A Novel MRI Brain Edge Detection Using

PSOFCM Segmentation and Canny Algorithm

Romesh Laishram
Electronics & Communication Engineering.
Manipur Institute of Technology, Imphal, Manipur, India
E-Mail: [email protected]
Wahengbam Kanan Kumar, Anshuman Gupta, Khairnar Vinayak Prakash
M.Tech (VLSI Design), Dept. of Electronics and Communication Engineering
Maharishi Markandeshwar University, Ambala, India
Email: [email protected], [email protected], [email protected]

Abstract— Introduction of many Image processing and Edge detection and clustering are two basic segmentation
segmentation tools has undoubtedly presented the procedures methods among the various techniques. Edge detection is
of mapping the human brain in a more efficient way. This basically aimed at identifying the image brightness
paper attempts to pull out a new and a practical approach for discontinuities especially along the edges where the intensity
enhancing the underlying delicate architectures of the human tends to change sharply. Canny edge detection is a premium
brain images captured by a Magnetic Resonance Imaging technique used for detecting the edge in an image. Clustering
(MRI) machine in a much better way. Edge detection is a is a process whereby a data set is replaced by clusters, which
fundamental tool for the basic study of human brain are collections of data points that "belong together". Thus, it
particularly in the areas of feature detection and feature is natural to think of image segmentation as image clustering
extraction. The edge detection methodology presented in this
i.e. the representation of an image in terms of clusters of
paper relies on two basic stages: Firstly, the original MRI
pixels that "belong together". The specific criterion to be
image is subjected to image segmentation which is done using
Particle Swarm optimization incorporating Fuzzy C Means
used depends on the application. Pixels may belong together
Clustering (PSOFCM) technique. And secondly, canny edge because of the same colour or similarity measure. This work
detection algorithm is used for detecting the fine edges. is an enhanced form of edge detection methodology that aids
After implementation it was found that this technique yields to obtain the best results out of MRI brain images. The
better edge detected image of the human brain as compared to method uses fuzzy C means (FCM) [2] clustering algorithm
other edge detection methods as discussed below for segmenting the image prior to edge detection process.
The main purpose of using PSO is to reach the Global
Keywords—Edge ,MRI,PSOFCM,Canny,segmentation. minima of the clustering objective function. In [3] the
application PSO in image segmentation problem is
I. INTRODUCTION investigated.
Magnetic Resonance Imaging (MRI) or nuclear magnetic
resonance imaging (NMRI) or magnetic resonance The “Particle Swarm Optimization” model was brought
tomography (MRT) is a medical imaging technique used in into light for the first time by Russel Ebenhart and James
radiology to visualize internal structures of the body in Kennedy in 1995. As the name suggest the basic operation
detail. It makes use of the property of nuclear magnetic simulates the behaviour of flocks of birds or the sociological
resonance (NMR) to image nuclei of atoms inside the body. behaviour of a group of people or schooling of fish.
It provides excellent contrast between the different soft Modelled on the mechanism of evolution and natural
tissues of the body, which makes it especially useful in movement this algorithm provides an alternative tool for
imaging the brain, muscles, etc. The images produced by an locating optimal solutions. Here swarm of several particles
MRI scanner are best described as slices through the brain. are used as the population size to find the best solution which
MRI has an added advantage of producing images, which is analogous to that of Genetic Algorithm in which some
slices the brain images in both horizontal and vertical planes. probable solutions are initialises as the population size. Each
For further processing, these MRI images must be particle keeps track of their current positions which
categorized and analysed using edge detection. correspond to potential solutions of the function to be
minimized. Each particle also keeps track of the speed and
Image segmentation is an attempt to partition a digital direction of travel by the particle. Similar to genetic
image into multiple segments which is also known as sets of algorithm a fitness value is also associated with each particle
pixels or super pixels. The goal of segmentation is to which directly depends on the particle’s current position.
simplify or change the image into something that is more Additionally each particle also remembers its personal best
meaningful and easier to analyse. positions which on comparison with the personal best
solutions of other particle are used to determine the overall
best solution or population. Moreover, this algorithm is also III. PARTICLE SWARM OPTIMIZATION
metaheuristic as it makes few or no assumptions about the INCORPORATING FUZZY C MEANS (PSOFCM)
problem being optimized and can search very large spaces of PSO algorithm works by having a population (called a
candidate solutions. To be more specific it does not use swarm) of candidate solutions (called particles).These
gradient of the problem being optimized, which means PSO particles are moved around in the search-space according to
does not require that the optimization problem being a few simple formulae. The movements of the particles are
differentiable as is required by classic optimization methods. guided by their own best known position in the search-space
So it can therefore be used on problems that are partially as well as the entire swarm's best known position. When
irregular, noisy, change over time, etc. improved positions are being discovered these will then
II. FUZZY C MEANS ALGORITHM come to guide the movements of the swarm. The process is
repeated and by doing so it is hoped, but not guaranteed, that
Segmentation is greatly being improved by using the a satisfactory solution will eventually be discovered.
FCM algorithm [4]-[6] instead of using K-Means Clustering The mathematical steps used to formulate the PSOFCM
algorithm. It divides the images into number of homogenous
algorithm is discussed below:
classes effectively. It has some success to detect the noise
from an image. Consider a D-dimensional search-space S 
 D and a swarm
The Traditional FCM algorithm is an iterative algorithm consisting of I particles. The i-th particle is in effect a D-
that produces optimal C partitions, centers V= {v1, v2,…, dimensional vector:
vc}. Let unlabelled data set X={x1, x2,…, xn} be the pixel Xi = (xi1, x i2, x iD)T  S, (in our program we initialized the
intensities, where n is the number of image pixels to
determine their membership. The FCM algorithm tries to number of particles in the swarm or swarm size as 20)
partition the dataset X into C clusters. The standard FCM The velocity of this particle is also a D-dimensional vector,
objective function is defined as follows. Vi = (vi1, v i2,……, v iD)T 

c n m 2 The best previous position encountered by the i-th particle
J m (U, V)    (u ik ) d (x k , v i ) (1)
i1k 1 in S is denoted by:
2 Yi = (yi1, y i2,……, y iD)T 

where d (x k , v i ) represents the square of the Euclidean
These particles’ initial velocities and positions are randomly
distance between the pixel intensity value xk and the
initialized. Let Y* be the global best position amongst all the
c
centroid value vi along with constraint  u ik  1 ,and the particles, and t be the program iterations. During each
i1 iteration, the velocity and position of each particle/swarm is
degree of fuzzification m≥1.A data point xk belongs to the
updated using (4) and (5) as given below.
specific cluster vi that is given by the membership value uik
Further in this algorithm we are trying to minimize a
of the data point to that cluster. Local minimization of the
objective function J m (U, V) is accomplished by repeatedly function Yi = f (Xi) using Fuzzy c means algorithm. The
adjusting the values of uik and vi according to the following objective function of the fuzzy c means clustering algorithm
equations. is used for achieving the global minima of the clustering

1  1
objective. In this way by minimizing the objective function



value FCM keeps track of the best particle or best position

c  d 2 x , v  (m1)  ever.
U ik 
 k i  (2)

 d (x k , v j ) 

 2
j 1 

  Vi(t+1) = W Vi(t)+c1r1(Yi(t) - Xi(t)) + c2r2(Y* - Xi(t))

Where Vi is calculated using the following equation
n and Xi (t+1) = Xi (t) + Vi (t+1)

m
 (u ik ) x k Where
Vi  k  0 (3)
n Yi = Yi, if f (Xi) ≥ f (Yi)
m
 (u ik ) = Xi, if f (Xi) < f (Yi)

k 0
As Jm is iteratively minimized, the centroid matrix is more
stable. This iteration is terminated when the difference Y* 
0, Y1, ….. Yi} such that
between the maximum of current centroid value, maximum f (Y*) = min (f (Y0), f (Y1),……., f (Yi))

of previous iteration centroid value is less than the 0.0001.
The value 0.0001 is predefined termination threshold. r1 and r2 are elements from two uniform random sequences
Finally, all homogeneous pixels are grouped into the same
in the range (0,1).W is called inertia weight matrix
class to evaluate the Fuzzy C-means algorithm.
representing weights of Vi(t) as a contribution to Vi(t+1).
It is quite clear from (7) that Y* is the global best position should not create false edges.
amongst all the particles. The value of each component of The resulted final images with their best edges details are
very much essential and helpful for further brain MRI image
Vi vector is clamped to range [-vmax, vmax] to prevent
processing and analysis.
PSOFCM from leaving the search space. The main idea of this present work is the application of PSO
c 1 and c2 are acceleration co-efficient which controls the based optimization in FCM problem. However for medical
displacements of a particle in a single iteration. image processing the desired result should be highly
In this work we used W = 0.5 and c 1 = c2 = 2 objective. The present work may be use as a preprocessing
stage for further analysis.
IV. SIMULATION RESULTS
In this section we present the simulation results of the
MRI image edge detection problem. The performance of
PSOFCM (with canny) based edge detection algorithm is
REFERENCES
compared with the GAFCM (with canny) for four different
MRI images as shown in figure 1-4 respectively. The [1] Romesh Laishram, W.Kanan Kumar Singh, N.Ajit Kumar,
Robindro.K, S.Jimriff, “MRI Brain Edge Detection Using GAFCM
simulation is performed in MATLAB environment. Segmentation and Canny Algorithm”, International Journal of
Advances in Electronics Engineering – IJAEE,volume 2 - Issue 3,
The comparison result shows that PSOFCM based canny ISSN:- 2278-215X, pp. 168-171,December 8,2012
edge detection method yields better result than GAFCM [2] Rafael C. Gonzalez, Richard E.Woods, “Digital Image Processing”,
based canny edge detection. It can also be seen that more Pearson Education, Second Edition, ISBN 81-7758-168-6, 2005.
number of edges are also fetched and the image is also [3] Amit Konar, “Computational Intelligence Principles, Techniques and
segmented and clustered properly as compared to GAFCM Applications”, Springer edition, ISBN 3-540-20898-4 Springer Berlin
Heidelberg, New York
based canny edge detection method; each and every edge is
fetched without missing even a single curve. This new [4] R.Venkateswaran, S.Muthukumar, “Genetic Approach on Medical
Image Segmentation by Generalized Spatial Fuzzy C- Means
algorithm goes even to the most delicate parts of the brain Algorithm”, 2010 IEEE International Conference on Computational
for extracting the fine edges and shows astounding Intelligence and Computing Research, pp: 210—213..
performance. This shows that the novel technique that is [5] Yingjie Wang, “Fuzzy Clustering Analysis Using Genetic
being incorporated in this paper (i.e. PSOFCM based canny Algorithm”, and ICIC International @ 2008 ISSN 1881-803 X, pp:
edge detection) is the best edge detection methodology for 331—337.
MRI Brain image edge detection as compared to GAFCM [6] A.Halder , S.Pramanik , A.Kar, “Dynamic Image Segmentation using
based canny edge detection technique [1]. Fuzzy C-Means based Genetic Algorithm”, International Journal of
Computer Applications (0975 – 8887) Volume 28– No.6, August
2011, pp: 15 – 20.
V. CONCLUSION
[7] Huang Ying, Wang Weixing, and Li Weisheng, “Anisotropic Filter
In this paper we have presented PSO based optimization Based Modified Canny Algorithm”. Fourth International Conference
of FCM algorithm i.e. PSOFCM and its application in human on Fuzzy Systems and Knowledge Discovery, Vol: 1, pp: 736 -740,
Brain MRI image segmentation for automatic detection of Aug 2007.
abnormalities. The segmented image is further processed for [8] Natarajan.P, “New Fangled MRI Brain Edge Detection Using
Enhanced Canny Algorithm”, 2011 IEEE International Conference on
edge detection using canny edge detector. The result Computational Intelligence and Computing Research, pp: 84—87.
obtained through PSOFCM yields better edge detected image
[9] Davoud Sedighizadeh and Ellips Masehian, “Particle Swarm
compared to GAFCM segmentation [1]. It may also be noted Optimization Methods, taxonomy and applications”, International
it was proven that GAFCM based canny edge detection is Journal of Computer Theory and Engineering (1793-8201) Vol. 1,
better than ordinary canny edge detection. No. 5, December,2009, pp: 486-502
The whole comparison is based on canny edge [10] J. Kennedy and R.C. Eberhart, "Particle swarm optimization" ,
detection’s principles: Proceeding of the 1995 IEEE International Conference on Neural
Networks (Perth,Australia), IEEE Service Centre, Piscataway, NI,
1) Good detection: the algorithm should mark as many (1995), Iv: 1942-1948
real edges in the image as possible.
2) Good localization: edges marked should be as close
as possible to the edges in the real image.
3) Minimal response: a given edge in the image should
only be marked once,and where possible,image noise
 (a) (b) (c)
Figure 1. (a) Original Image (b) PSOFCM with Canny algorithm (c) GAFCM with Canny algorithm

(a) (b) (c)

Figure 2. (a) Original Image (b) PSOFCM with Canny Algorithm (c) GAFCM with Canny algorithm

(a) (b) (c)

Figure 3. (a) Original Image (b) PSOFCM with Canny Algorithm (c) GAFCM with Canny algorithm

(a) (b) (c)

Figure 4. (a) Original Image (b) PSOFCM with Canny Algorithm (c) GAFCM with Canny algorithm