation.
Little Positive, and Very Positive (VN,
LN, AZ, LP, VP). A value can belong to
different
representations,
this
way
3.1.5. Indepth Analysis
its
description becomes uncertain, fuzzy. For
instance, the fuzzy value F=3.5 is LP with
belonging degree n=0.25 and VP with
n=0.75.
Oxygen
toxicity
plays a role in the
development
chronic
of
lung
disease in newborn
infants requiring mechanical ventilation. In
3.1.4. Advantages of fuzzy logic
• Optimizes the already existing solutions,
premature infants, varying levels of oxygen
exposure are implicated in the development
with the purpose to achieve a more simple
of retinopathy of prematurity. Because of
and efficient end
these effects, control of oxygen delivery to
• Reduces the price of the end product on the
ventilated newborns has become a priority in
bases of simplified procedure of
neonatal intensive care.
•
Clarifies
the
system.
System
is
understandable, easier for maintenance and
Among the many ventilator parameters that
• Higher resistance to errors and changes of
affect patient respiratory status, the inspired
the system
oxygen
• Increases the system robustness without
commonly adjusted on an acute basis to
decreasing its
control oxygen delivery and maintain patient
Thus on the basis of the performed
experiments, we can conclude that fuzzy
concentration
(FIO2)
is
most
oxygen saturation levels. Manual control of
the FIO2, however, may lag the clinical
condition of the patient. That is, a patient
oxygenation at a target level set by the
may have an increased oxygen requirement
physician.
as
demonstrated
by
a
lower
oxygen
saturation, but the manual increase of FIO2
Instead of controlling the ventilator directly,
may be delayed by human response times
the system currently operates by displaying
(i.e. a clinician may not be present to
suggested FIO2 changes to the physician,
respond immediately). Conversely, a patient
who then decides whether to execute the
may have a decreased oxygen requirement
recommended change. This ensures medical
as clinical conditions improve, yet the
safety until the system is fully tested for
amount of oxygen delivered may not be
clinical efficacy. A clinical trial of the FIO2
immediately decreased. The latter scenario
control system is currently taking place in
may be more common because of the
the neonatal intensive care unit (NICU) of
perception that a patient with high oxygen
Children's Hospital, Boston, MA.
saturation is "doing well" and does not
require immediate intervention.
Fuzzy logic controllers
Fuzzy
control
techniques
have
We have designed and implemented a
recently been applied to various medical
microcomputer
processes, such as pain control and blood
based
system
to
automatically and continuously control the
pressure
FIO2 delivered to mechanically ventilated
classical control theory, a fuzzy logic
newborn infants. This system utilizes a
approach to control offers the following
fuzzy logic controller based on "rules"
advantages:
generated by neonatologists who routinely

control.
When
compared
to
It can be used in systems which
provide care for ventilated infants. The goal
cannot
of this control system is to maintain patient
mathematically. In particular,
be
easily
modeled
systems



with
non-linear
understandable controller design
responses that are difficult to
and faster computation for real-
analyze may respond to a fuzzy
time applications.
control approach.
In the context of FIO2 control in the
As a rule-based approach to
newborn infant, a fuzzy logic approach can
control, fuzzy control can be
simplify the many complex factors and
used to efficiently represent an
interactions
expert's knowledge about a
oxygenation. For example, a ventilated
problem.
infant may exhibit decreased oxygen levels
Continuous variables may be
in the blood (as measured by SaO2) for any
represented
linguistic
of the following reasons: failure to make
constructs that are easier to
respiratory effort, a plug in the endotracheal
understand,
the
tube, or an increase in pulmonary shunting.
controller easier to implement
Each cause may require differing changes in
and
FIO2 to maintain target SaO2 levels, and
modify.
by
making
For
instance,

determine
factors
may
patient
instead of using numeric values,
many
temperature may be represented
oxygenation. At different times, the same
as "cold, cool, warm, or hot".
magnitude of change in FIO2 may result in
Fuzzy controllers may be less
completely different oxygenation states,
susceptible to system noise and
even within the same patient.
parameter changes, thus making
other
that
influence
FIO2 control in the newborn thus
them more robust.
demonstrates
some
of
the
previously
Complex processes can often be
mentioned features which make classical
controlled by relatively few
control techniques difficult to apply: the
logic rules, allowing a more
system to be controlled is complex with
many factors and interactions, it is very
difficult to model mathematically, and
standard manual control. In contrast, the
system responses to FIO2 changes are often
fuzzy
non-linear and unpredictable.
implemented within a week, and preliminary
controller
was
designed
and
results of clinical testing show promise that
COMPARISON OF HUMAN,
the controller maintains target oxygenation
CONVENTIONAL AND FUZZY
saturations better than manual control while
reducing oxygen exposure.
CONTROLLERS
3.1.6. SYSTEM DESCRIPTION
Performance index
140
120
100
80
60
40
20
0
Fuzzy
We chose SaO2 as our measurement
parameter
Human
and
FIO2
as
our
control
parameter for the operational model of
maintaining patient oxygenation.
Conventional
1
2
3
4
6 Different FIO2 Levels
5
6
3.1.7. SaO2 Measurement principle
The
pulse
oximeter
measures
arterial percent oxygen saturation (SaO2 )
It is noteworthy that our initial approach to
and pulse rate using the principles of
the problem of FIO2 control involved the
spectrophotometry
construction
of
a
modified
and
adaptive
photoplethysmography.
feedback-loop controller. After many weeks
Light from two light-emitting diodes at two
spent grappling with the previously defined
different wavelengths [typically 660 nm
problems and continually modifying the
(red) and 940 nm (infrared)] passes through
controller, clinical testing on ventilated
the tissue and is sensed by a photodiode. As
infants
revealed
that
our
"classical"
the heart ejects oxygenated blood into the
controller did not perform any better than
tissue, the instrument ignores all absorption
([[Delta]]SaO2), and the slope of SaO2
in the steady-state tissue and measures only
(SaO2-slope) as the specific inputs to the
the absorption in the tissue that is expanded
fuzzy controller.
by the pressure pulse. This expanded
Although many ventilator parameters affect
volume contains arterial oxygenated blood.
patient oxygenation (e.g. mean airway
There
are
transmitter/sensor
pressure, ventilatory rate, tidal volumes,
geometries. In the transmission mode, the
etc.), the FIO2 is used on maintain the
light source and sensor are on opposite sides
desired oxygenation status when the patients
of the tissue being measured (such as a
overall respiratory status has been stabilized.
finger or ear lobe); the light passes through
The design of the fuzzy controller then
the tissue. In the reflective mode, the sensor
follows standard methods, with fuzzification
and light source are on the same body
of the input parameters, construction of
surface (such as the chest), and the light
fuzzy inference rules, and defuzzification or
reflects from the tissue. The finger sensor
calculation of a "crisp" output value that
provided operates as a transmission sensor,
represents the controller's action.
where the multisite sensor can be used either
To fuzzify the input parameters, the values
way.
of [[Delta]]SaO2 and SaO2-slope were
SaO2
two
pulse
divided into fuzzy regions, with 7 regions
oximetry is a well established method of
chosen for [[Delta]]SaO2 and 5 regions
following patient oxygenation status. It's
chosen for the SaO2-slope. Triangular
advantages over direct measurement of
membership functions were assigned to each
blood
region, as illustrated in Figure 1.
oxygen
as
measured
levels
by
include
rapid
equilibrium with changes in blood oxygen
Using the fuzzy input parameters, the
levels, continuous monitoring, and it is
inference rules that form the body of the
noninvasive. We used the error between the
controller were constructed in the standard
patient's
declarative form: IF situation THEN action.
SaO2
and
the
target
SaO2
The combination of 7 [[Delta]]SaO2 fuzzy
values into a final value (a "crisp" value)
regions and 5 SaO2-slope fuzzy regions
that represents the controller output. We
yields 35 rules. The logic of these inference
used the weighted mean of all the rule
rules are based on the expert knowledge of
outputs to produce a single output value, in
the neonatologists. Some example rules
this case a change in the FIO2.
follow:
Although there are relatively few fuzzy
Rule: IF the [[Delta]] SaO2 is small-
inference rules, continuously calculating the
negative
crisp output in real-time may not always be
AND the SaO2-slope is medium-negative
feasible. To help minimize time-delays, we
(situation)
compiled the fuzzy inference rules into a
THEN increase the FIO2 by a medium-
look-up table at runtime. Thus, during actual
positive amount.
fuzzy control operation, evaluating the
(action)
inputs becomes a simple and fast table look-
Rule: IF the [[Delta]]SaO2 is large-negative
up producing the controller output.
AND the SaO2-slope is large-negative
THEN increase the FIO2 by an
The actual operation of the FIO2 controller
extremely-large-positive amount.
is as follows:
Rule: IF the [[Delta]]SaO2 is small-negative
1) SaO2 values are obtained for the patient
AND the SaO2-slope is small-positive
every 1-2 seconds.
THEN do nothing.
2) Every 10 seconds, the [[Delta]]SaO2 and
All 35 rules are summarized in Table 1.
the SaO2-slope are calculated.
For any pair of [[Delta]] SaO2 and SaO2-
[[Delta]]SaO2 = (ave. SaO2 values over last
slope inputs, we apply each of the inference
10seconds)
rules in turn. Each rule will yield an action
- (target SaO2)
value. The defuzzification step then involves
SaO2-slope = least squares regression of
choosing a method to combine all the action
SaO2 values over last 10 seconds
3) The calculated [[Delta]] SaO2 and SaO2-
manual control, and it reduced the overall
slope are used as indices for the compiled
oxygen exposure. Further clinical trials will
fuzzy controller look-up table. A suggested
test the actual clinical efficacy of this FIO2
FIO2 change is returned as the controller
controller, and additional patient data will
output.
allow more fine tuning of the fuzzy control
parameters
FIO2
fuzzy
control
the
shape
of
the
membership functions and the choice of
3.1.8. System Components
The
(e.g.
system
is
fuzzy regions).
implemented on an Apple Macintosh and is
The ease of implementing this fuzzy
programmed in Macintosh Common Lisp.
controller illustrates some of the advantages
The SaO2 data is obtained from a Nellcor N-
of this approach. No complex mathematical
200 pulse oximeter through a RS-232 serial
models were required, the simple rule-based
port on the back of the oximeter.
nature of the controller is easy to understand
and modify, expert knowledge about the
3.1.9. SUMMARY
Controlling oxygen exposure in newborn
infants is a delicate balance. The infants
must receive enough oxygen to ensure
adequate tissue oxygenation and to prevent
ischemia. Conversely, too much oxygen
may produce toxic effects.
The FIO2 fuzzy controller shows promise in
the preliminary trials to control patient
oxygen saturation levels. It was able to
maintain a target SaO2 better than routine
problem is utilized, and the controller was
easily designed
for
non-linear
system
responses.
Current research in fuzzy control include
combining it with other techniques such as
neural networks and genetic algorithms and
adaptive or self-modifying fuzzy control. As
more medical processes become candidates
for computerized control, the numerous
options offered by these approaches will
enhance the ability to produce a safe and
clinically efficacious control system.
Oxygen toxicity plays a role in the
maintain patient oxygenation at a target
development of chronic lung disease in
level set by the physician.
newborn
mechanical
Instead of controlling the ventilator directly,
ventilation. In premature infants, varying
the system currently operates by displaying
levels of oxygen exposure are implicated in
suggested FIO2 changes to the physician,
the
of
who then decides whether to execute the
prematurity. Because of these effects,
recommended change. This ensures medical
control of oxygen delivery to ventilated
safety until the system is fully tested for
newborns has become a priority in neonatal
clinical efficacy.
infants
development
requiring
of
retinopathy
intensive care. Among the many ventilator
parameters that affect patient respiratory
4. Conclusion
status, the inspired oxygen concentration
With each new technology released,
(FIO2) is most commonly adjusted on an
the automation systems gain more capability
acute basis to control oxygen delivery and
and
efficiency.
Automation
inherently
changes with technology because it applies
maintain patient oxygen saturation levels.
the latest advances in different fields to gain
Manual control of the FIO2, however, may
industrial
efficiencies
in
the
user’s
respective industry. In the new and different
lag the clinical condition of the patient.
business environment of the 21st century,
Experts
have
designed
and
the companies that can adapt, innovate and
implemented a microcomputer based system
utilize the latest in automation will generate
to automatically and continuously control
the
FIO2
delivered
to
significant growth and success.
mechanically
ventilated newborn infants. This system
utilizes a fuzzy logic controller based on
"rules" generated by neonatologists who
routinely provide care for ventilated infants.
The goal of this control system is to
5. References

“Fuzzy Sets & Fuzzy Logic”-
George J. Klir.

“Control
Instrumentation
Engineering” by Liptak

“Instrumentation
for
Process
Measurement and Control” by N.A.
Anderson

“Automatic Control Systems” by
Kuo

“Biomedical Instrumentation” by
R.S. Khandpur

“Biomedical Instrumentation” by
Carr & Brown

“Biomedical Instrumentation” by
Crommwell

European magazine for applied
electronics,
automation
&
programming

"Clear Thinking On Fuzzy Logic"
by Lawrence A. Berardinis

"How To Design Fuzzy Logic
Controllers", MACHINE DESIGN

http://www.sesormag.com-March
2001

http://www.bamboowebautomation.
com

http://www.isa.org

http://www.control-systemsprinciples.co.uk