new image processing techniques using elitism immigrants multiple objective of genetic algorithms for disease detection_4

New Image Processing Techniques Using Elitism
Immigrants Multiple Objective of Genetic
Algorithms for Disease Detection
Khalil Ibrahm Mohammed Abuzanounneh
IT Department, College of Computer, Qassim University Saudi Arabia
ka.abuzanouneh@qu.edu.sa
Abstract— Image processing and analyzing images in the
medical field is very important, this research diagnoses and
describes developing of diseases at an earlier stage, a detection
of diseases types by using microscopic images of blood
samples. Analyzing through images changing is very important,
the main objective is completed by analyzing evolutionary
computation into its component parts, using elitism immigrants
multiple objectives of genetic algorithms (EIMOGAs), artificial
intelligence system, evolution methodologies and strategies,
evolutionary algorithm. EIOMGAs are the type of Soft
Computing a model of machine intelligence to derive its
behavior from the processes of evolution in nature [1].
The goal of applying EIOMGAs is to enhance the quality of the
images by applying the image converting process segmentation
to get the best image quality to be very easy to analyze the
images. EIOMGAs are the unbiased estimator for optimization
technique, and more effective in image segmentation, and it is
the powerful optimization technique especially in a large
solution space to implement enhancement process. The
powerful of EIOMGAs system in image processing and other
fields leads to increase popularity and increasingly in different
areas of images processing and analyzing for solving the
complex problems. The main task of EIOMGAs is to enhance
the quality of the image and get required image recognition to
achieve better results, faster processing and implement a
specialized system to introduce different approaches based on
GAs with image processing to obtain good quality and natural
contrast of images [2].The development with comparisons used
between the different techniques of representation and fitness
analysis, mutation, recombination, and selection, evolutionary
computation is shown to be an optimization search tools. All
features of microscopic samples images and examines change in
geometry, texture, colors and statistical analysis will be applied
and implemented in this system.
Index Terms— Elitism Immigrants Multiple Objective,
Microarray Image Processing, Data Mining, Digital image
processing.
I. INTRODUCTION
Image processing is a section of artificial intelligence
concerned with the enhancement, and analysis of images
performed by a computer, and it has become the most
important visualization and interpretation methods in
biology and medical fields. It has a development of new
and powerful tools for analyzing, detecting, transmitting,
storing, and displaying medical images, the medical
images is challenging to found the development
integrated systems, design, implementation, and
successful testing of complex medical systems using in
the medical aspect, the analyzing process through images
is to collect information, diagnosis diseases, diseases
detection, and control and therapy evaluation [3].The
segmentation and morphological techniques of Digital
image processing (DIP) can be applied for analyzing and
diagnosis a lot of medical images diseases such as WBCs,
the white blood cells play the main goal in the diagnosis
and analysis different diseases, the extracting information
is very important for hematologists. The different
techniques in an image processing are used to analyze the
cells to be more accurate and diagnosis systems for
remote diseases. There are some complications to extract
some data from RBCs in the cells wide variety in shape,
edge, position, and size. Moreover, when the illumination
is imbalanced, the image contrast between cell
boundaries and the background varies based on the
capturing process conditions [4].
In the last few years, the image processing techniques got
rapidly grown, where hematologists can be used images
segmentations of blood automatically, blood slides and
blood boundaries for detecting diseases in the diagnosis
system.
The research study is focusing on RBCs segmentation
process for human blood system using elitism immigrants
multiple-objective of Genetic Algorithms and digital
images processing. The main goal is to analyze RBCs
using EIMOGAs that has been developed in the last
years. The using of EIMOGAs in the segmentation
techniques of the digital image processing can be applied
set of constraints to finding data about the ratio cytoplasm
to classify and identify various types of cells such as a
lymphocyte, basophil, and neutrophil. The segmentation
methods have been applied in many works and different
area of images processing, related to region growing,
border detection, edge, watershed clustering, and
mathematical morphology and filtering processes.
The author proposed an automatic medical system for the
segmentation technique and border identification for
whole objects based on image boundary among the
images database system that is taken from a blood slides
and the original image [5], the using of images processing
are used as they are not expensive and do not require
complex testing and labs equipment, the system focus on
International Journal of Computer Science and Information Security (IJCSIS),
Vol. 15, No. 12, December 2017
252 https://sites.google.com/site/ijcsis/
ISSN 1947-5500

Thalassemia disease, Thalassemia features in microscopic
images and changes in gene geometry, texture, statistical
analysis, and colors contrast of RBCs, therefore, the
microarray technology can be applied to get a robust
genomic system for studying and analyzing the thousands
behavior of genes simultaneously. The images analysis
which was obtained from the microarray technology
strongly helped in the diagnosis, detection, and treatment
of most diseases.
In this research can be developed an automated
diagnosis system for analyzing and testing data from
microscope images directly and detects diseases cases,
for that purpose, the digital image processing performs
many operations such as modify image rotation,
extracting data from the image, locating genes in the
images, and the data mining will be normalized the
extracted data and getting the effective genes [6].
I. GENETIC ALGORITHM
Charles Darwin is invented Genetic Algorithm as the
natural selection process to take input and calculates an
output when a set of solutions can be produced. In the last
few years, GAs was created to represent processes in the
natural system that is important to evaluate and perform
an efficient search in the global domain and to have many
optimal solutions and more than that. GA is very
effective in the contrast improvement in quality and
produces an image based on the natural contrast in
different scale levels. GAs are the systematic random
search techniques to apply generic methods for solving
complex problems and optimization process.In the image
process, GA can use less information related to the
segmentation problems to be solved than the traditional
optimization systems, which almost require the derivative
objective functions. The fitness function is based on an
individual of images, and additionally, GAs can be used a
set of different operators (reproduction, crossover, and
mutation) to generate new solutions and use it to get an
optimal solution for the new images that may contain new
chromosomes [7]. Basically, the new children or
chromosomes in Genetic Algorithm are obtained of a
combination of features of their parents from original
images.
The elitism based immigrants multiple objectives of
Genetic Algorithms (EIMOGAs) is a new technique will
be used in the image processing to produce a set of newly
enhanced pixels of the image to be much better than the
original image and contains good features, the Image
segmentation will be applied EIMOGAs techniques to
enhance and improve image quality for extraction more
details about the degraded images. The techniques of
image colors have some problems such as colors image
enhancement applied in the true colour (RGB), where the
colour spaces are not suitable for the human system, and
the distribution colours in the images are inappropriate
the normal visual limits to human perception[8], one
technique is not enough to be suitable for one type of
image degradations in the RBCs. EIMOGAs have the
ability to select optimal colors and segmentation regions
to choose appropriate features of the analysis size and
select the heuristic thresholds to solve complex problems
[7].
II. THALASSEMIA DISEASES
There are some main factors will be used in this research
to analyze blood color of RBCs, the cells shape, and the
cells number, the experiments diagnosis will be checked
whether the required factors are negative or positive
results, a lot of diseases can happen to cause the size
changing cell, shape cell, and the blood cell color. The
researchers can be used blood count analyses, blood
images analyze, iron analyzes, and the HPLC analyze to
check whether the patients are having thalassemia
diseases or not. In this research, we proposed a system
that can be applied to diagnosis thalassemia disease based
on EIMOGAs techniques, The purpose in this work is to
help both patients and doctors and health care regarding
the reducing time for pathology, the reducing effort, and
more accurate in achieving outputs. In this research will
be studied two types of thalassemia disease are alpha
thalassemia and beta thalassemia. Thalassemia diseases
cause a reduction in the lifespan of red blood cells , the
disease is a result of an imperfection in the genes that
regulate the haemoglobin formation, which is a core
ingredient of the red blood cells ,hence thalassemia is
hereditary blood disorder characterized by abnormal
haemoglobin production and very common in subtropical
and tropical areas, for instance, thalassemia disease was
infected 280,000,000 people in 2013, with about 439,000
having a dangerous disease, the most common among
Middle Eastern people, African descent, Italian, Greek,
and South Asian ,both females and Males have similar
disease rates, the resulted in 16,800 deaths in 2016,
down from 35 thousand, deaths in 1990, so the blood
characteristics should be analyzed to make a good
diagnosis.
The automated diagnostics system have been developed,
using available rule-based tools to cover a blood broader
range related diseases containing anemia various types,
the alternative automated diagnostic tools are required, in
order to find the diagnostic goal, and the differentiation
among thalassemia patients, thalassemia traits, and
normal people.
The classification problems of thalassemia patients will
be formulated in the pattern recognition problems as
input process [9]. The test patterns and samples will be
blood-related features that are the red blood cell,
characteristics reticulocyte, and blood platelet, that is
extracted and used in the blood samples.
In the data mining techniques, the researchers are used
different rules and patterns to extract data based on the
clustering, summarization, association, and classification
using the machine learning techniques to test Beta
Thalassemia [10]. There are research studies illustrated
the Thalassemia testing indicators as Haemoglobin (Hb)
A2, Mean Corpuscular Haemoglobin, and Mean
Corpuscular Volume. In the Knowledge research, the
principal components analyses research were used to
discover β-Thalassemia, there are several algorithms for
machine learning are applied in the β-thalassemia
ISSN 1947-5500

classification based on new data set which is different
from the other researchers, the classifiers of data mining
are applied to differentiate among thalassemia traits in
different levels as iron deficiency patients, normal people
and the patients with other blood diseases [10].
III. IMAGE PROCESSING AND SEGMENTATION
The image segmentation is the partitioning operation of
an image into a collection of pixels connected sets, and it
also the most significant task in image processing, and
for better analysis and diagnosis, the original image will
be partitioned into different sizes and pieces. The most
important task in the image segmentation is to explore the
appropriate parameter selection based on GAs.The
purpose of image segmentation are:
1. The regions segmentation to cover the image
coordinates.
2. The linear structures segmentations that including line
segments and curve segments.
3. The 2D shapes segmentations, such as ellipses, circles,
and strips (regions, long, symmetric) for instance, the
cluster pixels inside salient image boundaries, the regions
corresponding to objects surfaces, or objects natural
parts.
The applications of image segmentation include: the
image recognition segmentation is using for face
recognition, the medical image segmentation such as
diagnosis operations, locating cancer diseases and other
dangers pathologies. The image segmentation process
was used in the agricultural imaging for crop diseases
detection. Traffic control system was used to identify
shape and size of objects, and moreover, it used to
identify moving scene objects using video compression
system, the Image segmentation has been divided into
two parts of approaches: the region based approach and
boundary based approach, in the first part, the purpose is
to determine if a pixel belongs to an object or not[11], in
the second part the goal is to locate the boundary curves
between the background and the objects.
There are four different types of image segmentation:
a) Segmentation greyscale. b) Segmentation texture.
c) Segmentation motion. d) Segmentation depth.
The Main algorithms of region segmentation are divided
into three categories:
1. Region-based segmentation technique:
Thresholding method can be used as a simple technique
to segment an image for the objects separating from the
background using a pixels features values that are
compared with a threshold values in order to determine
the class of the pixels, this method starts with the first
one pixel of a potential region and expands it by
inserting adjacent pixels for any image includes different
regions, the image will be segmented based on the
different areas of the image which each piece has a
range of features values, the thresholds are significant to
select these thresholds, and it very effectively and
useful in the segmentation quality of the images, finally,
the statistical test processes used to take a decision
which pixels will be inserted into a region
segmentation or not.
2. Clustering-based image segmentation technique
is dividing the image into different classes which do not
require prior information. In the same type of classes, the
data should be collected together in similar classes and
the data which contains a different type of classes will be
in different classes as possible.
3. Edge-based image segmentation Technique is
the main features of the original image, which include
valuable data useful in image analysis and diagnosis of
object classification and explores the detection of
boundaries among the different region's image [12]. The
boundaries discontinuities occur among the pixels of the
selected features such as intensities, textures, and colors.
IV. IMAGE SEGMENTATION USING GENETIC
ALGORITHM.
The parameter selection will be applied using EIMOGAs
to enhance the parameters selection of the images
segmentation and to improve its outputs. The pixel scale
and level of segmentation implement GAs will be used to
complete region labeling tasks of the image segmentation
processes, the proposed method should be used the image
adaptive segmentation including the following steps
[10]:
1. Compute the image statistics tables give us the
probability for a given degrees of a confidence level and
identically distributed normally to select suitable
threshold. 2. Generate an initial population of
segmentation image. 3. The image Segmentation applied
initial parameters selections. 4. Compute the
segmentation based on quality measures to satisfy
conditions of the fitness function.
5. To select new individuals should be used the
reproduction operator to generate new population
applying by using the mutation and crossover operators.
6. The image segment should be used new parameters to
calculate the segmentation quality of an image.
7. Analysis and modify the knowledge based on the
knowledge structures of the new image.
V. GENETIC ALGORITHM AND CHROMATIC FEATURES
In this research, we will be applied Elitism Immigrants
Multiple Objective of Genetic Algorithms (EIMOGAs)
with Chromatic features to describe the color distribution
and the grey-level of the images, which are the most
discriminative features of Red Blood Cells. the image
pixels is represented a segmented object such as (RBCs,
RBCs, Nucleus, Cytoplasm, Cells Parasites ), The GAs
selection operator is used to detect the edge of cells
boundaries that have the same colors of pixels from the
current population (RBCs images) that will be used new
generation.
The convergence process will be completed and achieved
in under the iterations required the number to detect
RBCs and complete blood counted for a new generation
and population.
In the next step of the population, solutions are
represented intensity cells colors and chromatic features
which can be detected and computed using EIMOGAs of
RBCs. In this stage of research will use generation,
ISSN 1947-5500

mutation, selection, elitism immigrants, integrate all
several immigrants of memory scheme and combined
into the EIMOGAs to improve its searching capacity for
the image process environment.The image process is a
stochastic process where pixels values are modeled as
random variables, the GAs can be applied to calculate the
probability density of grey level and color distribution as
its fitness function [13].
The fitness function is used to get robust convergence as
building simulations for RBCs image as possible with
reliable convergence and a high convergence with the
original image.
the elitism-based immigrants schemes Multi-objective of
genetic algorithms (EIMOGAs) efficiently improve the
GAs performance in the image processing environment
,and the best selection individual of color pixels based
on fitness function from the previous generation is used
to create immigrants included probability density of grey
level , colors gradient ,colors distribution, cells color and
boundaries shapes of RBCs into the population using a
genetic operation (selection, evaluation, mutation ,and
recombination ) , the new process of generation will be
the implemented using the elite process et from the
previous generation gt-1 to create new immigrants, as a
set of r(e) × n(t) individuals are generated based on
fitness function and mutation et−1 with a probability p(et-
1), where n(t) is the population number of the image
colors , and r(e) is the number ratio of elitism immigrants
for each color to the population number, the selection
operator of EIOMGAs selects set of cells color of RBCs
as the best solutions that have a better classification,
based on a fitness functions [14], and then it will be
carried forward for recombination image process.
The sensitivity analysis and the results are shown in the
final experiments. EIOMGAs are efficiently improve the
genetic algorithms performance in the image processing
environment, and the best individual from the previous
generation (RBCs Image) to next generation can be
selected and created immigrants with optimal solutions
into the population by evaluation and mutation process.
VI. PROPOSED SYSTEM
The proposed analysis system for RBCs segmentation
explants the phases are shown in Figure 1. The image
pre-processing of the blood smears is applied for
removing noises, improving and contrast variation and
luminance in the original images. In the second phase, a
segmentation processes are applied and implemented to
explore and isolate the interest objects of the image. The
third phase goal is to extract the objects characters to be
used in the next phase of the process, the Features
selection method is applied to decrease the redundant
data and built classification stage. The selected features
are selected for input to the classification method and
take the decision about the class assignment by using
EIOMGAs as shown in figure 1.
The main goal of the segmentation process is to separate
RBCs from another different ingredient of blood image.
The blood smear consists four components, the image
background, WBCs, RBCs, and cytoplasm. WBC should
be darker than the background, and RBCs seem on a
high-intensity scale [15]. And also, there are shapes
variation in cells and their nucleus.
Figure 1 shows the block diagram of the segmentation
scheme.
Fig.1.The Proposed Block Diagram of RBCs Analysis and
Methods Using EIMOGA
VII.IMAGE SEGMENTATION AND ACTIVE CONTOUR
MODELS
In the last few years, there are recent developments in the
medical imaging fields have brought a new techniques
research on image processing for improving medical
analysis and diagnosis in segmented images. This
technique has been developed to identify specific
structures in a magnetic resonance imaging (MRI). The
Active Contour methods are adaptable to the desired
features in the image.
There are several forms and different types of RBCs
images. The applying appropriate method for variable
shapes and segmenting for RBCs has been always a
challenge for researchers between segmentation methods,
the active contour model has a lot of enhancements and
implemented in the last few years, In the RBCs, the
image should be used active contour models which are
changeable curves to respond their change forms to avoid
deform objects boundaries in an image segmentation [16
].
The active contour models can be moved based on
internal or external forces extracted from the image
characteristics. The active contour adaptation occurs in
response to both internal and external forces, the external
forces model has described the gray level gradient, the
active contour models can be divided into two types: the
parametric models like the Snakes model, which defines
a resilient contour that can dynamically adapt to required
edges of the image objects, and the geometric models,
such as the Level Set model [13] it embeds the front to be
zero level set in the higher dimensional function, to
calculate the new function evolution, this evolution
operation is dependent on the image characteristics
extracted and geometric restrictions of the function.
In the processing scheme, the segmentations are
implemented on sub-images, the parametric snake model
is a curve x(s) defined in Equation.1 [4], to move through
RBCs Feature Extraction
RBCs Selection
RBCs Separation
RBCs Identification
RBCs Edge Detection
RBCs Segmentation
RBCs Samples by
Microscopic Blood Image
RBCs Classification
ISSN 1947-5500

the image spatial domain and minimize the energy
function E(s) defined in Eq.2.
( )
= ( ) , ( ) , ∈ 1,0 (1)

( ) =
1
2
( ∣∣ ( ) ∣ ) ∣∣∣ ( ) ∣ ( ) (2)
Where x'(s) denotes the first derivative, x″(s) denotes the
second derivative of x(s). While α, β are parameters of
weighting to control the rigidity and tension of snake,
respectively. Ext is the function of external energy which
is derived from the image to take smaller values of
features of boundaries.
The external and internal forces are used the image
gradients as a parametric active contour of Snakes
models, the external and internal forces will be used the
image gradients as a parametric active contour of Snack
models. The gradient-based model is better models to use
insensitivity to its initial parameters positions and wide
capture region of images. Gradient vector flow
detriments the object boundaries when addressed locked
to the object boundary, while on homogeneous regions
will be changed smoothly and will be more extended to
the image border, the gradient vector flow field is
selected as v(x, y) to be vector field is written in
equation.3 [17], which reduces the energy function as
defined in the equation. 4.
( , ) = ( , ) , ( , ) (3)

( ) = ( ∣∣ ( ) ∣ ) ∣∣ ( ) ∣ ( ) (4)
The internal energy E for the gray-level images I (x,y) is
identified as:
=
( ∣∣ ( ) ∣ ) ∣∣ ( ) ∣ (5)
While the external energy can be identified as:
= ( ( , ) ∗ ( , ) ∣ ( 6)
Where g denotes a two dimensional of the Gaussian filter
with a normal deviation, is identified the gradient
operator. This filter is applied to the image in order to
enhance the map image edge and to reduce an image
noise . The regions closer to edges will be given the
gradient image high rates. In this research, the cell
boundaries will be extracted using edge detection and
avoided missing off the edges occurring[18], so the
image smoothing using a Gaussian filter to reduce noises
with normal deviation is written as the following [17].
( , ) = ( , ) ∗ ( , ) (7)
Gaussian’s smoothing operator (GSO) is a 2-Dim used to
remove noise and detail with special properties as defined
in Eq.8.
( , ) =
1
2π

(8)
There is another problem needed to a solution using
additional parameters to improve the external force (k,
k1) to improve the capture range in heterogeneous
regions of the image edges. This problem will be
enhanced using a constant normal (k) [19], to control the
external force, the active contour model can be inflated or
deflated, based on the sign and magnitude (k1) of the
external force. in this paper, the author is proposed
applying the balloon model to prevent the snake from
stalling in the image homogeneous regions and should be
taken to select appropriate values to (k, k1 ) to make the
snake control edges and noise, without exceeding the
desired characteristics for contour regions , which is
written in Eqau.9.
= ( )
‖ ‖
( ( ), ( )) (9)
Energy Surface and Optimum Thresholding is the basics
approach to image segmentation is an amplitude
thresholding, a threshold T is chosen to separate the two
regions modes, the image point for I(x,y) >T is
considered as object points[19], otherwise, the point is
called a background point. The threshold method is
defined as:
( , ) =
0 , ( , )
1 , ( , )
(10)
Where T is set on the entire image basis I(x, y), and the
threshold is global. When T depends on spatial
coordinates x and y, based on a dynamic threshold, when
T depends on both I(x,y) and set property p(x,y) of local
image, the average of gray level in a neighborhood
centered on I(x,y), the threshold will be local and T is set
according to a fitness function is defined by:
( , ) = ( , ), ( , ) (11)
Template Matching is a new type technique in the image
segmentation based on prior knowledge of the detected
object in image analysis, using the presence detection of
an object in a scene, and identifies its position in current
given scene [20].
The object locating can be described using a template
T[x,y], in the image I[x,y], The best match Searching can
be minimized the mean squared errors as written below:
, = , ( ), ( ) (12)
In this research, the correlation technique of images will
be used for exploring match of a searched shape w(x, y),
of size k*l within an image I(x, y)of a larger size m × n.
Where w(x, y) is a search of shape, shape size denotes as
z (l, k) , maximum size as m(l, k), the summation will be
taken in the image region when w and I not separately,
the correlation function techniques have the sensitive
disadvantage to local intensities of w(x,y) and I(x,y), the
correlation coefficient C(s,t) can be used to remove
difficulty pattern matching of local intensities as
follows.
( , ) =
∑ ∑ ( , ) ̅( , ) ( , )
∑ ∑ ( , ) ̅( , ) ∑ ∑ ( , ) )
(13)
In the image, analysis can be used Hough transform
technology as a technique of a feature extraction image
for RBCs number and to get the number of red blood cell
count in the image. Then using machine learning
algorithms tool, which has developed a formula to
convert a number of red blood cells in the image to actual
count by Hough Transform, blood count calculates the
blood cells number in a cubic millimeter of blood
volume. In this research can be calculated the number of
ISSN 1947-5500

RBCs per cubic millimeter based on the cells number in
the given image [20].
RBCs count per_cumm = RBCs / ((S /M2*) * dilution_factor.
Where (cu, mm) is cubic millimeter, and (µl, mcl) is
microliters, them is magnification.T is a film thickness. S is an
image size.
The Circular Hough Transform technique can be used as a
measuring tool to calculate the accuracy using the result of Red
Blood Cells number compared with manual counting as the
following:
Accuracy =
RBCcount
ActualCount
∗ 100% ( 14)
The RBCs classification results need to use set of the
numerical analyzing as parameters: mean corpuscular
volume (MCV), RBCs distribution width, RBCs count,
mean corpuscular hemoglobin (MCH), hemoglobin
count (HB), and mean corpuscular hemoglobin
concentration (MCHC), for identifying the analyzing
combinations.
The hemoglobin (HB) is responsible for the red blood
cells color, HB Contents can be calculated by measuring
the gradient of colors, the threshold Technique an image
will partition into two parts: the foreground and the
background.
The binary algorithm was used to calculate the values
colors in an image. The classification tools used to
formulate a formula and calculate the hemoglobin
contents[21][22].
There are three parameters can be identified the red cells
characteristics including reporting units, formulas, and
definition to calculate each parameter as shown below.
MCV is the RBCs average size constituting the sample.
One femtoliter is 10-15 L, the adult's interval (80 - 100
fL) is defined by.
MCV =
Hematocrit ∗ 10
RBC ∗ (10 / )
∗ 100% ( 15)
MCH is the hemoglobin average weight in the RBCs, one
pictogram is 10 -12 grams, adults interval (26-32 pg) is
defined by MCH =

( / )∗
∗( / )
( 16)
MCHC is the hemoglobin average concentration in the
RBCs, adults interval (32 - 36 g/dL) is defined by.
MCHC =
Hb (g/gL) ∗ 100
Hematocrit
∗ 100 (17)
The blood samples need to be diluted, so there are some
of the RBCs in whole blood to be accurately counted in a
microscope[23]. The dilution factor (Df) is provided by
the industrialist and is around 200X as shown below in
equation 18.
Actual RBCs =
RBCs counted
( ) ∗ Ft ) ∗ Df
( 18)
VIII. RESULTS AND DISCUSSION
This section will be assessed the performance of the
proposed RBCs segmentation system. In our experiment,
22 blood samples images of thalassemia. The Image
captured was digitized using a digital video camera of
Sony high resolution, which was coupled to a Microscope
LCD Biological BX5.The experiments were implemented
using the Java Genetic Algorithms Package In this
research study, there are many methods to diagnosis the
Thalassemia diseases. There are three types of images
using proposed method to discover abnormal cell types,
the color red of blood cells and classification tools.
The research results are described in the following
sections:
1- The researcher has identified the abnormal cell
types using the various images of red blood cells. The
researcher experiments have classified the red blood cells
in different shapes. Figure.1 shows the different shapes of
red blood cells and the changes in the colors rates (Red,
Green, and Blue) of the images after the calculations
process have done, the analysis explained the relationship
between Thalassemia and non-thalassemia blood images
that it has taken as shown in Fig 1.
Image A Image B Image C
Fig.1. shows the results of thalassemia infected.
In these experiments , there are different results of Image
processing for thalassemia infected, image A is the
original image, Image B explained blood image, and
Image C shows results after applying pixel classification
using EIMOGAs as shown in figure1.
The target cell of blood image, which has hypochromic
microcytic and abnormal cell such as sickle cell, that
patient will be an affected person of a thalassemia, and
for further classification, the author uses other extracted
information from classification tools.
2. The RBCs calculations are made to extract the color
information (Red, Green, and Blue values) for each
image using image processing. In this research, the
studies have calculated the average intensity for each of
(red, green, blue color), that gave the average rate of
colors (red, green, blue) for each sample blood image.
The results of average intensity identified the R, G and B
rates of thalassemia normal image rate >185 and
thalassemia abnormal blood image rate <= 185.
The blood samples need to be diluted, so there are some
of the RBCs in whole blood to be accurately counted in a
microscope. The dilution factor (Df) is provided by the
industrialist and is around 200X as shown below in
equation 18. The results are reported as the number of
RBCs per cubic millimeter of blood.
Males Normal Values = 4.2 - 5.4 million RBCs / mm 3,
Females = 3.6 - 5.0 million RBCs / mm 3.
Stained thin blood should be taken by Digital microscope
to be more easily distinguished between platelets, RBCs,
and WBCs.
To differentiate between RBCs, WBCs, and Platelets,
RBCs is less stained as compare to WBC and platelets
leaving a bright spot and its intensity value similar the
background value.
ISSN 1947-5500

Table 1. Standard complete blood count System for healthy people.
Blood cell
type
Women Men
RBCs 4-5 M/µL 4.5 - 6.0 M/µL
WBCs 4.5 - 11 K/µL 4.5 – 11 K/µL
Platelets 150 – 450 K/µL 150 - 450 K/µL
Haematocrit 36% to 45% 42% to 50%
Haemoglobin 12 - 15 gm/100 ml 14 - 17 gm/100 ml
Gm /100ml: gram per 100 milliliters; ml: milliliter; gm: grams;
µ: Microliter; K: Thousand; M: Million.
After isolating process of RBCs, we need to apply a
counter for counting the RBC number in the image
process, so we have used a formula to calculate RBCs
number per cumm on the cells number in the given image
area (Ia) of the blood samples and the of the blood
sample film thickness (Ft) is 0.1 mm that is the standard
medical system. The magnification factor (Mf) which is
the magnification level under the microscope.
In the three experiments below, the assessing RBC
morphology procedure includes the smear examination in
the thinner edge where the erythrocytes (RBCs) are
randomly distributed, the most part singly, with
sometimes overlapping cells.
In the next experiment, figure 2 shows three images,
Image A is shown an acceptable area of RBC
morphology evaluation, we can see that the most cells
clearly can be distinguished with some overlapping cells.
In an image B is included area too thin, the RBCs appear
very flat such as shape cobblestone. An Image C is
shown the examined area is thicker than Image B, so the
cells will close together, the evaluation process of the
morphology will be used individual cells.
Image A Image B Image C
Fig 2. Segmentation thinner edge area.
The experiments results of the smear examination in the
thinner edge area and Assessing Erythrocytes (RBCs)
Morphology as shown in figure 2.
Table 2. The Result Comparison between Counted Manually
System and CHT using EIMOGA in an Image Processing.
Images Radius
Range
In pixels
Film
Thicknes
s
RBCs counted
Manually
System
RBCs
counted
proposed
System
Img1 5-12 0.1 3.768 3.542
Img2 5-14 0.1 4.349 4.161
Img3 4-15 0.1 6.821 6.742
Img4 4-14 0.1 5.783 5.621
Img5 5-13 0.1 4.981 4.813
Where: Magnification Factor(Mf)=300*300; Dilution Factor(Df) =200 ;
Table 1. Type Sizes for Camera-Ready Papers
After the calculations that have done. In Fig 5 shows the
changes in the colors rates (Red, Green, and Blue) of the
two Images.The analysis explained the relationship
between Thalassemia and non-thalassemia blood images
that it has taken.
In the research observations, based on the research details
and results, could able to analysis possible diseases
combinations when high or low above according to
values of parameters.
In figure 3, shows if there is a normal or abnormal of
MCV, MCH, RBC and MCHC attributes of a patient’s
blood sample, and this analysis can be diagnosed the
current disease for that patient.
(a) (b) (c) (d) (e)
(f) (g) (h) (j)
Fig 3. Different Images of RBCs samples using
EIMOGA
Figure.3 shows different Images of RBCs samples using
EIMOGA as the following: the images (a) and (b) are
healthy RBCs. the images (c) and (d) are infected RBCs
with ring parasites. The Images (e) and (f) are infected
with trophozoites parasites. The images (g), (h) (j) are
RBCs infected with schizont parasites.
Image A Image B
Fig4. Compression between original and enhanced image.
In figure 4, Image A corresponds to original Image, and
Image B displays enhanced image after applying
EIMOGAs approach for image filtering.
Image C Image D Image E
Fig 6. Shows different method segmentation results.
Figure 6 shows three different method segmentation
results of images. Image C is gotten using EIMOGAs
algorithm of the histograms approach. Image D is
obtained by applying EIMOGA for pixels classification.
Image E is used EIMOGA for clustering approach.
ISSN 1947-5500

Image A Image B
Fig 7.Shows EIOMGAs technique to enhance Images results.
The experiments results using EIOMGAs technique to
enhance the contrast of images,Image A is gotten using
EIMOGAs algorithm of the histograms approach. Image
B is obtained by applying EIMOGA for pixels
classification as shown in figure 7.
In this experiment as shown in table 3, the Original
Image number is 5 of microscopic image, the accuracy
average is 93.33% was implemented. The comparing
process is implemented between the input image and an
image after using Circular Hough Transform, there are
some of RBCs is not calculated in CHT method due to
deformable shape and another condition.
Table 3. Shows the RBC results are counted for 5 RBCs
Images using CHT method.
Original
Image
Actual
RBCs
CHT
method
Counted
Accuracy of
CHT
Img_RBCs_1 903 837 92.69
Img_RBCs_2 937 871 92.96
Img_RBCs_3 978 916 93.66
Img_RBCs_4 1018 957 94.01
Img_RBCs_5 1123 1048 93.32
Average= 93.33
In the next experiment, we have the same number of
Original Images from the previous experiment
microscopic image, the accuracy average is 97.05 was
implemented. The comparing process is implemented
between the input image and an image after using
EIMOGA to enhance analysis process, the final results
were more accuracy for the counted process of RBCs by
EIOMGA as shown in table 4.
Table 4. Shows the Accuracy Results for 5 RBCs Images using
EIMOGAs method.
Original
Image
Actual
RBCs
EIMOGAs
Counted
RBCs
Accuracy of
EIMOGA
Img_RBCs_1 903 877 97.12
Img_RBCs_2 937 913 97.44
Img_RBCs_3 978 949 97.03
Img_RBCs_4 1018 987 96.95
Img_RBCs_5 1123 1086 96.71
Average= 97.05
IX. REFERENCES
[1] Khalil Ibrahim Mohammad Abuzanouneh. “Hybrid
Multi Objectives Genetic Algorithms and Immigrants
Scheme for Dynamic Routing Problems in Mobile
Networks”. International Journal of Computer
Applications 164(5): 49 -57, April 2017.
[2] Khalil Ibrahim Mohammad Abuzanouneh “Parallel
and Distributed Genetic Algorithm Multiple
Objectives to Improve and Develop of Evolutionary
Algorithm, International Journal of Advanced
Computer Science Applications, Volume7 Issue 5,
2016.
[3] Ritter Nand CooperJ(2007)Segmentation and border
identification of cells in images of peripheral blood
smear slides. In Proceedings of the Thirtieth
Australasian Conference on Computer Science
62:161–169.
[4] Gonzalez RC, Woods RE, Eddins SL (2003) Digital
image processing using MATLAB. Prentice-Hall,
Upper Saddle River.
[5] Ritter Nand CooperJ(2007)Segmentation and border
identification of cells in images of peripheral blood
smear slides. In Proceedings of the Thirtieth
Australasian Conference on Computer Science
62:161–169.
[6]
[7] Cseke I (1992) A fast segmentation scheme for white
blood cell images. In Pattern Recognition. Conference
C: Image, Speech and Signal Analysis, Proceedings.
11th IAPR International Conference on vol. III:530–
533.
[8] SobelI(1978) Neighborhood coding of binary images
for fast contour following and general array binary
processing. Comput Graph Image Process 8:127–135.
[9] Prewitt JMS (1970) Object enhancement and
extraction. Picture Processing and Psychopictorics,
pp. 75–149.
[10] NHLBI. "How Can Thalassemias Be Prevented?". 3
July 2012. Archived from the original on 16
September 2016. Retrieved 5 September 2016.
[11] ZackG, RogersW, Latt S(1977)Automatic
measurement of sister chromatid exchange frequency.
J Histochem Cytochem 25:741–753.
[12] Otsu N (1975) A threshold selection method from
gray-level histograms. Automatica 11:285–296.
[13] Ongun G, Halici U, Leblebicioglu K, Atalay V,
Beksac M and Beksac S (2001) Feature extraction and
classification of blood cells for an automated
differential blood count system. Neural Networks.
Proceedings. IJCNN '01. International Joint
Conference on vol. 4:2461–2466.
[14] Kan Jiang, Qing-Min Liao and Sheng-Yang Dai
(2003) A novel white blood cell segmentation scheme
using scale-space filtering and watershed clustering.
Machine Learning and Cybernetics, 2003
International Conference on, vol. 5:2820–2825.
[15] Dorini LB, Minetto R and Leite NJ (2007) White
blood cell segmentation using morphological
operators and scale-space analysis. In SIBGRAPI '07:
Proceedings of the XX Brazilian Symposium on
Computer Graphics and Image Processing.
[16] Kass M, Witkin A, Terzopoulos D (1988) Snakes:
active contour models. Comput Graph Image Process
1:321–331
ISSN 1947-5500

[17] Canny J (1986) A computational approach to edge
detection. IEEE Trans. Pattern Anal Mach Intell, pp.
679–698.
[18] Xu C, Prince JL (1998) Snakes, shapes, and gradient
vector flow. IEEE Trans Image Process 7:359–369.
[19] Xu C and Prince J (1997) Gradient vector flow: a new
external force for snakes. In 1997 IEEE Computer
Society Conference on Computer Vision and Pattern
Recognition, 1997. Proceedings, pp. 66–71.
[20] Kumar BR, Joseph DK, and Sreenivasc TV (2002)
Teager energy based blood cell segmentation. Digital
Signal Processing. DSP 2002 14th International
Conference on, vol. 2:619–622.
Dr. Khalil Ibrahim Mohammed Abuzanouneh is
assistant professor in Computer Collage at the
Department of Information Technology, Qassim
University. In teaching, he has been focusing on applying
genetic algorithms and solves problem based on genetics
approaches in Computer Education. In research, his
current interests include optimization problem and data
compression and security. Dr. Khalil received his PhD
degree in Computer Engineering Ph.D.in Computer
Engineering Sciences “Computer Aided Design”.
.
ISSN 1947-5500

new image processing techniques using elitism immigrants multiple objective of genetic algorithms for disease detection_4

More Related Content

What's hot (19)

Similar to new image processing techniques using elitism immigrants multiple objective of genetic algorithms for disease detection_4 (20)

Recently uploaded (20)

new image processing techniques using elitism immigrants multiple objective of genetic algorithms for disease detection_4