Segmentation of anotomical regions of brain is fundamental problem in medical image analysis.Although edge information is the main clue in image segmentation, it can't get a better result in analysis the content of images without combining other information. The segmentation of brain tissue in the magnetic resonance imaging (MRI) is very important for detecting the existence and outlines of tumors.
While surveying the the literature,it has found that no so many work has been done in segmentation of brain tumor based on bilateral symmetry information in MATLAB Envirntreatmentment. In the paper, an algorithm about segmentation based on the symmetry character of brain MRI image has been developed on 2D MRI image. Our goal is to detect the position and boundary of tumors automatically. Experiments were conducted on real pictures, and the results show that the algorithm is flexible and convenient.
KEYWORDS: Brain tumor, Magnetic Resonance Imaging(MRI), MATLAB, Image Segmentation
Get your grade
or your money back
using our Essay Writing Service!
The body is made up of many types of cells. Each type of cell has special functions. Most cells in the body grow and then divide in an orderly way to form new cells as they are needed to keep the body healthy and working properly. cells lose the ability to control their growth, they divide too often and without any order. The extra cells form a mass oftissue called a tumor. Tumors are benign or malignant.The aim of this work is to design an automated tool for brain tumor quantification using MRI image data setsSsegment the tumor in the brain make suegeon able to see the tumor and then ease the treatment.
During the image processing, edge information is the main clue in image segmentation. But, unfortunately, it can't get a better result in analysis the content of images without combining other information. So, many researchers combine edge information with some other methods to improve the effect of segmentation  . Nowadays, the X-ray or magnetic resonance images,CT Scan,ultrasound has became irreplaceable tools for tumors detecting in human brain and other parts of human body . Although MRI is more expensive than the X-ray inspection, the development of its applications becomes faster because of the MR inspection does less harm to human than X-ray's.
Segmentation of medical images has the significant advantage that interesting characteristics are well known up to analysis the states of symptoms. The segmentation of brain tissue in the magnetic resonance imaging is also very important for detecting the existence and outlines of tumors. But, the overlapping intensity distributions of healthy tissue, tumor and surrounding edema makes the tumor segmentation become a kind of work full of challenge.
In this paper, we make use of symmetry character of brain MRI to obtain better effect of segmentation. Symmetry is one of the most important characteristics of vision. It is a fast and high level first approach to object understanding. On Earth, because of the gravity, the bilateral symmetry is the most important as it is necessary to maintainobjects equilibrium. Most of the human-made objects have abilateral symmetry. Moreover, humans and other beings as fishes, animals, birds, insects have a bilateral symmetry
too.In particular reference to humans, symmetry is important because it can be a sign of disease. If the human body bilateral symmetry is not respected, that is most of the time due to some abnormalities.
Our goal is to detect the position and boundary of tumours automatically based on the symmetry information of MRI. For detection of tumour in 2D the software used is MATLAB. .Also, a Graphical User Interface (GUI) has been designed, which is user friendly environment to understand and run the work done by the one click of a mouse. This user friendly graphical user interface (GUI) was developed with the help of MATLAB.The rest of the paper is organized as follows.
The Section 2 introduces some related works. Our algorithm is presented at section 3. And, section 4 gives some experiment results. Section 5 is our conclusion.
2. Related works
In most of time, the edge and contrast of X-ray or MR image are weakened, which leads to produce degraded image. So, in the processing for this kind of medic image, the first stage is to improve the quality of images. Many researchers have developed some effective algorithms about it  .
Always on Time
Marked to Standard
After the quality of image been improved, the next step is to select the interesting objects or special areas from the images, which is often called segmentation. Many techniques have been applied on it. In this paper, we mainly discuss the brain tumour segmentation from MRI.
For now, there are also some very useful algorithms, such as mixture Gaussian model for the global intensity distribution , statistical classification , and texture analysis
, neural networks  and elastically fitting boundaries
, etc. An automatic segmentation of MR images of normal brains by statistical classification, using an atlas prior for initialization and also for geometric constraints was introduced in .
Even through, Brain tumours is difficult to be modelled by shapes due to overlapping intensities with normal tissue and/or significant size. Although a fully automatic method for segmenting MR images presenting tumour and edema structures is proposed in  , but they are all time consuming in some degree.
As we know, symmetry is an important clue in image perception. If a group of objects exhibit symmetry, it is more likely that they are related in some degree. So, many researches have been done on the detection of symmetries in images and shapes   .
In our applications, we developed an algorithm based on bilateral symmetry information of brain MRI. If the human body bilateral symmetry is not respected, that is most of the time due to some abnormalities. For example, the symmetry measurement can aid in the detection of breast cancers  or neurological disorders . Asymmetry was also used brain tumors detection on MRI images .
Our purpose is to detect the tumour of brain automatically. Compared with other automatic segmentation methods, more effective the system model was constructed and less time was consumed
Our algorithm composes of three steps. The first is to define the bilateral symmetrical axis. The second is to detect the region of brain tumour and the third to segment brain tumor region.
Symmetry axis defining
The first step of our algorithm is mainly based on symmetry character of brain MRI. The bilateral symmetry character is very obviously in four MR Images of brain presented in Figure.1.
Figure.1. The bilateral symmetry character is very obviously.
If without tumour in the brain or the size of tumor is very small, the symmetry axis can be defined with a straight line x = k,(y >= 0) , which separates the image into two bilateral symmetry parts, show as Figure.2.
Figure.2. The bilateral symmetry axis is defined with a straight line.
This kind of symmetry is not very strictly. And, compared with normal brain MRI, the symmetry characteristic is distorted for the existing of brain tumour, such as the circumstance shown in Figure.3.a.
Figure.3. The symmetry axis can't be defined with a straight line in the brain MRI with tumours, so a curve line is more convenient to describe it.
For more convenient to describing symmetry axis, a curve line ( y = f (x), x > 0, y > 0 ) is defined, which is shown in Figure.3.b.
At first, we get the edge map of source image like Figure.5.b. From the edge map, the edge point set Pe can be obtained. And then, we calculate the edge-centroid Gi of every line according to equation (1).
Gi = 1/k ∑ p i, j p i, j Ïµ Pe (1)
where, Gi is the abscissa of centroid in the ith line, k is the edge point number in the ith line, whose abscissas are pi,1,....,pi,k . So, based on the edge-centroids, we can use the least square method to get the symmetry curve line y approximatively.
Automatic brain tumor detection
After the bilateral symmetry axis is defined, we can go alone the segmentation of brain tumour.
Suppose I (x, y) representing the MRI plane, we use dx= ∂I / ∂x, dy=∂I / ∂y to represent the grads in two directions. Based on the dx ,dy , we can get the edge map E(x, y) of MRI. Then, the edge map E(x, y) is covered with a set of grids, show as Figure.4. For every pixel (x, y) , it is belong to the grid g(x, y) , and at the centre of g (x, y)
This Essay is
a Student's Work
This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.Examples of our work
Pixel (x,y) at center of mesh g(x,y)
Figure.4. A set of grids used to cover on the edge map; from left to right, the sizes of meshes decrease.
At first, using grid with biggest meshes, we calculate the grid-grads ggE of each pixel of E(x, y) with equation (2).
ggE (x,y)= ∑ (di+dj) (2)
(i, j)≡g(x, y)
Then, we calculate the contrast Contra of every two bilateral symmetry points with equation (3), in which (x, y) and (x', y') are two bilateral symmetry points
Contra(x, y) = (ggE (x', y') - ggE (x, y))/R
Contra(x', y') = (ggE(x, y) - ggE (x', y'))/R (3)
R= ggE(x, y) + ggE (x', y') (4)
For those pixels, whose Contra is below the threshold δ , we use the grid with smaller meshes to recalculate from equation (2) to equation (4) until the size of meshes is the smallest.
After the iterations have finished, we calculate the E(x, y) again with equation (5).
E(n) (x, y) = Contra(x, y) * E(n−1) (x, y) (5) where n is the reiteration time.
Then, all of edges within the regions having symmetry characteristic will be weakened. In other words, the more symmetrical the two regions have, the more the edges are weakened. At the same time, the edges not symmetrical are enhanced.
In the end, according to the enhancing effect, the unsymmetrical regions can be detected, which is caused by brain tumour.
Take the MRI image from database in MATLAB envirnment.Decide the size of image.Convert the image from RGB to gray image.
4.2 Edge Detection:-
Take the image in double precision.Find out gradient of images.Find out edges using canny edge detection.See edge detection image.
4.3 Find Mean points of image:-
Labels all edge regions.Find Fitting out edge centroids (Mean points) by using following formula
Mean Point =∑yi/no.of pixel detected
4.4 Fitting of mean points(curve fitting):-
Use least square method to fit the curve.2nd degee polynomial regression is used to fit the curve.
4.5 Generate the 100 indices at +ve direction of axis and 100 indices at -ve direction of axis.
4.6 Find out the Grid grad
GG(x,y)=(Gradient in +ve direction)+(gradient in -ve direction)
4.7 Find R
R=Sum of Grid gradient in both direction
4.8 If R~=0
Find out contra of every two bilateral symmetrical points as follows
ContraLR=[(Grid Grad in +ve Direction )- (Grid Grad in -ve direction)]/R
ContraRL=[( Grid Grad in -ve Direction)-( Grid Grad in +ve direction)]/R
4.9 Find out contra is below threshold or not.
4.10 If Grater than threshold fault is detected
`4.11Then, all of edges within the regions having symmetry
characteristic will be weakened. In other words, the more
symmetrical the two regions have, the more the edges are
weakened. At the same time, the edges not symmetrical are
In the end, according to the enhancing effect, the
unsymmetrical regions can be detected, which is caused by
GUI used in experiment is as shown in Fig
Figure.5.Grafical user interface for our expriment
First click Refresh button.It refress the program.Then Select the input from database.Database has two types of input tumor images and non tumor images.Then click Bilateral axis button it gives bilateral axis of selected input.Then click Possible tumor area button it gives possible tumor area exist region in MRI selected image ,if not exist it says tumor area does not exist.
Figure.6. GUI showing select input image,Bilateral Axis,Possible tumor area
Next click show result button it give segmentation of possible tumor area exist .
Figure.7.GUI shows possible tumor area does not found
Figure.8.It shows segmented possible tumor area
Lastly exist button ,if it click it ask Exist now ? it has two button Yes and No.If click Yes it exist if No then we can continue to observe another images to find out tomor area by above procedure
Figure.9.It shows GUI to exit or continue of program
So by clicking on push buttons we can check the results sequentially without any knowledge of MATLAB.Thus a user friendly envirornment has been designed which is helpful in understanding the work.
In this paper, we have presented an algorithm to detect possible tumor region from brain MRI. At first, MRI of health brain has an obviously character - almost bilateral symmetrical. However, if there is a macroscopic tumor, the symmetry characteristic will be weakened. According to the influence on the symmetry by the tumor, we develop a segment algorithm to detect the tumour region automatically.
But, our algorithm can only deduce the result from single
MRI, which means that we only make use of the 2D information of MR images. In the future, we will use the set of MR images (2D) to construct a 3D model. It can detect more small tumors, and the accuracy will also can be improved.