DIAGNOSIS OF BREAST CANCER

USING FUZZY EXPERT SYSTEM

NehaSharma(ResearchScholar)
CTgroupofInstitutions,Jalandhar,Punjab,India
Email:Nehu.sharma46@gmail.com
VishalKhanna(ResearchScholar)
CTgroupofInstitutions,Jalandhar,Punjab,India
Email:Vishalkhanna001@gmail.com
Abstract In a way to deal with real heath care
problems, we proposed a fuzzy based decision- making system for
diagnosis of breast cancer. Breast cancer is the most far- reaching
disease today, so primary detection of Breast Cancer is very
significant. The proposed paper was provided with artificial
intelligent techniques such as fuzzy logic to give correct decision
making. The fuzzy rule- based makes use of expert knowledge to
deal with patients symptom and give an accurate decision
according to rules constructed.
Index Terms Fuzzy expert system, Artificial Intelligent,
Breast Cancer, FIS

I. INTRODUCTION
The medical analysis of a disease is most far reaching
problem in todays medical world. The medical area is one of
the recent branches that need engineering techniques to gain
access to imprecise information. With new advances in
medical engineering and other control systems that have been
acquired by the usage of artificial intelligent techniques.
Artificial intelligence has made a progressive research that
includes fuzzy logic, artificial neural networks and genetic
algorithms. All these techniques work altogether and provide
necessary information from one type to another to deal with
challenging real-life problems. The most existing and
widespread type of artificial intelligence that they can provide
help and assistance to medical experts in the identification of
the disease is the development of the clinical diagnosis of
decision support system[5].
The farthest- reaching disease today is the Breast Cancer. It
is commonly found in women causing millions of deaths
yearly. The biggest issue in the diagnosis of breast cancer is
that the main cause or reason behind the disease is unknown.
The patients state of surviving can be improved by early
detection of breast cancer [6].The detection of breast cancer at
the earliest stage can improve the survival rate from 56% to
86%. Normally the diagnosis of the primary stage of the
disease is imprecise, as the physicians diagnose by studying
the history of patient [1].Although, numbers of methods are
used to diagnose breast cancer, such as Fine Needle
Aspiration, Mammography, Biopsy and Magnetic Resonance
Imaging (MRI), but these techniques are not so reliable due to

human errors. The medical experts detect breast cancer based

upon their past experience and knowledge which sometimes
may lead to wrong decisions [7]. Recent studies reported that
for the diagnosis of breast cancer, the fuzzy expert system is
one of the advantageous applications with the capability to
handle unclear and imprecise in prognosis and diagnosis of
breast cancer. Due to their robust behavior in the effect of
noise and uncertainty, yields better results [1].Breast cancer
tumors are classified as: Malignant Breast Tumor and Benign
Breast Tumor. Malignant breast tumors are those that once
appear in any part of the body will spread in the whole body
and is very difficult to remove. Benign breast tumors do not
spread in the whole body and it is easy to detect as compared
to malignant.
Artificial
intelligence in
Breast cancer

Mammo
graphy

Biopsy

FNA

MRI

Figure1: Artificial Intelligence in Breast Cancer

Fuzzy set theory was introduced by Prof. Lofti Zahed in

1965 [9], which make it attainable to define uncertain medical
attributes into human understandable form [5]. The yielding
performance of the fuzzy logic has been used in many various
applications. The biggest benefit of the fuzzy expert system
lies in the fact that researchers can model uncertain, complex
system into simple human understandable form by using
human experience and knowledge as fuzzy rules as set of
linguistic variables [2]. The current paper discussed an expert
system by making use of fuzzy logic to identify breast cancer

tumor from its prescribed symptoms. The accuracy

reasoning is determined using patient data set a record having
5 different attributes. By using medical expert knowledge
fuzzy rules are developed that can be used in decision making.
This present paper proposed knowledge- based expert system
for the diagnosis of breast cancer. The designed system was
provided with data mining artificial intelligent techniques such
as fuzzy logic techniques to give accurate and active decision
system. Detection of breast cancer at a primary stage is very
essential as to increase peoples survival rate.
The medical industry using the field of artificial intelligence
has successfully moved from clinical laboratory to real- time
applications. The fuzzy expert system can be employed as a
decision -making tool for monitoring various diseases using
fuzzy relationships. In this part, designing and implementing
results for the diagnosis of disease breast cancer using fuzzy
inference system are designed. After evaluating the degree of
accuracy, can be attained from a health center data set. Using 5
numbers of input parameters (symptoms) fuzzy inference
systems is developed. The fuzzy rule- based system utilizes
medical expert knowledge for understanding patients
symptoms and give an accurate decision according to fuzzy
rules are constructed.
Crisp
Input
values
Fuzzificati
on

Fuzzy Rule Base

Fuzzy
Inference
Engine

Defuzzificat
ion
Crisp Output
values

Figure2: Fuzzy Expert System

II. RELATED WORK

In this part, we shall discuss some work related to the fuzzy
expert system. There are various works done in a literature that
illustrate the implementation and model of medical experts
system.
Muhic I. (2010) proposed a new approach for diagnosis of
breast cancer using Fuzzy-C-Means clustering and pattern
recognization. The fuzzy-c-means clustering (FCM) algorithm
has been tested on clinical instances obtained from Wisconsin
University. This data set 683clinical instances in which 444

are benign and 239 are malignant breast cancer patients. By

using FCM, the clinical attributes are divided into two
clusters, one with malignant cluster and other with benign
cluster. Authors concluded that this approach yield 100% true
positive,87%true negative [4]. Balancia V. et al. (2011)
proposed a decision support fuzzy expert system for breast
cancer risk. Authors introduced fuzzy rules which can be used
to predict breast cancer risk and fuzzy results are compared
with clinical results. Authors concluded that the defined fuzzy
rules are near agreement with clinical results [2]. A novel
approach for classifying malignant and benign masses using
statistical measures was proposed by Surendiran B. et al.
(2011). Authors proposed a Hue, Saturation and Value (HSV)
weight function for classifying malignant and benign masses
by using statistical measures and 233 mammograms from
Digital Database for Screening Mammogram (DDSM) has
been used. Authors used gray weight value to smoothly
classify malignant and a benign masses. Authors concluded
that statistical measures are very much selective in
differentiating malignant and benign masses [3]. Tintu P. et al.
(2013) proposed an innovative approach to diagnose breast
cancer in Wisconsin Prognostic Breast Cancer (WPBC) data
sets using Fuzzy-c-means. Authors introduced fuzzy c-means
intelligent technique to diagnosing breast cancer and have
been tested to data set from Wisconsin University for
classification of cancer cases. Authors concluded that by using
Fuzzy-c-means technique, with higher efficiency and accuracy
would be capable of identifying cancer cases whether
malignant tumor or benign tumor cases [6]. Nagarajasri B. et
al. (2013) proposed a threshold neuro- fuzzy expert system for
diagnosis of breast cancer. Authors introduced ARM Cortex
M-3 neuro- fuzzy Expert system for diagnosis of breast cancer
for classifying malignant and benign mammography findings.
Authors concluded that this system increases a wide range of
accuracy and efficiency for diagnosis of breast cancer [8].
Khezri R. et al. (2014) proposed a fuzzy expert system for the
risk development of breast cancer prognosis. This model uses
Mamdani fuzzy inference system which has more capability to
interact with humans experts during diagnosis stage. The main
benefit of this implementation is that it can access risk
developing in breast cancer even in normal females. The
results of a fuzzy expert system were compared with
obstetrician decisions. Authors concluded that results of a
fuzzy expert system were 95% accuracy as compare to
previous work. This model is used for predicting breast cancer
risk and hence increases the survival rate of patient [1]. Walia
N. et al. (2015) proposed a decision support system for
diagnosis of tuberculosis using a fuzzy expert system. Authors
briefly explained the implementation between different input
attributes and their symptoms. Authors concluded that
proposed a fuzzy expert system for diagnosis of tuberculosis
give an accuracy of 78% and this system can give aid to
pulmonary physicians [5]. It is clearly understood from above
literature review on fuzzy logic that fuzzy logic has been
successfully applied in the various medical field for diagnosis
and prognosis of various diseases. In this, fuzzy rules are
developed to predict the breast cancer tumor.

and works similar to human beings. Also has an ability to

handle real life applications.

III. PROPOSED SYSTEM

This part explains the approach adopted in constructing the
overall fuzzy framework for decision - making system. The
fuzzy inference system is a framework which is dependent on
fuzzy set theory, gains a fuzzy representation of patients
symptom and accordingly induces fuzzy relationship. In order
to accomplish fuzzy description to fullest i.e. to attain very
high interpretability, the capability to grasp generalization is of
highly significant. The word generalization, we mean that
ability to explicit the state-action agreement as compact as
feasible. Generalization rules grant additional compact rule
base, rapid inference and better fuzzy interpretability. A fuzzy
dependent decision support system attains expert knowledge
and experience and understanding of IF-ELSE rules to design
fuzzy inference. Thus, a fuzzy expert system allows a simple
way for designing an accurate solution with assistance from an
uncertain region. The given fuzzy set corresponding with a
membership function defines the input attribute to its accurate
membership and it must in a range of (0, 1). A fuzzy set is a
set which has no crisp value and has fuzzy borderline.
The trapezoidal membership plot is a function having four
variables a, b, c, d where a and d represent feet of trapezoid
with membership degree 0 and b and c represent shoulders of
trapezoid with membership degree 1 is illustrated by equation:

(1)

Figure 3: Membership Plot for Change in Skin Color

Figure 4 : Membership Plot for Nipple Discharge

IV. FUZZY MODELLING AND WORKING

Fuzzy rules and fuzzy analysis is the keystone to the fuzzy
inference system that converts input variable (crisp value) into
the fuzzy variable for the prediction of actual stage of a
disease.
Fuzzy IF-ELSE rules are the rules constructed using input
(antecedents) variable and output (consequent) variable in the
manner as in table 1.1 IF A (INPUT) THEN D (OUTPUT)
where A and D contain some particular information related to
both input and output parameters. This current paper is
simulation related paper and medical decision - making system
is performed using fuzzy tool box in MATLAB 2013B
software. This proposed system is used to predict the breast
cancer tumor. The system consists of l number of input
variables and output variable taken during diagnosis of breast
cancer. The number of input attribute changes in skin color,
breast lump, nipple discharge, family history and fibro
adenoma. Each input attribute is associated with two or three
trapezoidal membership function
Mamdani inference system is used for diagnosing due to its
ability to illustrate expert knowledge in a progressive manner

Figure 5: Membership Plot for Family History

The figure3 displays the membership function of a change

in the skin color which consists of two membership plots
change in the skin occurs and not occurs. A trapezoidal
membership function is used for this input attribute. Change in
skin occurs having membership parameter [-3.6 -0.4 3.27
6.207] and change in skin color not occurs having membership
parameter [4.02 5.816 7.13 8.6]. The figure 4 displays the
membership plot for nipple discharge which consists of two

membership functions, occurs and not occurs. The occur

membership function uses trapezoidal membership function
and having membership parameter [-3.6 -0.4 0.788 5.02] and
not occurs membership function also uses trapezoidal
membership function and parameter value [2.83 4.6 5.4 8.6].
The figure shows the membership plot for family history
which consists of two membership functions, occurs and not
occurs. The family history occurs having parameter value [-3.6
-0.4 0.4 3.6] and not occurs having parameter value [1.4 4.6
5.4 8.6]
The defuzzification performs the reverse of fuzzification
process. The defuzzification converts the fuzzy output
obtained from inference system into the crisp variable. The
common defuzzification method used is the center of area. It
can be represented as:
=

(2)
Figure 8: Surface Plot

Figure 5 displays the rule viewer of the proposed system. It

indicates the results of entire proposed system. From the left
side at the top we get defuzzified values, we get tumor=5.28
which means person is suffering from malignant tumor
Figure 6 displays the surface plot of change in skin color and
Breast lump. As the graph indicates that as the change in skin
color increases and breast lump is becoming more and more
fixed, there is equally increase in tumor. The graph displays
that patient has 6.2 tumor means the person is suffering from
malignant tumor. Figure 7 displays three dimensional surface
plot between family history and nipple discharge. From the
graph, it is clear that as chances of family history are more,
there will be probably more chances of malignant tumor. As
the graph indicates that the person is having tumor of 5.6 as
the nipple discharge starts increasing and the chances of
family history are more which means that person is suffering
from malignant tumor
V CONCLUSION
Figure 6: Rule Viewer

Breast cancer is most far reaching disease today, so early

detection is very necessary to prevent people suffering from
breast cancer. Early diagnosis is the most high fidelity decision
which provides medical physicians to differentiate malignant
and benign cancer tumors.
In this paper, we have demonstrated a fuzzy framework on
decision support system for the diagnosis of breast cancer
tumor. The proposed fuzzy inference system predicts the
breast cancer tumor. Also, the proposed fuzzy system gives
essential and significant conditions for diagnosis of breast
cancer. The proposed method can deal with various inputs
which can be far better than to handle uncertainty during
diagnosis process. This present system can be extended by
increasing number of inputs.
[1]

Figure 7: Surface Plot

REFERENCES
Khezri R., Hosseini R., Mazinani M., A Fuzzy Rulebased Expert system for the Prognosis of Risk

[2]

[3]

[4]
[5]

[6]

Development of Breast Cancer, International Journal

of Engineering (IJE), vol.27, pp 1557-1564, 2014.
Balancia V.,Dumitrache I., Rae W., Evaluation of
Breast Cancer Risk by using Fuzzy Logic, World
Academy of Science, Engineering and Technology,
vol. 73, pp 53-64,2011
Surendiran B., Vadivel A., Classifying Malignant
and Benign Masses using Statistical Measures,
International Journal on Image and Video Processing,
vol. 02, pp 319-326, 2011.
Muhic I., Fuzzy Analysis of Breast Cancer using
Fuzzy-c-means and Pattern Recognition, Southeast
Europe Journal of Soft Computing, pp 50-55, 2010
Walia N., Singh H., Tiwari S., Sharma A., A
Decision Support System for Tuberculosis
Diagnosability, International Journal on Soft
Computing, vol. 6, pp 1-13,2015.
Tintu P., Paulin R., Detect Breast Cancer using
Fuzzy-c-means Techniques in Wisconsin Prognostic
Breast Cancer (WPBC) Data sets, International
Journal of Computer Applications Technology and
Research, vol.2, pp 614-617, 2013.

[7]

Caramahai M., Severin I., Balan H., Evaluation of

Breast Cancer Risk using Fuzzy Logic, World
Academy of Science and Technology, vol.56,pp3742, 2009.
[8] Nagarajasri B., Padmavathamma M., Threshold
Neuro Fuzzy Expert System for diagnosis of Breast
Cancer, International Journal of Computer
Application, vol.66, pp 6-10, 2013.
[9] Rawat K., Burse K., A Soft Computing Genetic
Neuro-Fuzzy Approach to Data mining and its
Application to Medical Diagnosis, International
Journal of Engineering and Advanced Technology,
vol. 3, pp 409-411, 2013.
[10] National
Breast
Cancer
Organization,
http://www.nationalcancer.org/breastcancer.
[11] Singh A., Gupta B. A Novel Approach for Breast
Cancer Detection and Segmentation in a
Mammogram, Proceedings of Eleventh Multiconference on Information Processing, vol. 54, pp
676-682, 2015.

Table 1.1: Rule based for fuzzy expert system

Then
Rule IF
No.
Change Breast
in Skin Lump
Color
1.
Occurs Fixed

Nipple
Family FibroDischarge History adenoma

Tumor

Malignant

Mobile -

Benign

3.

Not
Occurs
Occurs

Occurs

Malignant

4.

Occurs

Present -

Malignant

5.

Occurs

Absent

Malignant

6.

Benign

Not
occurs
-

Absent

Benign

Present

Benign

9.

Not
occurs
Not
Occurs
Not
Occurs
-

Fixed

Occurs

Malignant

10.

Fixed

Present -

Malignant

11.

Fixed

Absent

Malignant

12.

Benign

13.

Mobile Not
Occurs
Mobile -

Absent

Benign

14.

Mobile -

Present

Benign

2.

7.
8.