TABLE OF CONTENTS
Title page ………………………………………………………………………………………………………………… i
Approval Page ………………………………………………………………………………………………………… ii
Certification ……………………………………………………………………………………………………………. iii
Declaration ……………………………………………………………………………………………………………… iv
Dedication ……………………………………………………………………………………………………………… v
9
Acknowledgment ……………………………………………………………………………………………………. vi
Abstract …………………………………………………………………………………………………………………. vii
Table of Contents ……………………………………………………………………………………………………. viii
List of Figures ………………………………………………………………………………………………………… xiii
List of Tables …………………………………………………………………………………………………………… xvii
List of Plates ……………………………………………………………………………………………………………. xix
List of Acronyms ……………………………………………………………………………………………………… xx
CHAPTER ONE: INTRODUCTION…………………………………………………………………………. 1
1.1 Background Information………………………………………………………………………. 1
1.2 Statement of Problem ………………………………………………………………………. 3
1.3 Aims and Objectives………………………………………………………….…………….. 3
1.4 Significance of the Study…………………………………………………………………….. 4
1.5 Scope of Study…………………………………………………………………………………. 4
1.6 Thesis Outline…………………………………………………………………………… 5
CHAPTER TWO: LITERATURE REVIEW…………………………….………………. 6
2.1 Cardiotocogram (CTG)…………………………………………………………………….. 6
2.2 Cardiotocogram (CTG) Monitoring……………………………………………………. 8
2.2.1 Uncertainties and Imprecision in Fetal Heart
Rate and Cardiotocography (CTG) Analysis………………………………………… 8
2.2.2 Improvement in Cardiotocography ……………………………………………………… 10
2.2.3 Uncertainty in the Real World……………………………………………………………. 11
2.2.3.1 Sources of Uncertainty……………………………………………………………………….. 11
2.2.3.2 Uncertainty in Data……………………………………………………………………………. 12
2.2.3.3 Uncertainty in Knowledge………………………………………………………………….. 12
2.2.3.4 Uncertainty Handling in Expert Systems………………………………………………. 12
10
2.2.4 Uncertainty and Imprecision in Management Labour…………………………… 13
2.2.5 Stan Clinical Guidelines for Action with a Mature
Foetus ³ 36 weeks …………………………………………………………………………… 14
2.3 Electronic Fetal Electrocardiogram (ECG) Monitoring………………………….. 15
2.3.1 QRS Detection in Electrocardiogram…………………………………………………… 16
2.3.1.1 Pre-Processing: the Signal Enhancement Scheme for QRS Detection……. 17
2.3.1.2 Non-linear Prediction Filter…………………………………………………………………. 17
2.3.1.3 Linear Filtering Design by Least Squares Approximation ……………………… 19
2.3.1.4 Comparison of Pre-Processing Techniques………………………………………….. 21
2.4 The Fuzzy Logic System …………………………………………………………………… 22
2.4.1 Fuzzy Set Operators………………………………………………………………………….. 23
2.4.2 Fuzzy Models for Cardiotocogram and Electrocardiogram Analysis……. 24
2.4.3 Fuzzy State Model for Managing Complex Patterns in Time ………………. 26
2.4.4 Adding Memory……………………………………………………………………………….. 27
2.4.5 Using State Machine to Add Memory to Intelligent Systems……………….. 29
2.5 Review of Related Work …………………………………………………………………… 32
CHAPTER THREE: METHODOLOGY…………………………………………………………………… 40
3.1 Materials…………………………………………………………………………………………… 40
3.1.1 Data Collection………………………………………………………………………………….. 40
3.1.2 Cardiotocogram Machine……………………………………………………………………… 40
3.1.3 Ultrasound Machine………………………………………………………………………….. 44
3.1.4 Experiment Conducted for a Patient Whose
Foetus was with Distress……………………………………………………………………. 48
3.1.5 Interfacing Maternal Heart Rate (MHR) with
Fetal Heart Rate (FHR) Patterns ………………………………………………………… 50
3.2 Methods…………………………………………………………………………………………. 55
11
3.3 Fetal Condition Matrix (FCM) …………………………………………………………. 56
3.4 Design of the Fuzzy Model for CTG Analysis……………………………………. 57
3.4.1 Fuzzy Logic-Based (FL-B) System Model Design………………………………. 59
4.2 Membership Functions for the Linguistic Variables……………………………….. 61
3.4.3 Fuzzy Inference System (FIS) Editor……………………………………………………. 62
3.4.4 Membership Function (MF) Editor……………………………………………………… 63
3.4.5 Output Membership Function…………………………………………………………….. 66
3.4.6 Rule Editor………………………………………………………………………………………. 66
3.4.7 Rule Viewer…………………………………………………………………………………….. 67
CHAPTER FOUR: RESULTS AND DISCUSSION…………………………………………………… 74
4.1 Simulation of Results…………………………………………………………………………. 74
4.1.1 Determination of the Optimization Condition………………………………………. 75
4.1.2 Analysis of the Optimization Condition………………………………………………. 77
4.2 Results Obtained from Cardiotocogram Machine…………………………………. 77
4.3 Results Obtained from Ultrasound Machine………………………………………. 78
4.3.1 Comparison of the Accuracy of Cardiotocogram and
Ultrasound Machines… ……………………………………………………………………. 78
4.3.2 Determination of the Accuracy of the Measured Values ……………………… 79
4.3.3 Accuracy of the Measured Values using Cardiotocogram
Machine and Fuzzy Inference System…………………………………………………. 79
4.3.4 Determination of the Degree of Uncertainty between
Cardiotocogram Machine and Fuzzy Inference System…………………………. 80
4.3.5 Accuracy of the Reviewer’s Values using Cardiotocogram
Machine and Fuzzy Inference System …………………………………………………… 81
4.3.6 Determination of the Degree of Uncertainty of the Reviewer’s Values
between Cardiotocogram Machine and Fuzzy Inference System…………. .. 82
4.3.7 Comparison of the Primary Data and Secondary Data…………………………… 83
4.3.8 Graphically Representation of both the Primary Data and
Secondary Data………………………………………………………………………………… 84
12
4.4 Determination of the Surface Plot…………………………………………………………. 85
4.4.1 Analysis of the Surface Plot………………………………………………………………. 89
CHAPTER FIVE: CONCLUSION AND RECOMMENDATION FOR FUTURE WORK. 91
5.1 Conclusion………………………………………………………………………………………. 91
5.1.1 Fuzzy Inference System for Cardiotocogram Machine…………………………. 91
5.2 Recommendation for Future Work……………………………………………………… 92
5.2.1 List of Publications…………………………………………………………………………… 93
References
Appendix 1
Appendix II
Appendix III
CHAPTER ONE
INTRODUCTION
1.1 BACKGROUND INFORMATION
Childbirth is a critical period for the foetus and the mother. A good outcome of child labour is
generally desired but sometimes problems occur that may lead to injury like fetal brain damage,
other abnormalities or even death. Electronic fetal monitoring, introduced by E. J. Quiligan [1]
was expected to improve the patient care, but this has not yet happened. The most common
monitoring method is based on a continuous trace of the fetal heart rate pattern and maternal
contractions, known as the cardiotocogram (CTG). Difficulties in the interpretation of the
cardiotocogram have led to unnecessary medical intervention and failure to intervene when
necessary may lead to injuries and deaths. These problems have led to the development of a
number of computerized systems to assist with the analysis and interpretation of CTG data.
However, despite developments over two decades, there is no significant improvement in fetal
outcomes. The progress in computerized cardiotocogram analysis has been impeded by several
factors. There are inherent problems of imprecision and uncertainty in the clinical data and the
interpretation methods used. The solutions to this problem are yet to be addressed in
computerized cardiotocogram system. Cardiotocogram does not contain sufficient information
accurate for assessment of the fetal condition. Additional information may be obtained by a
proper analysis of changes in the fetal electrocardiogram (ECG), but the problems of uncertainty
and imprecision also exist in fetal electrocardiogram analysis [6].
A major difficulty is the presence of many natural sources of stress on the foetus during child
labour, such as the squeezing effect of uterus due to contraction which causes a temporary
reduction in blood supply and oxygen deficiency. The foetus has many protective mechanisms
that enable it to cope well with stressful events during child labour such as redirecting valuable
blood flow from the non-vital organs to the vital organs, adjusting its metabolic needs and
utilizing stored energy to maintain its heart function (anaerobic metabolism) [2]. Fetal distress
occurs when a foetus can no longer compensate for the effects of stress in labour, which can
22
quickly lead to fetal injury. If continuous information on the fetal condition is available then the
clinician is in a position to improve on the level of care by making more accurate and timely
decisions. Thus the clinician could respond to appropriate signs of fetal distress to prevent fetal
injury and at the same time avoid unnecessary intervention.
The normal fetal heart rate (FHR) pattern is characterized by a baseline frequency between 110
and 159 beats per minute, presence of periodic accelerations, a normal heart rate variability with
a bandwidth between 5 and 25 beats per minute and the absence of decelerations [3].
The FHR pattern is abnormal when the following features are observed. These are the baseline
frequency below 110 or above 160beats per minute, absence of accelerations for more than 45
minutes, absence of FHR variability and late decelerations. A baseline frequency between 100
and 110 can be considered as normal when the duration of pregnancy has exceeded 41 weeks.
A normal foetus is well adapted to the intermittent hypoxia (low oxygen level) by switching to
periods of anaerobic metabolism and redistributes blood to vital organs such as the brain and
myocardium. During this period, the fetal buffering system cannot maintain a prolonged normal
pH due to increase in organic acids produced by anaerobic metabolism, and these results in the
development of a metabolic acidosis.
The assessment of the fetal condition depends on the growth of the uterus and its contents, the
movements of the foetus perceived by the mother and the listening of the fetal heart beat with a
stethoscope. Absence of fetal movements during pregnancy is a serious diagnostic problem. The
decision to assist the delivery of the baby by artificial means depends on information gathered
through the application of cardiotocography [4].
A cardiotocography is a simultaneous record of the FHR and magnitude of uterine contractions.
As uterine contractions can impose stress on the foetus, the relationship between uterine activity
and FHR can provide information about the condition of foetus. The maternal uterine
contractions are recorded using an abdominal wall pressure sensor, whereas the FHR is recorded
from the maternal abdomen wall by an ultrasound transducer.
23
The aim of this research work is to develop reliable fuzzy logic system for Cardiotocography
based monitoring, to reduce the incidence of unnecessary medical intervention and fetal injury
during child labour, due to a high degree of uncertainty and imprecision in obstetric data and
knowledge.
1.2 STATEMENT OF PROBLEM
The outcome of labour is usually good for the foetus, but sometimes problems occur that can
result in permanent fetal brain damage or even death. Cardiotocogram interpretation is a difficult
task and this requires clinical experience and significant expertise. Difficulties in the
interventions and failure to intervene when necessary can lead to preventable injuries and deaths.
The computerized CTG analysis has been impeded by the inherent problems of imprecision and
uncertainty in the clinical data and the interpretation methods used. Also, the CTG does not
contain sufficient information for accurate assessment of the fetal condition.
The large amount of information produced by the fetal CTG requires continuous and vigilant
monitoring but this is not always practical in a busy labour ward. Continuous monitoring is a
very intensive process which would lead to fatigue and eventually mistakes. Computer assistance
would help to provide constant monitoring.
Therefore, the need to develop a reliable fuzzy logic system capable of reducing the incidence of
unnecessary medical intervention and fetal injury during child labour becomes imperative for
this research work.
1.3 AIMS AND OBJECTIVES
The aim of this thesis is to design a fuzzy logic model for Cardiotocogram analysis to handle the
imprecision and uncertainty in the clinical data and knowledge during labour. The main
objectives include the following:
· Determine and extract key robust features from the cardiotocogram waveforms.
24
· Develop a technique that will make the fuzzy model less sensitive to noise and provide a
higher level of accuracy in a precise manner.
· To validate the findings and improve on the outcome between the system and human
experts.
· Obtain a variety of data including normal and abnormal cases.
· Compare the Cardiotocogram and Ultrasound measurements obtained from the same
samples.
1.4 SIGNIFICANCE OF THE STUDY
Enhancing the accuracy of the Cardiotocogram machine will reduce neonatal mortality rate. The
reason is that with abnormal Cardiotocogram readings, immediate delivery of the foetus will be
done. Again, if the accuracy is made more precise and reassuring Cardiotocogram is obtained, it
will reduce the risk of intervention like Caesarean Section (CS).
When the accuracy of the Cardiotocogram machine is enhanced, it will enable the cross-channel
verification monitoring facility to compare all fetal and maternal heart rates and ensure that each
rate is measured separately.
The designed fuzzy model has improved performance in Cardiotocogram analysis by handling
the imprecision and uncertainty in the clinical data and has been able to enhance the accuracy of
Cardiotocogram analysis.
Therefore, improvement in the efficiency of Cardiotocogram will help clinicians to adequately
monitor labour and intervene when necessary.
1.5 SCOPE OF STUDY
The scope of this thesis will be limited to the application of fuzzy logic system to the
development of fuzzy model that enhances the accuracy of Cardiotocogram analysis.
25
Finally, the model will be implemented by developing MATLAB simulation used for the design
of fuzzy logic system to determine the optimality conditions of the fuzzy system and to indicate
the machine that provided lower classification of errors.
1.6 THESIS OUTLINE
This thesis is organized into five chapters. Chapter 1 presents the introduction and background
information of the study. Chapter 2 presents the literature review of Cardiotocogram and fuzzy
logic concepts. Related research work based on Cardiotocogram and fuzzy logic system is also
presented.
In Chapter 3, the methodology and processes leading to the development of the Fuzzy Inference
System (FIS) model are presented. It also describes in detail the simulation carried out in fuzzy
logic toolbox in MATLAB to compute the accuracy from the measured Cardiotocogram features.
Chapter 4 presents the results and discussions of simulation, the accuracy of the measured values
using Cardiotocogram and Fuzzy Inference System. Finally, in Chapter 5, we give conclusion
and recommendation for future work.
Do you need help? Talk to us right now: (+234) 08060082010, 08107932631 (Call/WhatsApp). Email: [email protected].
IF YOU CAN'T FIND YOUR TOPIC, CLICK HERE TO HIRE A WRITER»
