ELEC4810: Introduction to

Biosensors and Bioinstrumentation

Lecture Notes – Set #1
 Course Information:

INSTRUCTOR: JIANAN QU
Professor
ECE Department,
Room 2434
Tel: 8541
Email: eequ@ust.hk

OFFICE HOUR: 2:30 am - 4:30pm, Tuesday

or by appointment. (Rm. 2434)

Senior Technician: Bruce Lee (email: eelee@ust.hk)

TA team: Yingzhu He (email: yhebi@connect.ust.hk)

Yiming Fu (email: yfuaq@connect.ust.hk)
2
 ELEC4810: Introduction to
Biosensors and Bioinstrumentation
Lecture Notes – Set #1

Caduceus, Rod of Asclepius

a traditional
symbol of Hermes

ELEC4810

Science and Technology Symbols for the practice of

medicine originated from
ancient Greek mythology
 Course Information:
OBJECTIVES:

• This course introduces electrical and physics students to the concepts underlying the
design of medical electronic devices.
• Biomedical applications from electrical, chemical, and mechanical engineering
perspectives will be discussed in both descriptive and quantitative terms.
• Methods for acquiring and processing signals from the human body are studied mostly at
the system level, but practical examples of the circuit-level design of medical devices are
also presented. The instrumentation of non-invasive devices will be emphasized.
• After completing this course, one should understand the biophysical origins of the
electrocardiogram, the blood pressure pulse, and other major diagnostic variables and be
able to design systems to measure them.

PREREQUISITES: Basic Electronics (Elec 2400 or 2410)

GRADING:
Homework: (15%)
Lab reports: (15%)
Midterm: (20%)
4 Final: (50%)
 Course Information:
COURSE WEBPAGE: HKUST Canvas system (canvas.ust.hk)

TEXTBOOK:
"Medical Instrumentation: Application and Design", 4th edition, ed. by John G.
Webster

REFERENCES:
• “Introduction to Biomedical Equipment Technology”, by Joseph J. Carr and Johyn M.
Brown (4th edition, 2001)
• “Biomedical engineering and instrumentation: basic concepts and applications”, by Joseph.
D. Bronzino
• “Principles of biomedical instrumentation and measurement”, by Richard Aston
• “Biomedical instruments: theory and design”, by Walter Welkowitz
• “Biomedical engineering”, by A. Edward Profio

COURSE LECTURE NOTES

5
 Course Information:
MAIN TOPICS COVERED BY THE COURSE:
• Basic concepts of bio-signal analysis and measurement
• Biopotentials
• Blood pressure and flow measurement
• Respiratory monitoring
• Clinical chemistry
• Integrated biosensors and biochips (?)

COURSE FORMAT:

LECTURES will complement readings from lecture notes, your textbook and class handouts.

HOMEWORK ASSIGNMENTS will be given to provide practice at applying the concepts covered
in lecture. Group discussion of the homework (but not copying) is encouraged. Homework will
be assigned on a Wednesday and will be due the following Thursday. Exceptions, especially closer
to exams, will be announced in class or by e-mail. Absolutely no late homework assignments
will be accepted.

6
 Course Information:
COURSE FORMAT:

SEVEN EXPERIMENTS (main experiments start after mid October) will be assigned
to teams composed of 2~3 students, and each team will be expected to work together to
provide a solution. Grades for the experimental assignment will be based on accuracy of the
results and analysis of the experimental data.

A MIDTERM EXAM (scheduled on October 25, 2019) and a FINAL EXAM will test
both retention of concepts and facts, and the ability to apply problem-solving skills. Material
tested will be extracted from the readings, homework, lectures, and lab projects. The exam
will be closed book and closed notes.

ATTENDANCE at lectures is important for your mastery of the subject material. In

addition, labs require your attendance for credit and important announcements are often
made in class (students are responsible for any changes announced in class).

CHEATING -- Anyone found cheating will be dealt with according to University policy.

7
 Course Information:
COURSE OUTLINE (Syllabus)

I. INTRODUCTION: OVERVIEW OF COURSE

1. Types of medical electronic devices
2. General characteristics of biomedical signals

II. BASIC CONCEPTS

1. Mathematical descriptions of random signals
2. Measurement systems fundamentals
• Transduction
• Signal processing methods
• Close loop systems
• Equipment specifications

III. BIOPOTENTIALS
1. Origin: Ionic currents in a single cell
2. Action potentials (nerve and muscle)
3. Multiple cells: Generation of body-surface potentials
4. Electrocardiography (ECG)
8
 Course Information:
COURSE OUTLINE (Syllabus)

4. Electrocardiography (ECG)
• Physiology of the heart
• Dipole concept
• Lead systems
• Electrodes, amplifiers
• ECG instrument design (First lab assignment)
5. Electromyography (EMG) (Second lab assignment)
6. Electroencephalography (EEG)

IV. BLOOD PRESSURE AND FLOW MEASUREMENT

1. Physiology of the circulatory system
2. Pressure transducers (invasive)
• Strain-gauge (1-,2-,4-arm bridges)
• Inductive
• Capacitive
• Piezoelectric
• Optical
9
 Course Information:
COURSE OUTLINE (Syllabus)

3. Non-invasive pressure monitoring

• Manual cuff
• Oscillometric (Third lab assignment)
4. Flow measurement
• Indicator-dilution method (dye, thermal)
• Electromagnetic
• Doppler ultrasound (Six lab assignment)

V. RESPIRATORY MONITORING
1. Physiology of blood/gas exchange
2. Capnography
3. Oximetry (Fifth lab assignment)

VI. CLINICAL CHEMISTRY

1. Spectrometry
• Bee-Lambert law
• Light sources
10 • Wavelength selection: interference gratings
• Photodetection (Fourth lab assignment)
 Course Information:
COURSE OUTLINE (Syllabus)

2. Electrochemical sensors
• Potentimetric (pH, PCO2 electrodes)
• Amperometric (PO2 electrode)
3. Hematology counting
• Cell counting (Coulter principle)
• Cytometry

VII. INTEGRATED BIOSENSORS AND BIOCHIPS (?)

1. Special requirements of biomedical applications
2. Integrated and smart sensors: up-to-date requirements
3. Novel microarray technology
4. Biochip for clinical diagnostics

11
 Course Information:

SEVEN EXPERIMENTS:

1. Bio-signal exploration
2. Electrocardiography (ECG)
3. Electromyography (EMG)
4. Non-Invasive Blood Pressure Measurement
5. Measurement of the Finger Pulse
6. Pulse Oximeter
7. Advanced ultrasound imaging system

12
 ECE Outcome-based Education Course Syllabus Outline

Course Code: ELEC 4810

Course Title: Introduction to Biosensors and Bioinstrumentation

Course Website: HKUST Canvas (canvas.ust.hk)

Course Schedule: 12:00-13:20 pm on Monday and Wednesday.

Class Room: Zoom

Teaching Staff Contact Details:

Instructor:Professor Jianan Qu
Room 2426, Email: eequ@ust.hk, Tel: 2358-8541
Website: http://www.ece.ust.hk/~eequ
TA(s): WanjieWu (email: wwuas@connect.ust.hk)
Yingzhu He (email: yhebi@connect.ust.hk)

13
14
 ECE Outcome-based Education Course Syllabus Outline
Outcome-based Assessment:
LECTURES will complement readings from lecture notes, your textbook and class handouts.
HOMEWORK ASSIGNMENTS will be given to provide practice at applying the concepts covered in lecture.
Group discussion of the homework (but not copying) is encouraged. Homework will be assigned on a
Monday and will be due the following Tuesday. Exceptions, especially closer to exams, will be announced in
class or by e-mail. Absolutely no late homework assignments will be accepted. (CO1-2,4-5)
SEVEN EXPERIMENTS (main experiments start after mid October) will be assigned to teams
composed of 2~3 students, and each team will be expected to work together to provide a solution. Grades
for the experimental assignment will be based on accuracy of the results and analysis of the experimental
data. (CO3, 5-6)
A MIDTERM EXAM (scheduled in last week of October) and a FINAL EXAM will test both retention
of concepts and facts, and the ability to apply problem-solving skills. Material tested will be extracted from
the readings, homework, lectures, projects, and quizzes. The exam will be closed book and closed
notes. (CO1-2,4-5)
ATTENDANCE at lectures is important for your mastery of the subject material. In addition, in-class quizzes
require your attendance for credit and important announcements are often made in class (students are
responsible for any changes announced in class). Students not in class for quizzes will not receive credit
(there are no make-up quizzes).
CHEATING - Anyone found cheating will be dealt with according to University policy.

15
 ECE Outcome-based Education Course Syllabus Outline

Grading:
Homework (15%)
Lab reports: (20%)
Midterm: (20%)
Final (45%)

Teaching and Learning Activities:

Lectures: Delivered by the instructor on key concepts (CO1-2,4-5)

Labs: Conducted in ECE teaching lab and assisted by a senior technician and TAs. (CO3,
5-6)
Homework assignments / exams:
For students to apply their knowledge of electronic and information technology to solve
biomedical engineering problems (CO1-2,4-5)

16
 ECE Outcome-based Education Course Syllabus Outline
Student Learning Resources:

Lecture notes / handouts / optional readings will be available at the course website.

Textbook
"Medical Instrumentation: Application and Design", 4th edition, ed. by John G. Webster

References
“Introduction to Biomedical Equipment Technology”, by Joseph J. Carr and Johyn M. Brown (4th
edition, 2001)
“Biomedical engineering and instrumentation: basic concepts and applications”, by Joseph. D. Bronzino
“Principles of biomedical instrumentation and measurement”, by Richard Aston
“Biomedical instruments: theory and design”, by Walter Welkowitz
“Biomedical engineering”, by A. Edward Profio

17
 Chapter 1
A Perspective on Biosensing and Bioinstrumentation
What is a tool?
A device or instrument used or worked by hand or machine to carry out a particular function.
Why do we need a tool?

Homo Erectus

The fundamental purpose of tools is to enhance the capabilities of human being.

18
 What is a medical instrument?

• The sensor converts biological signal or information from human body to another
form (usually electrical signal). This biological signal is then processed, interpreted
and displayed for diagnosis purposes.

A few milestones of development of medical devices:

Feeling enhancement:

• A thermometer was invented by

Italian scientist Galileo in about
1596.

• The Clinical thermometer, which

has been universally used in
medicine was developed by Carl
Wunderlich in 1868.

19
 A few milestones of development of medical devices:
Hearing enhancement:

• In 1819, French physician, Rene T.H. Laennec reported his simple hollow tube, a stethoscope
to amplify the sound.

20
 A few milestones of development of medical devices:
Sight enhancement:

• Lens used as magnifiers in the ancient world was systematically studied by Isaac Newton. An
ophthalmoscope was invented in the mid-19th century allows physician examine body cavities,
such as ear, eye and nose, non-invasively.

21
 A few milestones of development of medical devices:
Enhancement of senses of taste and smell (chemical composition):

• In late-18th century, physicians noticed that the urine of diabetics had a sweet taste; the residue
of urine by evaporating can be used for medical diagnosis.
• In 1877, Englishman William Gowers invented a haemocytometer which can counter the
number of red cells in the blood.
• In 1885, the German Hugo von Zeimssen established one of the first clinical chemistry lab.

22
 A few milestones of development of medical devices:
Magnetic resonance imaging (MRI)
In 20th century….

X-ray

DNA Sequencer

Brain Wave

23
 A General Biomedical Measurement Systems

Functions:
• Sense variables which contain information about the physiological status of the patient
• Extract the desired information from the variables
• Display the information

A. Transducer, also called sensor converts energy or information from the measurand to
another form (usually electric). It is frequently the most difficult element to design in a
biomedical instrumentation system because it provides the interface between man and machine.
The signal quality is dependent on the performance of the sensor. Special emphasis will be given
to the details of the signal transduction when we study the specific types of instrumentation
systems in this course.
24
 Examples of transducers / sensors:

Electrodes Strain gauge Photodiode

25
 B. Signal processing
• Methods to separate signals from noise or, in other words, to extract the desired
information by eliminating or suppressing irrelevant information.

Consider a general time-varying random signal: T(t) = S(t) + N(t)

Signal to noise ratio (SNR):

Ratio of the root-mean square magnitudes of the signal and noise.

𝑆𝑆 2 (𝑡𝑡) 1/2
𝑆𝑆𝑆𝑆𝑆𝑆 = ( 2 )
𝑁𝑁 (𝑡𝑡)

How is this separation of signal from noise accomplished?

1.Thresholding
We can try to decide between what is signal and what is noise on the basis of their
relative magnitudes using the following criterion:

𝑆𝑆 𝑡𝑡 , 𝑇𝑇 𝑡𝑡 > 𝑇𝑇0
𝑇𝑇 𝑡𝑡 = �
𝑁𝑁 𝑡𝑡 , 𝑇𝑇 𝑡𝑡 < 𝑇𝑇0

26
 𝑆𝑆 𝑡𝑡 , 𝑇𝑇 𝑡𝑡 > 𝑇𝑇0
𝑇𝑇 𝑡𝑡 = �
𝑁𝑁 𝑡𝑡 , 𝑇𝑇 𝑡𝑡 < 𝑇𝑇0

How to choose 𝑇𝑇0 ?

One possibility is to set the threshold slightly above the rms value of the noise.

𝑇𝑇0 = 𝑘𝑘0 + 𝑇𝑇 𝑡𝑡 ⁄𝑆𝑆𝑆𝑆𝑆𝑆

Where 𝑘𝑘0 equal to a constant offset value with a magnitude slightly greater than 0.

Q: How to determine SNR

A: There is no way of solving this problem in a completely general and satisfactory way.

A common solution: Turn off/on the input. (Is this practical for measurement of bio-
signals from human body?)

• Change the measurement condition. Clearly, if the SNR is large the problem will be
simplified.
• Optimization of design of the sensor.

• Unfortunately, achieving the high SNR is rarely attainable in practice. Additional

27 processing steps are required.
 2. Filtering
- Probably the most common form of signal processing used to improve the SNR.

How a filter discriminates signal from noise is best understood by examining its effects in the
frequency domain.

Let’s plot the Fourier transforms of a time varying random signal T:

𝑆𝑆 𝜔𝜔
N 𝜔𝜔

Power spectrum of a corrupted signal separated into its signal and noise components
𝑆𝑆 𝜔𝜔 and N 𝜔𝜔 . The filter function is represented as Γ 𝜔𝜔

The first thing to recognize is that the spectra of the signal and noise components overlap.
The major concern is the extent of the overlap.

28
 𝑆𝑆 𝜔𝜔 ∞ 1⁄2
N 𝜔𝜔 ∫0 𝑆𝑆 𝜔𝜔 𝑑𝑑𝑑𝑑 𝐴𝐴1
𝐴𝐴1 𝑆𝑆𝑆𝑆𝑆𝑆 = ∞ =
𝐴𝐴2 ∫0 𝑁𝑁 𝜔𝜔 𝑑𝑑𝑑𝑑 𝐴𝐴2

𝐴𝐴′1 𝐴𝐴1
Consider a band pass filter with the frequency response Γ 𝜔𝜔 : >
𝐴𝐴′2 𝐴𝐴2
For an ideal filter with infinitely steep
lower and upper cutoff: A filter function is represented as Γ 𝜔𝜔
1⁄2
∞
∫0 𝑆𝑆 𝜔𝜔 T 𝜔𝜔 𝑑𝑑𝑑𝑑
𝑆𝑆𝑆𝑆𝑆𝑆 = ∞
∫0 𝑁𝑁 𝜔𝜔 T 𝜔𝜔 𝑑𝑑𝑑𝑑 𝐴𝐴𝐴2
𝑙𝑙 𝐴𝐴𝐴1
𝜔𝜔ℎ 1⁄2
∫𝜔𝜔 𝑆𝑆 𝜔𝜔 𝑑𝑑𝑑𝑑 𝐴𝐴1′
𝑙𝑙
= 𝜔𝜔ℎ =
∫𝜔𝜔 𝑁𝑁 𝜔𝜔 𝑑𝑑𝑑𝑑 𝐴𝐴′2
𝑙𝑙

This equation is useful to estimate the improvement in the SNR. However, the challenge is to
know the signal and noise spectrum before designing the filter Γ 𝝎𝝎 .
29
 Table 1.1 at page 10, 11 of the textbook listed the frequency range of major biomedical signals.

30
31
 • Most of signals are ranged in DC to 100 Hz.
• Major noise source from man’s environment: Electrical power lines, fluorescence lights, vibrations and etc.
• To prevent noise from entering the transducer: shielding and inherently insensitive to the noise source.
32
 • Most of signals are ranged in DC to 100 Hz.
• Major noise source from man’s environment: Electrical power lines, fluorescence lights, vibrations and etc.
• To prevent noise from entering the transducer: shielding and inherently insensitive to the noise source.

Simple approaches:
• A grounded metal plate or screen can shield the undesired electrical field;
• A rubber damper could serve as a shield against high-frequency vibration.
• Optical fiber sensor is inherently insensitive to electromagnetic noise radiated from nearby
equipment. It has much better performance than traditional electrodes.

Second line of defense against the entry of noise into transducer:

Differential sensing
• To simultaneously record the difference between two signal channels which are affected equally by
internally and externally noise.
• The signals are usually from the same source and have opposite polarities. Examples: differential
blood pressure transducer, differential skin electrode, and dual-channel thermal sensor.
Modulation
• To modulate measured variable before it is transduced into an electrical signal.
• The objective is to shift the frequency of the measured signal to a new band of frequencies located
outside of the band occupied by the noise source.
33
 C. Interpretation:
A clean, noise-free signal is worthless if the user does not know what it means.

Lead II with notations of electrocardiac activity

Depolarization in:
SH Sinoatrial node
Atria Right atrium and left atrium
AV node Atrioventricular node
H Bundle of His
BB Bundle branches
P Purkinje network

Repolarization in:
a Right atrium and left atrium
v Right ventricle and left ventricle

34
 C. Interpretation:
A clean, noise-free signal is worthless if the user does not know what it means.
How?
• Collecting numerous data and build up reliable data base
• Analyzing the data by using different statistical methods
• Learning from the statistical results, generating new information
• Interpreting
Examples:
The waveform of electrocardiogram (ECG) is full of meaning.
⇒ Pattern analysis, alarm system and etc.
Generate new information from measured variables: imaging reconstruction.

D. Display
• Never underestimate the importance of display the measured result. It determines the
success or failure of the design of a medical instrument.
• Most of modern medical instruments are modified from old models by introducing the
computer graphics.

E. Feedback
• Major function: to stabilize a given variable. Considering the user as part of the system,
almost all the biomedical devices employ some sort of feedback.

35
 A General Biomedical Measurement Systems

Functions:
• Sense variables which contain information about the physiological status of the patient
• Extract the desired information from the variables
• Display the information

36
 IMPORTANT NOTE:

Before designing any instrumentation system, an engineer must have a thorough

understanding of the nature of the signal that he/she wishes to measure. This is
especially true in the design of medical instruments, because of the unique
characteristics of biomedical signals.

What makes biomedical signal unique?

For example, what distinguishes the biomedical signals from those that an engineer would
deal with in the design of a mobile phone or music player?

37
 Biological Signal is Random Signal.

SIGNAL?
A detectable physical quantity or impulse (as a voltage, current, or magnetic field
strength, etc.) by which messages or information can be transmitted.

RANDOM?

Without definite aim, direction, rule, or method.

38
 Signals

𝑉𝑉 𝑡𝑡 = 𝐴𝐴⋅𝑐𝑐𝑐𝑐𝑐𝑐 𝜔𝜔𝜔𝜔 + 𝜑𝜑 + 𝐶𝐶

𝑉𝑉 𝑡𝑡 ≈ 2.0

Random signal:

Without definite aim, direction,

rule, or method.

39
 Why is biological signal Random?

HUMAN BODY TEMPERATURE – an important biological signal

Background:

Mammals such as man are 'warm-blooded' or 'homeothermic' as they can maintain their body
temperature within a narrow range despite fluctuations in environmental temperature. 'Cold-
blooded' or 'poikilothermic' animals such as reptiles cannot and their body temperatures fluctuate
depending on the temperature of their surroundings.

Why do we maintain body temperature?

• Do not need to depend on the external temperature

• Chemical reactions of the body are most efficient as the body temperature rises but above 43
degrees there may be membrane damage or denaturing of proteins, therefore the body
temperature must be kept below this. An oral temperature of 37 degrees represents a safe level
for a large mammal such as man.

40
 How well is our temperature maintained?

Some fluctuations in normal temperatures

both individual variations & variations
within individual
• varies with BMR (age, sex, hormone
level)
• varies with ambient temperature, muscle
actvity, emotional state, menstrual cycle
and state of health
• varies with circadian rhythm (~1 degree
lower in early mornimg than in the
evening for day wakers)

Core temperature or deep body temperature

fluctuates less than does temperature of the
skin and extremeties
(note: blood supply to various parts enables
more even distribution of heat.)

41
 Mechanisms by which heat is gained and lost

Heat gain:
(a) heat input from environment - sun, fire, warm objects etc
(b) heat production by body
• basal metabolism
• ingestion of food
• muscular activity - voluntary (exercise) or involuntary (shivering) - remember
most of the energy used in exercise is wasted as heat (about 70%)
• endocrine mechanisms - thyroid activity ( growth hormone, testosterone)
• emotion/fear (adrenaline)
• non shivering thermogenesis - brown adipose tissue - particularly newborn
Heat Loss
Heat generated in deeper parts of the body is first conducted to the body surface, this
42 depends on blood flow to the skin and on insulation of the body (fat)
 Mechanisms by which heat is gained and lost

Heat transfer to or from environment by four physical processes

• Radiation - transfer by electromagnetic radiation between objects not in contact - may

gain or lose heat - at 21 degrees about 60% of heat is lost by this method
• Conduction - heat exchange by direct contact - gain or loss of heat - rate of conduction
depends on difference in temperature & on thermal conductivity of the objects in
contact
• Convection - transfer of heat from body by moving gas (or liquid) - conduction to gas
first then movement (convection currents)
• Evaporation - evaporation of water from surface of body - insensible water loss about
10 - 50ml/hour - sweating is active evaporative heat loss initiated when temperature of
body rises. Sweating can produce 1700ml/hour.

Heat loss from skin depends on temperature differential, so as the ambient temperature approaches 35
degrees then heat loss by the first three mechanisms decreases and heat loss by evaporation increases and
eventually becomes the only means by which the body can lose heat.

43
 Regulation of Body Temperature

Regulation of body temperature is an example of a biological control system which has

1. Temperature receptors to sense the existing body temperature:

• Peripheral thermoreceptors, cold and warm receptors which are naked nerve endings located in
the dermis. Cold receptors increase their firing rate as the temperature decreases and hot
receptors as it increases, there are more cold receptors than hot receptors. They send
information along the lateral spinothalamic tract via the thalamus to the sensory areas of the
cerebral cortex
• Central thermoreceptors found in deep body areas, including temperature sensitive cells in the
anterior hypothalamus and also the spinal cord and abdominal viscera
44
 Regulation of Body Temperature
2. Integrator in hypothalamus

• Information from both sets of receptors goes to the posterior hypothalamus for
integration, this is the body's thermostat . To prevent heat gain or loss it has a set
point or normal temperature around which it operates. It responds to deviations
from this by initiating both involuntary and voluntary responses.

3. Effect organs to lose or produce heat

• mechanisms that decrease body temperature (mainly from anterior hypothalamus but
some local reflexes when skin temperature is changed)
 vasodilation - reduction of sympathetic tone to all skin blood vessels ->
increase of heat transfer to skin (can be eight times as much)
 sweating - sharp increase in evaporative heat loss when core temp rises above
37 degrees (sympathetic control) -increases by ten times for one degree rise
 inhibition of heat producing mechanisms, reduced voluntary movement
 behavioural - seek shade - cool drink etc

45
 3. Effect organs to lose or produce heat (continued)

• mechanisms that increase body temperature (mainly from posterior hypothalamus)

 vasoconstriction in skin
 piloerection ( not so good in humans)
 increase in heat production
 shivering - involuntary rhythmic muscle activity from hypothalamus
 voluntary muscle activity
 chemical heat production (thermogenesis)
 adrenaline - brown fat (adipose tissue) special tissue with lots of mitochondria -
uncoupled oxidation - heat produced - relatively little in adults, some in neonate
(note: relatively high surface area of neonate -> more dificult heat regulation),
occurs in some animals
 thyroxine - delayed chemical thermogenesis, increase in metabolic rate
 behavioural - curl up

46
 Regulation of Body Temperature - What can go wrong?

Fever (a controlled rise of temperature)

Fever is an elevation of body temperature as a result of infection or

inflammation - the body behaves as if its thermostat has been reset at a higher
temperature. Fever is triggered by the release of pyrogens (heat producing
substances) commonly endotoxins from bacteria or breakdown products
from degenerating tissues. These are phagocytosed by white blood cells
which then release endogenous pyrogens (cytokines such as interleukin 1
(IL1)). These cytokines, which are polypeptides, may act by triggering the
release of prostaglandins (locally produced hormones) which then act on the
hypothalamus to increase the temperature of the 'setpoint'. The new setpoint
is now maintained by heat regulating mechanisms (NOTE: shivering as fever
starts when the body feels it is too cold). Fever may be beneficial, the
increased temp may inhibit bacterial multiplication and antibody production
is increased when the temperature is higher. In the past fevers were induced
for the treatment of neurosyphilis and the growth of some tumors is slowed
by increased temperature. However, it may be dangerous if too high leading
to brain damage; heat stroke occurs when the temperature rises above 43
degrees and death commonly follows.
Fever 'breaks' with sweating and vasodilation when the 'setpoint' goes back to
normal and the body tries to cool down at the end of the illness.
Aspirin reduces fever (antipyretic) by acting on the hypothalamus, probably
by inhibiting the synthesis of prostaglandins.
47
 Regulation of Body Temperature - What can go wrong?

Hyperthermia (an uncontrolled rise of temperature)

• Any increase in body temperature above the normal range ie from sustained exercise, hot
humid weather
• Core temperature stabilises above normal temperature due to increased heat production
• Heat stroke may result if rate of heat gain exceeds rate of heat loss and body temperature
continues to rise
• Critical temperature above 42 degrees
• Dizziness, delerium, circulatory shock from loss of water and electrolytes
• fatal if not cooled down
• Heat exhaustion - less severe - fainting due to decreased blood pressure resulting from
excess sweating which reduces blood volume
• Malignant hyperthermia - abnormal rise in body temperature resulting from a gene
mutation that causes a great increase in heat production during muscle contraction
• Reduction in body temperature (eg: immersion in ice water fatal after 20-30min)
hypothalamus can no longer regulate temp if core
• Temp falls too low - depressed metabolic rate, drowsiness, coma frost bite - freezing
causing tissue damage (cold vasodilation for protection)
• Hybernation
• Induced hypothermia for surgery - allowing the circulation to be stopped because the
48 oxygen needs of the tissues are reduced
 BIOMEDICAL SIGNALS

Classification of signals:
SIGNAL

DETERMINISTIC RANDOM

NON-
PERIODIC NON-PERIODIC STATIONARY
STATIONARY

COMPLEX ALMOST NON- SPECIAL TYPES

SINUSOIDAL TRANSIENT ERGODIC
PERIODIC PERIODIC ERGODIC NON STATION

Deterministic signals can be described by explicit mathematical relationships.

• Periodic and "quasi-periodic"
=> Fundamental frequency, its harmonics or discrete frequency components
• Transient
=> Continuous frequency distribution

Random signals can not be exactly expressed and much more difficult to deal with.
49
 BIOMEDICAL SIGNALS

Classification of signals:
SIGNAL

DETERMINISTIC RANDOM

NON-
PERIODIC NON-PERIODIC STATIONARY
STATIONARY

COMPLEX ALMOST NON- SPECIAL TYPES

SINUSOIDAL TRANSIENT ERGODIC
PERIODIC PERIODIC ERGODIC NON STATION

Random signals can not be exactly expressed and much more difficult to deal with.

A random signal is a sample function or a random process which differs from

another in their time description. However, they possess the same statistical
properties.

A stationary process has the statistical property of which is not a function of time. We
can calculate the expectation by averaging the value of desired variable over a set of
sample function or ensemble.
50
 BIOMEDICAL SIGNALS

A random signal is a sample function or a random process which differs from

another in their time description. However, they possess the same statistical
properties.
Heart Rate Recording

51
 BIOMEDICAL SIGNALS

• A stationary process has the statistical property of which is not a function of time.
We can calculate the expectation by averaging the value of desired variable over
a set of sample function or ensemble.
• A non-stationary process is very difficult to process. Very often, we have to assume
the process is ergodic even the assumption is false.
• A common approach is to divide the non-stationary signal to many segments. In
each segment, the signal is assume as a stationary.

Examples of common biomedical signals: (a) pulse,

(b) electrocardiogram (ECG), (c)
electroenencephalogram (EEG), (d) heart sound

Figure 3 A set(ensemble) of samples of a random signal

showing its behavior with different time intervals

52
 BIOMEDICAL SIGNALS

Unique characteristics of biomedical signals:

1. Unpredictable (Random or stochastic)
- biological variability
- adjustments to environment
- interacting feedback "noise" or "kicks"
2. Easily influenced by measurement process
- psychological/physiological reactions
- energy shunting
- physical damage
3. Remote Sensed
- Need for non-invasiveness
- Propagation over long distance causes attenuation, distortion, and
intermingling
- Interfacing and boundary condition

Probability theory plays important role to mathematically

describe and analyze the biomedical signals.
53
 Probability theory plays important role to mathematically
describe and analyze the biomedical signals.

The mean or expected value

1T
S (t ) = lim ∫ S (t )dt → Central tendency
T →∞ T 0

𝐴𝐴 𝑐𝑐𝑐𝑐𝑐𝑐 𝜔𝜔𝜔𝜔 − 1
e.g. S (t ) = A sin (ωt ) 𝑆𝑆 𝑡𝑡 = lim
𝑇𝑇→∞ 𝑇𝑇
=0
𝜔𝜔

54
 Probability theory plays important role to mathematically
describe and analyze the biomedical signals.

The synchronous averaging:

Suppose that we have a quasi-periodic signal composed of k wavelets Wi(t), such
ECG signal:

The most representative wavelet (called a “template”) can be found by averaging:

1 k
WT (t ) = ∑ Wi (t )
k i =1

55
 Different signals
with the same
mean value.

T
The mean-square value:
S (t ) = lim ∫ S 2 (t )dt
2 1
• measure of signal power T →∞ T
0

The variance:
∫ [S (t ) − S (t )] dt
T 2
1
• measure of power in fluctuations about mean σ 2 = lim
T →∞ T
0

𝑆𝑆 2 𝑡𝑡
e.g. S (t ) = A sin (ωt )

S 2 (t ) = σ 2 = 0.5
σ = 0.707 - rms value

56
 The probability density function (PDF)
describes the probability that a value of a random variable lies in a range between x and x+dx.

Pr ( x < X < x + dx ) = f ( x )dx

- Infinitesimal interval
x2
P ( x1 < X < x2 ) = ∫ f ( x )dx
x1

- Interval [𝑥𝑥1 , 𝑥𝑥2 ] 𝑥𝑥1 𝑥𝑥2

+∞
Normalized as: ∫ f ( x )dx = 1
−∞

+∞

( )
∞ 2

X= ∫ xf (x )dx
−∞
σ 2 = ∫ x − X f (x )dx
∞

Can be continuous or discontinuous

57
 +∞
∫ f ( x )dx = 1
Examples of common PDFs −∞

Uniform:

 1
a< x<b
f (x ) = b − a
,
 0 , otherwise

( ) (b − a)
∞ 2
+∞ 2
a+b
X = ∫ xf ( x )dx = σ2 = ∫ x− X f ( x )dx =
−∞ 2 12
∞

Example: phase of sound waves back-scattered

from body in ultrasound imaging:
Phase: 0 ≤ 𝜙𝜙 ≤ 2𝜋𝜋

58
 Gaussian or normal distribution:

1  (x − x )2 
f (x ) = exp  − 2 
2π σ  2σ 

Central Limit Theorem:

Y= ∑ xn (xn is an independent random variables)

K → l arg e

PDF of Y ⇒ Gaussian, regardless of PDF of x

Distribution of single serum cholesterol

measurements over samples of size n
Example

Serum cholesterol is an important risk

factor for cardiovascular disease.

59
 The crosscorrelation and autocorrelation functions:
• Crosscorrelation function compares two different signals in order to assess the similarity
of them
• Autocorrelation function describes how similar a signal is to itself when measured at a
different time or location

For a stationary signals S1(t), S2(t)

T
Γ1, 2 (τ ) = lim ∫ S1 (t )S 2 (t + τ )dt
Crosscorrelation function:
1
T →∞ T
0

If two signal are similar or the same

T
Γ(τ ) = lim ∫ S (t )S (t + τ )dt
1
Autocorrelation function:
T →∞ T
0

Some properties:

60
 The Power Spectral Density or “Power Spectrum”:
• describes the distribution of power in different frequency bands.
• useful for random, as well as deterministic signals, but must be defined as
an average of samples of finite, but sufficiently long duration.

The power spectral density is defined as:

1 2
𝜙𝜙 𝜔𝜔 = lim 𝐹𝐹(𝜔𝜔)
𝑇𝑇→∞ 2𝑇𝑇

𝑇𝑇

Where: 𝐹𝐹 𝜔𝜔 = ℱ 𝑆𝑆(𝑡𝑡) = lim � 𝑆𝑆 𝑡𝑡 exp −𝑗𝑗𝜔𝜔𝑡𝑡 𝑑𝑑𝑑𝑑

𝑇𝑇→∞
−𝑇𝑇

is the Fourier transform of a single sample of the signal of duration 2T.

Some properties: No phase information 𝜙𝜙 𝜔𝜔 = ℱ ℜ(𝑡𝑡)

61
 The Power Spectral Density or “Power Spectrum”:
• describes the distribution of power in different frequency bands.
• useful for random, as well as deterministic signals, but must be defined as
an average of samples of finite, but sufficiently long duration.

Examples:

62
 The Power Spectral Density or “Power Spectrum”:
Examples:
Beat-to-beat heart rate variation (HRV)

Demonstration of the distinct frequency

bands. Upper panels, subject is supine;
lower panels, subject is standing. In the
panels on the left, the time records are
given, and in the right panels, the
corresponding power spectra are given. In
all panels, the subject is breathing
sinusoidally at 0.25 Hz. The peak at 0.25
Hz(the frequency of respiration) is larger
in the supine than standing position, and
the peak at 0.10 Hz is smaller in the
supine than standing position. The
ordinate in this figure is linear power
spectral density. All other figures are log
power.

63
 The Power Spectral Density or “Power Spectrum”:
Examples:
Power spectrum of heart rate variability in
three individuals. In a normal subject( uppermost
trace), and one each from Group II (non-SCD
group, middle trace), and Group III (SCD group,
lowest trace)
SCD – sudden cardiac death
64
Power spectrum of heart rate variability averages
for each of three groups. Group I (top curve) six
normal volunteers without evidence of heart disease.
Group II (middle curve) six patients with cardiac
disease who are noninducible upon endocardial
programmed electrical stimulation (PES), and who did
not have a history of SCD. Group III (bottom curve) six
patients with cardiac disease who 1) have had a
documented out of hospital sudden cardiac death
initiated by ventricular fibrillation and who were
resuscitated, and 2) were found to be inducible by PES.
 2. Filtering
- Probably the most common form of signal processing used to improve the SNR.

How a filter discriminates signal from noise is best understood by examining its effects in the
frequency domain.

Let’s plot the Fourier transforms of a time varying random signal T:

𝑆𝑆 𝜔𝜔
N 𝜔𝜔

Power spectrum of a corrupted signal separated into its signal and noise components
𝑆𝑆 𝜔𝜔 and N 𝜔𝜔 . The filter function is represented as Γ 𝜔𝜔

The first thing to recognize is that the spectra of the signal and noise components overlap.
The major concern is the extent of the overlap.

65
 𝑆𝑆 𝜔𝜔 ∞ 1⁄2
N 𝜔𝜔 ∫0 𝑆𝑆 𝜔𝜔 𝑑𝑑𝑑𝑑 𝐴𝐴1
𝐴𝐴1 𝑆𝑆𝑆𝑆𝑆𝑆 = ∞ =
𝐴𝐴2 ∫0 𝑁𝑁 𝜔𝜔 𝑑𝑑𝑑𝑑 𝐴𝐴2

𝐴𝐴′1 𝐴𝐴1
Consider a band pass filter with the frequency response Γ 𝜔𝜔 : >
𝐴𝐴′2 𝐴𝐴2
For an ideal filter with infinitely steep
lower and upper cutoff: A filter function is represented as Γ 𝜔𝜔
1⁄2
∞
∫0 𝑆𝑆 𝜔𝜔 T 𝜔𝜔 𝑑𝑑𝑑𝑑
𝑆𝑆𝑆𝑆𝑆𝑆 = ∞
∫0 𝑁𝑁 𝜔𝜔 T 𝜔𝜔 𝑑𝑑𝑑𝑑 𝐴𝐴𝐴2
𝑙𝑙 𝐴𝐴𝐴1
𝜔𝜔ℎ 1⁄2
∫𝜔𝜔 𝑆𝑆 𝜔𝜔 𝑑𝑑𝑑𝑑 𝐴𝐴1′
𝑙𝑙
= 𝜔𝜔ℎ =
∫𝜔𝜔 𝑁𝑁 𝜔𝜔 𝑑𝑑𝑑𝑑 𝐴𝐴′2
𝑙𝑙

This equation is useful to estimate the improvement in the SNR. However, the challenge is to
know the signal and noise spectrum before designing the filter Γ 𝝎𝝎 .
66
67
Summary:

Basic statistical properties of importance for describing single stationary

random records are
 Graphic illustration:

Four special time histories:

(a) Sine wave.

(b) Sine wave plus random noise.

(c) Narrow-band random noise.

(d) Wide-band random noise.

68
 Graphic illustration:

Probaility density function plots:

(a) Sine wave.

(b) Sine wave plus random noise.

(c) Narrow-band random noise.

(d) Wide-band random noise.

69
 Graphic illustration:

Autocorrelation function plots:

(a) Sine wave.

(b) Sine wave plus random noise.

(c) Narrow-band random noise.

(d) Wide-band random noise.

70
 Graphic illustration:

Autospectral density function plots:

(a) Sine wave.

(b) Sine wave plus random noise.

(c) Narrow-band random noise.

(d) Wide-band random noise.

71