International Journal of Trend in Scientific Research and Development (IJTSRD)

Volume 5 Issue 5, July-August 2021 Available Online: www.ijtsrd.com e-ISSN: 2456 – 6470

Utilization of Super Pixel Based Microarray Image Segmentation

Mr. Davu Manikanta, Mr. Parasurama N, K Keerthi
Department of ECE, JNTUK University College, PPDCET Vijayawada, Andhra Pradesh, India

ABSTRACT How to cite this paper: Mr. Davu

In the division of PC vision pictures, Super pixels are go probably as Manikanta | Mr. Parasurama N | K
key part from 10 years prior. There are various counts and Keerthi "Utilization of Super Pixel
methodology to separate the Super pixels anyway whole all of them Based Microarray Image Segmentation"
the best super pixel looking at strategy is Simple Linear Iterative Published in
International
Clustering (SLIC) have come to pivot continuously recently. The
Journal of Trend in
concentrating of small scale group quality verbalization from MRI Scientific Research
imaging is more useful to perceive tumors or some other dangerous and Development
development contaminations, so the fundamental DNA (cDNA) (ijtsrd), ISSN: 2456-
microarray is a grounded device for analyzing the same. 6470, Volume-5 | IJTSRD46274
Issue-5, August
The division of microarray pictures is the essential development in a
2021, pp.2101-2105, URL:
microarray assessment. In this paper, we proposed a figuring to www.ijtsrd.com/papers/ijtsrd46274.pdf
dividing the cDNA small show picture using Simple Linear Iterative
Clustering (SLIC) based Self Organizing Maps (SOM) method. In Copyright © 2021 by author (s) and
any case, the proposed figuring is taken up a moving task to look at International Journal of Trend in
the bad quality of pictures in addition. There are two phases to Scientific Research and Development
separate the image, introductory, a pre-setting up the applied picture Journal. This is an
to diminish fuss levels and second, to piece the image using SLIC Open Access article
based SOM approach. distributed under the
terms of the Creative Commons
KEYWORDS: MRI, DNA, SLIC, Super pixels, SOM, Quality Attribution License (CC BY 4.0)
clarification microarray (http://creativecommons.org/licenses/by/4.0)

I. INTRODUCTION
A super pixel is a heap of basic proximal and acumen. To locate the basic attributes for a specific
homogeneous pixels. The homogeneous would be goal is a basic district of assessment. These attributes
portrayed similar to measure, significance of are called illuminating qualities. The divulgence of
concealing, significance of surface, etc The Super enlightening attributes is essential to the expert for
pixels are known to save the close by picture credits, judge a patients and for the affiliation that are making
for instance, object cutoff points, shape and area, and drugs over the a few years, a ton of exertion has been
decrease the cost of assessment of various PC vision set in the improvement of answer for the informative
division issues. Along these lines, super pixel over qualities openness. Till now the undertaking is
division capably decreases the no. of units to manage inconceivably trying and some phenomenal strategies
an image. The SOM approach is best anytime to think are set up to beat the standard Approaches. The
about the image pixels and here we are introducing microarray picture assessment.
another system of microarray picture division using
Feature confirmation when utilized for microarray
SLIC over SOM. The critical central focuses for
quality clarification information is called quality
using this advancement are; to handle features with
choice. Beside scrutinize of dimensionality there are
more significant regions and to diminish the data
different issues looked in quality affirmation like
objects for the sub-sequent estimations.
mislabeled information, dull information, pointless
Quality clarification microarray (GEM) tests amass and crazy information, and issue of cross-stage
fundamental trademark data gathering regular appraisals, wrong and tendency issue and burden in
information from tests like tissues, cell lines recorded typical data recovery. Particular quality choice
GEM information hold quality data over all models in strategies and assessments are proposed recorded as a
the insight. As a matter of fact incalculable qualities printed copy which can diminish the dimensionality
is assessed and recorded simultaneously. In different by removing unessential, tedious and clamorous
viewpoints these models can be diverse under attributes.

@ IJTSRD | Unique Paper ID – IJTSRD46274 | Volume – 5 | Issue – 5 | Jul-Aug 2021 Page 2101
 International Journal of Trend in Scientific Research and Development @ www.ijtsrd.com eISSN: 2456-6470
1.1. Related work EXISTING NOISE REMOVAL METHODS
Basic direct iterative social affair (SLIC) is an Mean Filter
embrace of means for Super pixels age with two Median Filter
colossal partitions like the no. of distance calculations Vector Filtering techniques
in the overhaul is profoundly diminished by the Vector Median Filter
multifaceted nature to be straight in the measure of Vector directional filter
pixels N and independent of the no. of Super pixels k Directional Distance Filter
and the other one is a weighted distance survey Center weighted Vector median filter
consolidate the tone and spatial proximity while Adaptive Center Weighted Vector Median Filter
simultaneously giving request over the size of pixel Wavelets based filtering
and strength of the Super pixels.” The “section of Empirical Mode Decomposition
camouflaging picture has been wound up being DISADVANTAGES OF EXISTING WORK
irritating considering the way that it combines a
1. Noise Removal
goliath degree of data arranging. Regardless of the Highly depending on the intensity characteristics
way that stunning undertakings have been given to, a of the image.
few issues are currently not totally tended to. Dong
and Xin master watched out for a blend system which 2. Segmentation
circuits solo division and encouraged division. The 1. The need of human intervention and correct the
free division is refined by a two-level methodology, potential alignment and rotation problems.
i.e., lesser covering and concealing get-together. The 2. Sensitive to contaminations and large number of
managed division joins covering learning and pixel missing spots.
gathering. Reenacted propping (SA) has been used for 3. Parameters about the sub-array and spots are
finding the ideal packs shape SOM models". The required ie., number of spots in each row and
calculation of SLIC Super pixels age is given under. column.
4. Segmentation using clustering algorithm is carried
In the past, scientists have only been able to conduct out using single feature.
these genetic analyses on a few genes at once. With 5. No post processing is required.
the development of DNA microarray technology,
however, scientists can now examine how active PROBLEM STATEMENT
thousands of genes are at any given time. NOISE REMOVAL:
If we use VMD+DWT Technique then we can
The functional organization of gene cell has been perform Computationally efficient, Rotationally
investigated by the developed tools in large scale symmetric(perform the same in all directions)
analysis leads to the success of developing many
techniques in bioengineering. Out of these techniques, SEGMENTATION
DNA microarray technology has developed more Clustering Algorithms are not restricted to a
powerful by enabling the biologists to investigate particular spot size and shape, does not require an
thousands of DNA genes concurrently that lies in the initial state of pixels and no need of post
entire organism. processing.
2. CREATION OF MICROARRAY IMAGE MULTIPLE FEATURE CLUSTERING
SEGMENTATION
In the microarray image segmentation problem,
not only the pixel intensity, but also the distance
of pixel from the center of the spot and median of
intensity of a certain number of surrounding pixels
influences the result of clustering.
Based on this observation, in this research,
multiple feature fuzzy c-means clustering
algorithm is proposed, which utilizes more than
one feature.
PROPOSED WORK
Noise Removal:
Variational Mode decomposition and Wavelet
Figure 1: Microarray Image Creation Process based Filtering

@ IJTSRD | Unique Paper ID – IJTSRD46274 | Volume – 5 | Issue – 5 | Jul-Aug 2021 Page 2102
 International Journal of Trend in Scientific Research and Development @ www.ijtsrd.com eISSN: 2456-6470
Segmentation:
Multiple feature clustering
Super pixel-based segmentation methods
NOISE REMOVAL ALGORITHM USING VMD
WAVELETC

Figure 2: Noise removal using VMD + Wavelet

FLOW DIAGRAM OF FILTERING USING
VMD+ WAVELET
VMD: Variational mode decomposition VMF:
Vector Median Filter
NOISE REMOVAL USING VMD-DWT
METHOD
PSNR(Peak signal to noise ratio)
PSNR, is the ratio between the maximum possible
power of a signal and the power of corrupting noise

Here, MAXf is the maximum possible pixel value of the image

PSNR(Peak signal to noise ratio) VALUES OF FILTERING METHOD
The value of σ denotes the standard deviation value used in imnoise() function in matlab to add
Gaussian noise to microarray image
Method σ = 0.015 σ = 0.025 σ = 0.036
Wavelet(Universal Threshold) 22.02 16.98 14.89
Wavelet(SURE shrink) 21.90 16.11 13.76
BEMD (Universal Threshold) 34.11 28.86 22.86
BEMD (SURE shrink) 35.89 29.74 23.11
VMD+ Wavelet
36.16 30.98 24.43
(Universal Threshold)
VMD+ Wavelet (Sure Shrink) 38.91 32.34 26.98

QUALITATIVE ANALYSIS Perform Edge detection using Bidimensional

Empirical Mode Decomposition
Perform morphological filling on the edge.
QUALITATIVE ANALYSIS

Fig3 :Sub-gridding
AUTOMATIC SPOT DETECTION
ALGORITHM
The steps of the automatic spot detection algorithm
Figure 4: Spot Detection
are as follows:
Consider the sub-gridded microarray image.

@ IJTSRD | Unique Paper ID – IJTSRD46274 | Volume – 5 | Issue – 5 | Jul-Aug 2021 Page 2103
 International Journal of Trend in Scientific Research and Development @ www.ijtsrd.com eISSN: 2456-6470
SEGMENTATION (SINGLE FEATURE
CLUSTERING ALGORITHMS
K-means
Moving K-means
Fuzzy C-means
QUANTITATIVE VALUES OF SINGLE
FEATURE CLUSTERING ALGORITHMS

SEGMENTATION USING MULTIPLE

FEATURE CLUSTERING ALGORITHM
In microarray image segmentation, the position of
the pixel and median value of surrounding pixels
also influences the result of clustering and
subsequently that leads to segmentation.
Based on this observation, multiple feature
clustering algorithm is developed for segmentation
of microarray images
QUANITATIVE VALUES OF MULTIPLE
FEATURE FCM COMPARED WITH SINGLE
FEATURE K_MEANS AND FCM
MSE Values MSE Values
Method Compartment Compartment
No 1 No 8
K-means 282.781 346.47
Fuzzy c-means
(Single 216.392 228.69
Feature)
Multiple
feature Fuzzy 198.327 186.276
C-means
TABLE 3: Multiple feature clustering algorithm
SEGMENTATION USING MULTIPLE
MSE Values
FEATURE CLUSTERING ALGORITHM
In microarray image segmentation, the position of SUPER PIXEL BASED FCM SEGMENTATION
the pixel and median value of surrounding pixels
also influences the result of clustering and
subsequently that leads to segmentation.
Based on this observation, multiple feature
clustering algorithm is developed for segmentation
of microarray images

@ IJTSRD | Unique Paper ID – IJTSRD46274 | Volume – 5 | Issue – 5 | Jul-Aug 2021 Page 2104
 International Journal of Trend in Scientific Research and Development @ www.ijtsrd.com eISSN: 2456-6470
CONCLUSION [6] Stamos Katsigiannis, Eleni Zacharia, Dimitris
The microarray image analysis was done in three Maroulis, “Grow-Cut Based Automatic cDNA
stages: Noise reduction,, segmentation and intensity Microarray Image Segmenta-tion” IEEE
extraction. Existing microarray analysis mechanisms Transactions on Nanobioscience, vol. 14, no. 1,
are semi-automatic in nature, requiring human January 2015.
intervention, initialization of parameters etc. In this https://doi.org/10.1109/TNB.2014.2369961.
work we are developed a truly automatic system for [7] Rajendra Nagar, Shanmuganathan Raman,
microarray image analysis.
“SymmSLIC Symmetry Aware Superpixel
New algorithms are developed at each and every stage Segmentation” The IEEE Internation-al
of microarray analysis. In this research, a new Conference on Computer Vision (ICCV), 2017,
algorithm for noise removal in microarray image is pp. 1764-1773. https://doi.org/10.1109/ICCVW.
developed without effecting the edge information of 2017.208.
spot. For segmentation, algorithms based on multiple [8] S. Shirakawa, T. Nagao. “Evolutionary image
feature clustering and super pixels based segmentation
segmentation based on multi objective
are developed. clustering”. In Proceedings of IEEE Con-gress
REFERENCES on Evolutionary Computation, IEEE,
[1] Durga Prasad Kondisetty, Dr. Mohammed Ali Trondheim, Norway pp. 2466–2473, 2009.
Hussain, “A Robust Method for Reducing
[9] Rezaul Karim, Md. Khaliluzzaman, Sohel
Image Noise in Microarray Images International Mahmud, “A Review of Image Analysis
Journal of Pure and Applied Mathematics,
Techniques for Gene Spot Identification in
Volume 117 No. 19 2017, 441-447.
cDNA Microarray Images” Next Generation
[2] [Durga Prasad Kondisetty, Dr. Mohammed Ali Information Tech-nology (ICNIT), 2011 The
Hussain, “ Gene Microarray Analysis using Self 2nd International Conference.
Organizing Maps” Interna-tional Journal of
[10] Maziidah Mukhtar Ahmad, Asral Bahari
Pure and Applied Mathematics, Volume 117
Jambek, Mohd Yusoff bin Mashor, “A Study on
No. 19 2017, 441-447.
Microarray Image Gridding Tech-niques for
[3] Xudong Jin and Yanfeng Gu, “Superpixel- DNA Analysis” 2nd International conference on
Based Intrinsic Image Decomposition of electric design (ICED) August 19-21 2014,
Hyperspectral Images” IEEE Transactions on Malaysia. [11] Revathi T, Sathish A, Sumathi P,
Geoscience and Remote Sensing, vol. 55, no. 8, “Microarray Analysis us-ing Fuzzy C-Means
August 2017. Clustering Algorithm”. International Journal of
https://doi.org/10.1109/TGRS.2017.2690445. Innovative Research in Electrical, Electronics,
[4] Guifang Shao, Tingna Wang, Wupeng Hong, Instrumentation and Control Engineering Vol.
Zhigang Chen, “An Improved SVM Method for 2, Issue 1, Jan 2014.
eDNA Mieroarray Image Segmentation”. The [11] David Moena Q, “Microarray Image Gridding
8th International Conference on Computer Sci- by using Self-Organizing” IEEE transactions
ence & Education (ICCSE 2013) April 26-28, 978-1-4244-5089- 2011. [13] R. M. Farouk, M.
2013. Colombo, Sri Lanka. A. SayedElahl, “Robust cDNA microar-ray
image segmentation and analysis technique
[5] Natarajan P, Sushmitha G, “Brain Tumor
based on Hough circle transform” International
Detection in MRI Brain Images using Threshold
Journal of Pure and Applied Mathematics,
Operation” International Journal of Pharmacy &
Volume 117 No. 19 2017, 441-447.
Technology, ISSN: 0975-766X, August-2016.

@ IJTSRD | Unique Paper ID – IJTSRD46274 | Volume – 5 | Issue – 5 | Jul-Aug 2021 Page 2105