See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/312494841

Determination and Classification of Blood Types using Image Processing

Techniques

Article  in  International Journal of Computer Applications · February 2017

DOI: 10.5120/ijca2017912592

CITATIONS READS

2 7,005

4 authors:

Ravindran G. Joby Titus

Sri Ramakrishna Institute of Technology Sri Ramakrishna Institute of Technology
4 PUBLICATIONS   3 CITATIONS    17 PUBLICATIONS   2 CITATIONS   

SEE PROFILE SEE PROFILE

Pravin Manikandan Pandiyan Pitchai

Sri Ramakrishna Institute of Technology Sri Krishna College of Technology
9 PUBLICATIONS   2 CITATIONS    13 PUBLICATIONS   5 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Integrated Feature Extraction for Image Retrieval View project

SWARM ROBOT IN INTELLIGENT VEHICLE SPEED MONITORING SYSTEM View project

All content following this page was uploaded by Joby Titus on 18 January 2017.

The user has requested enhancement of the downloaded file.

 International Journal of Computer Applications (0975 – 8887)
Volume 157 – No 1, January 2017

Determination and Classification of Blood Types using

Image Processing Techniques
G. Ravindran T. Joby Titus M. Pravin P. Pandiyan
Assistant Assistant Assistant Assistant Professor
Professor(Sr.Grade), Professor(Sr.Grade), Professor(Sr.Grade), Department of EEE
Department of ECE Department of ECE Department of ECE Sri Ramakrishna
Sri Ramakrishna Sri Ramakrishna Sri Ramakrishna Institute of
Institute of Institute of Institute of Technology,
Technology, Technology, Technology, Coimbatore-641010,
Coimbatore-641010, Coimbatore-641010, Coimbatore-641010, India
India India India

ABSTRACT being the result interpreted according to the occurrence or not

Determining of blood types is very important during of agglutination. The combination of the occurrence and non-
emergency situation before administering a blood transfusion. occurrence of the agglutination determines the blood type of
Presently, these tests are performed manually by technicians, the patient. Thus, the software developed in image processing
which can lead to human errors. Determination of the blood techniques allows, through an image captured after the
types in a short period of time and without human errors is procedure of the slide test detect the occurrence of
very much essential. A method is developed based on agglutination and consequently the blood type of the patient.
processing of images acquired during the slide test. The image
processing techniques such as thresholding and morphological
2. BLOOD AND ITS TYPES
operations are used. The images of the slide test are obtained A blood type (also called a blood group) is a classification of
from the pathological laboratory are processed and the blood based on the presence or absence of inherited antigenic
occurrence of agglutination are evaluated. Thus the developed substances on the surface of red blood cells (RBCs). These
automated method determines the blood type using image antigens may be proteins, carbohydrates, glycoproteins, or
processing techniques. The developed method is useful in glycolipids depending on the blood group system. Blood
emergency situation to determine the blood group without groups are identified by antigens and antibodies in the blood.
human error. Antigens are any substance that stimulates the immune system
to produce antibodies. Antigens can be bacteria, viruses or
Keywords fungi that cause infection and disease. Antibodies, also called
Blood samples; morphological techniques; Luminance; immunoglobulin are proteins manufactured by the body that
quantification.. help fight against foreign substances called antigens. When
antigens enter the body, it stimulates the immune system to
1. INTRODUCTION produce antibodies. The antibodies attach or bind themselves
Before the blood transfusion it is necessary to perform certain to antigens and inactivate it. The role of antibodies is to bind
tests. One of these tests is the determination of blood type. with antigens and inactivate them so other bodily processes can
There are certain emergency situations which due to the risk take over, destroy and remove the foreign substances from the
of patient life, it is necessary to administer blood immediately. body. There are many types of blood group. But, the major two
The tests currently available require moving the laboratory, it types of blood groups are,
may not be time enough to determine the blood type and is  ABO blood system
administered blood type O negative considered universal
donor and therefore provides less risk of incompatibility.  Rhesus blood system
However, despite the risk of incompatibilities be less
sometimes that cause death of the patient and it is essential to The ABO blood system is the most important blood group
avoid them. Thus, the ideal would be to determine the blood system in human blood transfusion. The associated anti-A,
type of the patient. Secondly, the pre-transfusion tests are anti-B antibodies are usually immune globin M, abbreviated as
performed by technicians, which lead to human errors. Since IgM antibodies. ABO blood system determines whether the
these human errors can translate into fatal consequences, person belongs to blood A or B or AB or O. There are four
being one of the most significant causes of fatal blood major blood groups determined by the presence or absence of
transfusions is important to automate the procedure of these two antigens A and B on the surface of red blood cells:
tests. Various blood type classification, diffusive reflectance, Group A – has only the A antigen on red cells
ABO Rh-D blood typing using simple morphological image
processing[1]-[4].There is a scope for determining blood types Group B – has only the B antigen on red cells
using image processing techniques. Image segmentation Group AB – has both A and B antigens on red cells
algorithm for blood type classification and various image
processing parameters are analyzed[5],[6],[9]. Image features, AB blood types have both A and B antigens and no A or B
such as color, texture, shape are analyzed [7],[8]. Low quality antibodies. As they lack antibodies, they can receive any type
ancient document images and antibody agent analysis using of blood and known as UNIVERSAL RECIPIENT.O blood
image processing is explained[10],[11]. The slide test consists have neither A nor B antigens, so their blood cells will not be
of the mixture of one drop of blood and one drop of reagent,

12
 International Journal of Computer Applications (0975 – 8887)
Volume 157 – No 1, January 2017

agglutinated by any recipient’s antibodies, therefore they are When T is constant over an entire image, the preceding
known as universal donor. equation is referred as global thresholding. When the value of
T changes over an image, it is called as variable thresholding.
3. EXISTING BLOOD GROUP TEST The term local or regional thresholding is also used to denote
To work out blood group of a person, red cells of that person variable thresholding in which the value of T at any point (x,y)
are mixed with different antibody solutions. If, for example, in an image depends on properties of neighborhood of (x, y). If
the solution contains anti-B antibodies and the person has B T depends on the spatial coordinates (x, y) themselves, then
antigens on cells, it will clump together. If the blood does not variable thresholding is often referred as dynamic or adaptive
react to any of the anti-A or anti-B antibodies, it is blood thresholding. Thresholding may be single or multiple. In single
group O. A series of tests with different types of antibody can thresholding, the image consists of a single light object on a
be used to identify blood group. If the person has a blood dark background. But in multiple thresholding, image consists
transfusion, the blood of the person will be tested against a of two or more light objects on a dark background. This
sample of donor cells that contains ABO and RhD antigens. If multiple thresholding method classifies the pixel at (x,y) as
there is no reaction, donor blood with same ABO and RhD belonging to the background if f(x, y) ≤T1, to one object class
type can be used. It indicates that the blood has reacted with if T1 ≤ f(x, y) ≤T2 and to the other object class if f(x, y) ≥T2.
certain antibody and is therefore not compatible with blood
containing that kind of antibody If the blood doesn’t The threshold of a multiple threshold image is given by,
agglutinate, it indicates that blood doesn’t have antigens a if f(x, y) > 𝑇2
binding the special antibody in the reagent. In existing system,
g x, y = b if T1 < 𝑓(𝑥, 𝑦) ≤ 𝑇2--------------------------- (2)
blood group is determined manually. In this system, adding
c if f(x, y) ≤ T1
solutions such as anti-a, anti-b, anti-d to the three samples of
blood took place. After some time, agglutination may or may Where a, b and c are three distinct intensity values.
not occur. Depending upon the agglutination, blood group can
be determined by the person manually. Disadvantages of this 4.1 Types of Thresholding
system are more chances of human errors are possible. Only 4.1.1 Global thresholding:
experts can tell the blood type by seeing at the agglutination Generally in image processing, the preferred approach is to
process. use an algorithm capable of choosing a threshold
automatically based on image data. This thresholding is
4. PROPOSED SYSTEM known as global thresholding. In Images with uniform
In our proposed system, reagents are mixed with three samples contrast distribution of background and foreground like
of blood. After sometime, agglutination may or may not occur. document images, global thresholding is more appropriate.
After the formation of agglutination, the slide is captured as an The following iterative procedure is used in this technique.
image and allowed to process in MATLAB image processing
toolbox. By using this system, more chances of human errors 1. Select an initial estimate for the global threshold T.
can be reduced. Image processing techniques used for blood
group identification are 2. Segment the image using T. This will produce two
groups of pixels: G1, consisting of all pixels with
 Pre-processing techniques intensity values greater than T and G2, consisting of
pixels with values less than or equal to T.
 Thresholding
3. Compute the average intensity values m1 and m2 for
 Morphological operations the pixels in regions G1 and G2 respectively.
 HSL Luminance plane 4. Compute a new threshold value:
 Quantification 1
T = 2 (m1 + m2) ----------------------------------- (3)
In this proposed work various pre-processing techniques such
as color plane extraction, gray conversion were used. Image Repeat steps 2 through 4 until the difference in T in
preprocessing can significantly increase the reliability of an successive iterations is smaller than a predefined value, ΔT.
optical inspection. Several filter operations which intensify or Parameter ΔT is used to control the number of iterations in
reduce certain image details enable an easier or faster situations where speed is an important issue. In general, the
evaluation. Users are able to optimize a camera image with just larger the ΔT is, the fewer iterations, the algorithm will
a few clicks. Filtering contains numerous image filters for perform. It can be shown that the algorithm converges in a
image optimization miscellaneous filter for edge enhancement, finite number of steps, provided that the initial threshold is
noise suppression, character modification, etc. Image chosen between the maximum and minimum intensity levels
processing includes It includes several functions for image in the image. In terms of segmentation, the algorithm works
processing. Contrast increase by static or dynamic binarisation, well in situations where there is a reasonably clear valley
lookup tables or image plane separation. Resolution reduction between the modes of the histogram related to objects and
via binning. Image rotation. An image f(x,y) is composed of background.
light objects on a dark background. This technique is used to 4.1.2 variable or local thresholding
extract the light objects from the dark background. It is done Global thresholding methods typically fail when the
by using a threshold value T. Any image point (x, y) at which background illumination is highly nonuniform. One solution
f(x,y) >T is called an object or foreground point; otherwise, the to this problem is to attempt to estimate the shading function,
point is called background point. use it to compensate for the nonuniform intensity pattern, and
The threshold of a binary image is defined as, then threshold the image using any global thresholding
methods. Another approach is used to compensate for
a, if f x, y > 𝑇 irregularities in illumination, or in cases where there is more
g x, y = ----------------------------------- (1)
b, if f x, y ≤ T than one dominant object intensity, is to use variable

13
 International Journal of Computer Applications (0975 – 8887)
Volume 157 – No 1, January 2017

thresholding. This approach computes a threshold value at Quantification:

every point (x,y) in the image, based on one or more specified
properties of the pixels in a neighborhood of (x,y). Quantification is the measure of intensity of pixels to measure
the region of interest. The first option is the measure of
4.1.3 Adaptive thresholding: reposition region of interest, dynamically restores the region
Adaptive thresholding typically takes a grayscale or color of interest based on the coordinate system defined. The
image as input and in the simplest implementation, outputs a second option is reference coordinate system allows selecting
binary image representing the segmentation. For each pixel in the coordinate system associated to the region of interest.
the image, a threshold has to be calculated. If the pixel value These features identify the occurrence of agglutination in the
is below the threshold it is set to the background value, blood sample and reagent based on the value of standard
otherwise it assumes the foreground value. In the local deviation. If no agglutination occurs in samples and the values
adaptive threshold technique, a threshold is calculated for of standard deviation do not exceed 16 then quantification
each pixel, based on local statistics such as range, variance, or takes the values between 0 and 10. If the standard deviation
surface –fitting parameters of the neighborhood pixels. It can values greater than or equal to 16 then quantification shows
be approached in different ways such as background the values between 20 and 70.
subtraction, water flow model, mean and standard deviation of
pixel values, and local image contrast. Some drawbacks of the
local thresholding techniques are region size dependent,
individual image characteristics, and time consuming.
Therefore some researchers use a hybrid.

4.2 Morpholgical operations:

Morphology is a tool of extracting image components that are
useful in the representation and description of region shape,
such as boundaries, skeletons, and the convex hull. In
morphological operation, there are two fundamental
operations such as dilation and erosion, in terms of the union
of an image with translated shape called a structuring element.
This is a fundamental step in extracting objects from an image
for subsequent analysis. The fundamental operations in
morphological operations can be listed as

4.2.1 Dilation
Dilation is the process that grows or thickens the objects in an
image and is known as structuring element. Graphically, Fig 1: Steps for determining blood types
structuring elements can be represented either by a matrix of
0s and 1s or as a set of foreground pixels. The dilation of A by 5. METHODOLOGY AND ANALYSIS
B is set considering all the structuring element origin The image for analyses is collected from the laboratory and
locations where the reflected and translated B overlaps at least digital images stored in JPEG format. These images are pre-
one element. It is a convention in image processing that the processed using color plane extraction. The riginal slide test
first operand of AB be the image and the second operand is image is shown in figure.
the structuring element, which usually is much smaller than
the image.

4.2.2 Erosion
Erosion shrinks or thins objects in binary image. The erosion
of A by B is the set of all points z. Here, erosion of A by B is
the set of all structuring element origin locations where no
part of B overlaps the background of A. In image processing
applications, dilation and erosion are used most often in
various combinations. An image will undergo a series of Fig.2. Input image
dilations and erosions using the same, or sometimes different,
structuring elements. The most important combinations of 5.1 Color plane extraction
dilation and erosion are opening and closing. The color plane contains the color information in images. The
foreground and background color of each image has different
4.3 HSL luminance values. In this work only green color component is extracted
HSL luminance stands for Hue, Saturation and Luminance. because it contains the maximum value in the RGB color
Hue is expressed in a degree around a colour wheel, while plane. The green plane extraction is shown in figure.
saturation and brightness are set as a percentage. Shade uses a
standard window colour picker with a scale of 0 to 239(which
can be regarded as 1 to 240) for each quality, which makes
calculations easy. HSV stands for Hue, Saturation and Value.
A third model, common in computer vision applications, is
HIS. In each cylinder, the angle around the central vertical
axis corresponds to hue and saturation. Hue in HSL and HSV
refers to Saturation and differs dramatically. RGB devices has
unique HSL and HSV spaces Fig.3. Color plane Extraction

14
 International Journal of Computer Applications (0975 – 8887)
Volume 157 – No 1, January 2017

5.2 Thresholding operation uniformly reduces the size of the objects in relation
It is the simplest method of image segmentation. From a gray to their background and dilation expands the size of the
scale image, thresholding is used to create binary images. The objects. By using dilation and erosion, secondary operations
gray scale samples are clustered as background and object. It like opening and closing can be done. Morphological
may be viewed as an operation that involves test against a operations are used to eliminate noise spikes and ragged
function T of the form edges. Closing operation is used to fill the holes and gaps. It is
the process of dilation which is followed by erosion. It can be
T = T x, y, p x, y , f(x, y) ----------------------------- (4) observed that the segmented image is filled using operation is
shown in Figure.
Where f(x, y) is a gray level at the point (x, y) and p(x,
y) denotes some local property of the point. A threshold
image is defined as,
1, if f x, y > 𝑇
g x, y = ------------------------------ (5)
0, if f x, y ≤ T

Thus pixels labeled 1 corresponds to objects and pixels

labeled 0 corresponds to background. Based on T dynamic
and adaptive conditions are chosen. When T depends only on Fig 6. Filling Holes
f(x,y) and the value of T solely relates to the character of
pixels, this thresholding technique is called global Opening operation is used to smoothens the contours of cells
thresholding. Clustering is the task of grouping a set of and parasites. It is process in which erosion is followed by
objects in such a way that objects in the same group are more dilation. Therefore, the opening is the dilation, followed by
similar to each other than to those in other groups. It can be the erosion of the result. It can be noticed that it smoothens
observed that both background and object are separated as the contours of cells by removing small objects is shown in
shown in figure. Figure.

Fig 7. Removing small objects

Fig 4. Thresholding
5.5 HSL luminance plane
5.3 NIblack thresholding It stands for hue, saturation and lightness. In this cylindrical
Niblack’s algorithm calculates a pixel-wise threshold by coordinate representation of points in RGB color plane is
sliding a rectangular window over the gray level image. The used. The result of HSL plane is shown in figure
computation of threshold is based on local mean and standard
deviation s of all pixels in the window and is given by the
equation

Tniblack = m + (k × s) ---------------------------------- (6)

Where m is the average value of the pixel, and k is fixed to -

0.2 and s is the standard deviation. If threshold T depends on
both f(x,y) and p(x,y),this thresholding is called local
thresholding. This method divides an original image into
Fig 8. HSL Luminance Plane
several sub regions. It can be observed only the segmented
part of an image shown in figure. 5.6 Quantification
Quantify function expressed as a number or measure of
quantity. It measures intensity only in the region of interest
selected. Area (percentage of surface examined for full
image), mean (average value of the pixel), standard deviation,
minimum and maximum values of pixel intensity are
determined. Using the value of standard deviation, occurrence
of agglutination is identified. If standard deviation value does
not exceed 16, it is confirmed that no agglutination occurs. If
the standard deviation value is equal or greater than 16, it is
Fig 5. Niblack thresholding confirmed that agglutination occurs. The result of
quantification is shown in Figure.
5.4 Morphology
It includes pre or post processing operations such as dilation,
erosion, morphological filtering and granulometry. The
fundamental operations are dilation and erosion. The erosion

15
 International Journal of Computer Applications (0975 – 8887)
Volume 157 – No 1, January 2017

7. REFERENCES
[1] Neha Srivathsa; Dhananjaya Dendukuri, “Automated
ABO Rh-D blood type detection using
smartphone imaging for point-of-care medical
diagnostics” IEEE Conference Publications, Year: 2016,
Pages: 4345 – 4348.
[2] Fabien Picot; Julien Pichette, "Imaging system based on
Fig 9. Quantification diffusive reflectance spectroscopy for bloodvessels
detection during brain biopsy procedure” IEEE
6. RESULTS Conference Publications,Year: 2016, Pages: 1 – 1
[3] Tejaswini.H.V, M.S. Mallikarjuna Swamy,
“Determination and classification of blood types using
image processing techniques” ITSI Transactions on
Electrical and Electronics Engineering (ITSI-TEEE)”,
Volume-2, Issue-2-2014, pp.29-33.
[4] Ana Ferraz, Filomena Soares “A Prototype for Blood
Typing Based on Image Processing” The Fourth
Fig.10. Sample blood sample analysis International Conference on Sensor Device Technologies
and Applications, Copyright (c) IARIA, 2013.
[5] Miss Hetal J.Vala, Prof.Astha Baxi “A Review on Otsu
Image Segmentation Algorithm” Volume 2, Issue 2,
February 2013.
[6] Miss.Madhuri, G.Bhamare “Automatic Blood cell
Analysis by using Digital Image processing:A
preliminary study”, Vol-2, Issue-9, September-2013.
[7] T.Romen Singh, Sudipta Roy, O.Imocha Singh, Tejmani
Sinam and Kh.Manglem Singh “A New Local Adaptive
Fig.11. Blood sample morphology analysis
Thresholding Technique in Binarization”, IJCSI
The method developed proves that it is effective and efficient International Journal of Computer Science Issues, Vol. 8,
method to detect the agglutination and determines the blood Issue 6, No 2, November 2011.
type of the patient accurately. The use of image processing
[8] Er.Nirpjeet kaur, Er.Rajpreet kaur, “A review on various
techniques enables automatic detection of agglutination and
methods of Image thresholding”, Er.Nirpjeet kaur et al. /
determines the blood type of the patient in a short interval of
International Journal on Computer Science and
time also helpful in emergency situations. In future it is
Engineering (IJCSE), Vol. 3 No. 10 October 2011.
intended to improve the system developed by making it
smaller so that it can be portable and incorporate GSM [9] Qingqiang Yang, Wenxiong Kang “General Research on
technology, to send a message to the mobile of technician of Image Segmentation algorithms”, I.J. Image, Graphics
the laboratory in order to avoid unnecessary travel. and Signal Processing, 2009, P.No:1-8.
[10] Khurram Khurshid, Imran Siddiqi, Claudie Faure, Nicole
Vincent “Comparison of Niblack inspired Binarization
methods for ancient documents” DDR, Volume 7247 of
SPIE, Page 1-10.
[11] M.R.Brown, P.Crim, “Organizing the antibody
identification process” Clin Lab Sci, Vol.20, 2007,
Pp.122-126.

IJCATM : www.ijcaonline.org
16

View publication stats