Comparison of Acute Leukemia image segmentation

using RGB, HSI and YUV color spaces

1Rohini Patil 2
Prof. Mandar Sohani 3
Prof.Sachin Bojewar
TEC,Nerul,navi mumbai . VIT,Wadala,Mumbai. VIT,Wadala,Mumbai.
1aditipatil01@gmail.com 2mandar.sohani@vit.edu.in 3sachin.bojewar@vit.edu.in

ABSTRACT- as to cluster pixels into salient image regions, i.e., regions

corresponding to individual surfaces, objects, or natural parts of
The field of computer vision is concerned with information
objects. Image segmentation has applications separate from
extraction from images. Image segmentation is a fundamental step
computer vision. In segmentation, an image is partitioned into
in many areas of computer vision including stereo vision and
different non-overlapping homogeneous regions, where the
object recognition. Segmentation is the process of assigning a
homogeneity of a region may be composed based on different
label to every pixel in an image such that pixels with the same
criteria such as gray level, color or texture. The research in the
label share certain visual characteristics The recognition of acute
area of image segmentation has led to many different techniques,
leukemia cell images based on color image is one of the most
which can be broadly classified into histogram based, edge based,
challenging tasks in image processing . The use of color
region based, clustering, and combination of these techniques
segmentation for acute leukemia images is proposed. The
technique segments each leukemia image into two regions; blast The segmentation of color images requires more information
about the scene while less attention is required to develop the
and the background. The blast contains white blood cells. In this
algorithms than the gray scale images. Color images allow more
work, color segmentation of acute leukemia images is carried out
reliable image segmentation and are also very rich source of
using different color spaces and the evaluation of color spaces is
information as well as they will provide a better description of a
carried out using a segmentation algorithm.
scene than grayscale images. Contrast enhancement and image
General terms: Pattern Recognition. segmentation play an important role in most medical image
Keyword: Image segmentation, Acute leukemia ,RGB, analysis tasks. Contrast enhancement at the pre-processing stage
HSI, YUV. becomes the most important process for a successful feature
extraction and diagnosis of leukemia.
1.INTRODUCTION
The term leukemia refers to a group of cancers of the blood
The field of computer vision is concerned with information
cells. In leukemia, white blood cells become abnormal, divide and
extraction from images. Image segmentation is one of the most
grow in an uncontrolled way. Leukemia is a broad term covering a
challenging tasks in image processing and is a very important
spectrum of diseases. Leukemia is clinically and pathologically
preprocessing step in the problems in the area of image analysis,
subdivided into a variety of large groups. The first division is
computer vision, and pattern recognition.
between its acute and chronic forms: The word acute means the
Segmentation refers to the process of partitioning a digital image disease develop and progress rapidly ,while chronic leukemia is
into multiple segments pixels, The goal of segmentation is to characterized by the excessive build up of relatively mature, but
simplify and/or change the representation of an image into still abnormal, white blood cells. Typically taking months or years
something that is more meaningful and easier to analyze as well
to progress. Image segmentation can be used in identifying blast
in leukemia. Due overlapping between the blood cells, it is still a
challenging task to segment and count the blasts in leukemia
2. METHODOLOGY
The primary goal of acute leukemia blood cell image
segmentation is to extract morphological component such as blast
from its complicated blood cells background using RGB,HSI and
YUV color spaces. The algorithm for RGB color space is as given
below.
is based on S component. This procedure involves conversion
from RGB to HSI and back conversion is required. The algorithm
for HSI color space is as given below.
1:Capture acute leukemia slide image at 40x magnification and
save it into bitmap.
2:Extract the S-component information from original image.
3:Plot the histogram to determine the threshold value.
4:Select the threshold value using the S component from HSI
color space from the histogram.

2.1 Acute leukemia image segmentation 5: Implement N x N (N=5) the median filter to the resulted
images.
based on RGB color space
6: Convert the resulted image to RGB & display the resulted
image.
1)Capture acute leukemia slide image and save into bitmap (.bmp)
extension . The transformation procedure from RGB to HSI involves

2) Applying the Contrast Stretching to increase the contrast of the following equations.

image captured.
3) Select the threshold value by using histogram. Hue= if B <= G
3600 - if B > G
4) Implement the median filter N X N(N = 5) to the resulted
images.
1/2[(R-G)+(R-B)]
5) Display the resulted image in RGB color space.
=cos-1
[(R-G)2+(R-B)(G-B)] 1/2
For RGB color space, local contrast stretching (LCS) is an
enhancement method performed on acute leukemia images for Saturation = 1- 3 min (R, G, B)
locally adjusting each RGB pixel value to improve the R+G+B
visualization of structures in both darkest and lightest parts of the Intensity = 1 (R+G+B)
acute leukemia images at the same time. 3

IOutput (x,y) =255.[Iinput (x,y)-min]/(max-min)

Where, Ioutput (x, y) is the color level for the output pixel
Iinput (x, y) is the color level for the input pixel
max is the maximum value for color level in the input image.
min is the minimum value for color level in the input image.
2.2 Acute leukemia image segmentation based
2.3 Acute leukemia image segmentation based
on YUV color space
The YUV color space is derived from the RGB space. The Y
component determines the brightness of the color called as
luminance , while the U and V components determine the color
itself ie.chroma. Again this procedure involves conversion from
RGB to YUV and back conversion is required. The algorithm for

on HSI color space YUV color space is summarized as given below.

In HSI color space Hue is a color attribute which describes pure

1:Capture acute leukemia slide image at 40x magnification and
color where as saturation gives a measure of the degree to
save it into bitmap.
which a pure color is diluted by white light. Intensity component,
2:Extract the U-component information from original image.
I, is decoupled from the color information in the image. The
3: Plot the histogram to determine the threshold value.
segmentation of acute leukemia images using HSI color space
4: Select the threshold value using the U component of YUV
color space from the histogram.
5:Implement N x N (N=5) the median filter to the resulted
images.
6: Restore the YUV color to the resulted images.
7: Convert the resulted image to RGB & display the resulted
image
Transformation procedure is summarized as follows:
1)RGB to YUV conversion:
Y=(0.299*R)+(0.587*G)+(0.114*B);
U=(-0.14713*R)+(-0.28886*G)+(0.436*B);
V=(0.615*R)+(-0.51498*G)+(0.10001*B);
2)YUV to RGB conversion:
R=(0.7720*Y)+(-0.4633*U)+(1.1398*V);
G=(1.1161*Y)+(-0.1587*U)+(-0.5806*V);
B=(1.0000*Y)+(2.0321*U)+(-0.0000*V);
3. RESULT AND DISCUSSION
In this we study acute leukemia image segmentation based on
RGB, HSI and YUV color space. following figure1shows the
original acute leukemia image.
Fig1.Original Acute leukemia image
Fig2(a)Contrast stretched image
(b)LCS histogram
The above figure 2 shows the resultant contrast stretched image
after applying local contrast stretching on RGB color space.
contrast stretched image along with the histogram shows three
peaks. As the acute leukemia image which consist of: the
background, red blood cell and blast(cytoplasm and nucleus). Due
to this ,In this case, it is quite difficult to select the threshold value
which is suitable for the image.
Fig3(a)HSI image of acute leukemia
(b) S plane image
(c)S plane histogram
By observing the figure3 (a),(b),(c)which shows HSI image by
converting from RGB to HSI. Based on the s component threshold
is selected for the identification of the correct size of the blast and
it is shown in figure5(b)
Fig4 (a)YUV image of acute leukemia
(b) U plane image
(c)U plane histogram
Figure4shows the images after applying the YUV color space on
the original acute leukemia image .The resultant image in
figure4(a) containing YUV image. figure(b) shows U plane image
and figure(c)shows the resultant histogram of U plane image
.After deciding the appropriate threshold the resultant image is
shown in figure5(c).
For Figure 5 (a),shows after applying RGB based local contrast
stretching, this will result on the elimination of all cytoplasm blast
.By applying S component from HIS color space shown in
figure5(b).Based on U component the resultant blast are shown in
figure 5(c).

Fig5:(a) Segmented image of RGB with median filter
(b) Segmented image of HSI with median filter
(c) Segmented image of YUV with median filter
4.CONCLUSION
This paper shows a comparison of acute leukemia image
segmentation using HSI and RGB as well as YUV color spaces
for detection of blast ie the abnormal white blood cell in the blood
slide images. The method based on RGB YUV color space has not
performed well. For image segmentation using RGB and YUV
color space, Besides that ,the shape of the blast after the
segmentation process is not quite similar to the original blasts.
While the method based on HSI color space using Scomponent
can provide almost similar pixel values and shape to original
blasts. The results also show that the histogram is also useful to
select the threshold value using HSI color space. For proper
treatment management early detection of the disease is necessary
hence this result will be used for hematologists in their diagnostic
process and accelerating diagnosis of leukemia diseases..
5.REFERENCES
[1] Nor Hazlyna Harun, ,MY Mashor, ,NR.Mokhtar, ,Aimi alihah,
A.N, ,Rosline Hassan, R.A.A. Raof, ,MK Osman Comparison of
Acute Leukemia Image Segmentation using HSI and RGB Color
Space in ISSPA-2010
[2]N.H.AbdHalim,,M.Y.Mashor,A.S.AbdulNasir,N.R.Mok htar,
H.Rosline# Nucleus Segmentation Technique for
AcuteLeukemia in 2011 IEEE 7th International Colloquium on
Signal Processing and its Applications
[3] W. Skarbek and A. Koschan, "Colour Image Segmentation,"
Institute for Technical Informatics, Technical University of
Berlin, Berlin 1994.
[4] N.R.Mokhtar, Nor Hazlyna Harun, M.Y.Mashor,H.Roseline
,Nazahah Mustafa, R.Adollah ,H. Adilah, N.F.Mohd Nasir ,
Image Enhancement Techniques Using Local, Global, Bright,
Dark and Partial Contrast Stretching For Acute Leukemia Images
in proceedings of WCE-2009in London.
[5] Aimi Salihah, A.N. , M.Y.Mashor, Nor Hazlyna Harun, Azian
Azamimi Abdullah, H.Rosline, Improving Color Image
Segmentation on Acute Myelogenous Leukaemia Images Using
Contrast Enhancement Techniques (IECBES 2010),
[6] Akram A. Moustafa and 2Ziad A. AlqadiColor image
Rconstruction using new RGI model.in journals of computer
science-2009.
[7] Laurent Busin , Nicolas Vandenbroucke , and Ludovic
Macaire ,color spaces and image segmentationin july 2007.
[8] R. Adollah, M.Y. Mashor, N.F. M. Nasir, H. Rosline,
H.Mahsin, and H. Adilah, "Blood Cell Image Segmentation," in
FMBE. vol. 21, 2008, pp. 141-144.
[9] Aimi Salihah, A.N., M.Y.Mashor, Nor Hazlyna Harun,
H.Rosline Colour Image Enhancement Techniques for Acute
Leukemia Blood Cell Morphological Features in 978-1-4244-
6588-0/10/$25.00 201 0 IEEE
[10] R.c. Gonzalez and R. E. Woods, Digital ImageProcessing, 3
ed: Prentice Hall, 2007.
[11] Roy A. Dimayuga1, Gerwin T. Ong2, Rainier Carlo S.
Perez3, Gefferson O. Siy4, Saman C. Sohrabi Langroudi5,and
Miguel O. Gutierrez6 Leukemia detection using digital image
processing in matlab
[12]eMedicinehealth.com Leukemia. Retrieved March 15, 2010
fromhttp://www.emedicinehealth.com/leukemia/article_em. html