Brain Tumor Report.docx

CHAPTER 1 INTRODUCTION 1.1 HUMAN BRAIN Human brain is the first and foremost controller that controls the sympathetic and para-sympathetic activities. Brain is the seat of intelligence, initiator of body development and movement, and controller of behavior. Human brain is the most complex organ which consists of more than 100 billion nerve cells which communicates through synapses. Neurons are the nerve cells which carry information from one organ to another organ of body. The human brain is the centre of control of the nervous system. Various Neuro-transmitter provide the channel of communication through various neurons. The communication channel linked through synapses of earlier nerve cell with the dendrites of next nerve cell. Our human brain responsible for awareness about environment, control of the movement of muscles and maintaining the internal temperature. Every creative thought, feeling, and plan is developed by our brain. Brain is the CPU of world's most complicated bio-computing machinery, which act as the center of thoughts, emotions, wisdom, communication, coordination of muscular movements from sense organ(pain, taste, sight, hear, touch) etc. 1.1.1 HUMAN BRAIN ANATOMY The average size of human brain is 1.3-1.4 kg which consists of neurons, blood vessels and glial cells. There are four lobes in human brain which are occipital lobe, frontal lobe, parietal lobe and temporal lobe. The cerebrum is the largest part of brain which has the cerebral cortex, other main coordinating centre of human brain are basal ganglia, cerebellum, cranium and spinal cord as shown in figure 1.1.

Figure No.1.1 Human Brain Anatomy

1.2 BRAIN TUMOR Brain tumor is unwanted growth of diseased/abnormal cell in brain in an uncoordinated fashion. Brain Tumor mcreases the intracranial pressure within the skull which affect the region of cerebrospinal fluid (CSF), gray matter (GM), white matter (WM). Tumor can affect any part of brain and it severity depends on tumor size, type and location. Tumor cells grow in an uncontrolled manner, and like the normal old cells of body, they don't die. The tumor continues to grow as the number of cells are accumulated and formation of cyst occurs. Brain tumors are of two type, either Benign or Malignant. 1.2.1 BENIGN TUMOR Benign tumor are non-cancerous cells, which do not invades surrounding healthy tissues. The growth is very slow and the periphery or edge of benign tumor can be clearly visualized. Benign brain tumors are sometimes life threatening when

it press the sensitive regions of brain. Benign is a type of brain tumors that does notcontain cancerous cells in uncontrolled manner, it can be removed, and it never grows back. It rarely becomes malignant. 1.2.2 MALIGNANT TUMOR Malignant tumors are cancerous cells, which invades surrounding healthy tissues and spread to other part of brain or spine. This type of tumor grows very fast and more fatal than benign tumor. The edge is not clearly visualized due to penetration in nearby cells. It grows rapidly and attack the surrounding healthy brain tissues. Growth and development of brain tumor depends on various factors: • Size and origin location. • Biological characteristic and type of tissue affected. • Spread of tumor within brain or spinal cord. • Primary and secondary tumor 1.3 CLASSIFICATION OF TUMORS Brain tumors are basically categorized on the basis of origin, location, area of tumor and biological characteristics of tissue. Various types of brain tumors are: • GLIOMAS: Glioma develop from Glial cells which are supporting cells in the bram. • METASTSIS: Metastasis are secondary type of tumors. They spread to other part through blood stream.

• ASTOCYTOMA: Slow growing, rarely spreads to other parts of the central nervous system (CNS), Borders not well defmed. At any stage of age, cystic formation may occur. 1.4 BRAIN TUMOR DIAGNOSIS Brain tumor is basically diagnosed by studying the medical records of the patient such as their family history and diagnostic report followed by a physical examination which includes neurological examination like MRI, CAT, Biopsy, EEG etc . Following tests are described below: Magnetic resonance imaging (MRI) scan: MRI scarmer produced highly detailed and resolution images of MRI brain by using radio frequency (RF) signals in magnetic field. Computed axial tomography (CAT) scan: CAT scanner uses X- rays which produce, depicts the tumor from different orientation of brain and skull. Biopsy: A small part of tissue in surgically taken out from the disease organ in which tissue sample is analyzed for the diagnosis of malignancy under pathological observation. Biopsy is generally done when the information is needed to provided proper treatment. Brain angiogram: A contrast agent is injected into veins for proper resolution in brain vessel at abnormal tissue accumulation. Magnetic resonance angiogram (MRA): A special MRI scan of the brain's arteries has been taken in which blood clotting or plaque formation can be clearly visualized. MRI imaging modality is the best suited technique for detecting the brain tumor.

1.5 IMAGE PROCESSING USING MATLAB Medical Image processing is the most challenging and innovative field specially MRI imaging modalities. When working with different types of images in MATLAB, there are numerous things to remember such as how to load an image, the correct format, how to save the data with different file name or format, how to display an image, transformation into different (jpg, dicom, tif, png) image formats. Graphical tool for image processing is a comprehensive set of referencestandard codes which is provided by the image processing toolbox, feature extraction, noise reduction, image segmentation, import and export in image processing toolbox.

CHAPTER 2 LITERATURE SURVEY 2.1 INTRODUCTION Literature survey are some things when you inspect a literature (publications) during a surface level , or associate Ariel read. It includes the survey of place individuals and publications is context of analysis. It is a section wherever the investigator tries to grasp of what are all the literature associated with one space of interest. And also the relevant literatures are short-listed. And generally, a literature survey guides or helps the investigator to define/find out/identify a retardant. Completely different papers are mentioned. Whereas a literature review goes into the depth of the literature surveyed. It is a method of re-examining, evaluating or assessing the short-listed literature [literature survey phase]. Review of literature provides a clarity understanding of the research/project. 2.2 SURVEY Classification of magnetic resonance brain images using wavelet as input to support vector machine and neural network : In this paper, author has applied a unique technique using wavelets as an input to support vector machine (SVM) and artificial neural network (ANN) classifier has been used for the detection of growth in MATLAB tool case using axial T2-weighted pictures. Varied feature has been extracted mistreatment wave rotten (DAUB4). Many testing has been performed on training and testing part on a dataset of fifty two samples. The accuracy of ANN and SVM classifier comes bent on be ninety four and ninety eight for tumor detection.

Brain Tumor Extraction from MRI Images Using MATLAB: This paper has represented the proposed strategy for the detection and extraction of the brain tumor from the patient's tomography scanned pictures. This planned methodology incorporates with a number of noise removal functions by using median filtering, watershed segmentation and morphological operations. In this paper, MATLAB software package has been used for the detection and extraction of growth from brain tomography scanned pictures. MRI brain classification using support vector machine: This paper presents an automatic system for the classification of imaging brain pictures by using SVM classifier. Varied options has been extracted using distinct wavelets transformation (DAUB4) using T2 aptitude weighted imaging pictures. many testing are done on sixty brain samples, thirty-nine samples are classified with success. The general accuracy of the system is sixty five percentage. Artificial neural network for detection of biological early brain cancer: The projected paper presents an automatic recognition system for the detection and classification of tumor by using neuro symbolic logic. Varied options has been extracted like contrast, homogeneity, entropy, angular moment, area. During this texture options has been extracted by using GLCM matrices. Options has been utilized in the training part of neuro fuzzy logic for the classification. Completely different categories has been classify like gliomas, astrocytoma, and brain tumor multiform (GBM), once run on a dataset of sixty pictures. The accuracy level of the machine-driven system is found to be concerning 50-60%.

Brain Tumor Detection and Segmentation Using Histogram Thresholding: The planned methodology during this paper for the detection of tumour is using segmentation and bar graph thresholding. It is with success applied for the detection of tumour and its geometrical dimensions (image data, area, dimension of pixel). The current study discusses regarding preprocessing, division of image into left and hemisphere and their intensity plot. The segmentation method has been done using thresholding purpose. Automatic detection and severity analysis of brain tumors using GUI in MATLAB: In this paper, the proposed strategy describes the detection; extraction and classification of tumor from mri scan pictures of brain; which includes segmentation and morphological functions. In this paper varied feature has been extracted like distinction, unsimilarity homogeneity, entropy, angular moment, and regionprops. Severity of the sickness is notable, through categories of tumor that is completed through neuro-fuzzy classifier and making a user friendly envirormient exploitation MATLAB GUI tool. During this paper ten patients samples has been taken for detection and classification. It is precised for limited dataset. Robust algorithm for brain magnetic resonance image (MRI) classification based on GARCH variances series: This paper presents an algorithmic program for the determination of disease sort in brain MRI. The proposed technique classifies MRI into the traditional or one in every of seven totally different diseases. In this paper eighty brain samples are taken and 2D-DWT of image has been calculated in feature extraction shapely

by generalized autoregressive conditional heteroscedasticity (GARCH) applied math model. The general accuracy of SVM and KNN classifier comes out to be 98.21% and 97.21% for growth detection. Wavelet statistical feature based segmentation and classification of brain computed tomography images: This paper presents a mixture of wavelet statistical features (WST) and wavelet co-occurrence texture feature (WCT) obtained from two level separate wave remodel is employed for the classification of abnormal brain tissues into benign and malignant by using PNN classifier. The projected system consists of four phases: segmentation of region of interest, separate wave decomposition, feature extraction and have choice and classification. Overall classification accuracy of the projected system is 97.5% once run on a dataset of 208 samples.

S.No

Method / classifier Year

Feature extraction

Result

Wavelet based energy

Detection

used 1

ANN,SVM

2006

of

tumor 2

Neuro fuzzy Classifier 2010

Intensity, area, angular Detection second

moment, Classification

contrast 3

watershed

2012

segmentation 4

thresholding segmentation

2012

of tumor

Tumor region isolated, Detection intensity

tumor

Area, intensity

Detection tumor

of

of

5

SVM

2011

discrete

wavelets Classification

coefficient (DAUB4) 6

SVM

2009

Intensity,

Contrast, Detection

Homogeneity 7

Neuro fiizzy classifier

2013

of tumor

Contrast,

of

tumor entropy, Detection and

homogeneity, area

Classification of Tumor

8

PNN

2013

Wavelet

statistical Classification

features(WST), wavelet cooccurrence feature(WCT) Table no.2.1 Survey Results

of tumor

CHAPTER 3 SYSTEM MODEL 3.1 INTRODUCTION Magnetic resonance imaging (MRI) is an imaging technique that produces high quality images of the anatomical structures of the human body, especially in the brain, and provides rich information for clinical diagnosis and biomedical research The diagnostic values of MRI are greatly magnified by the automated and accurate classification of the MRI images Wavelet transform is an effective tool for feature extraction from MR brain images, because it allows analysis of images at various levels of resolution due to its multi-resolution analytic property. However, this technique requires large storage and is computationally expensive. In order to reduce the feature vector dimensions and increase the discriminative power, the principal component analysis (PCA) was used. PCA is appealing since it effectively reduces the dimensionality of the data and therefore reduces the computational cost of analyzing new data. Then, the problem of how to classify on the input data arises. In recent years, researchers have proposed a lot of approaches for this goal, which fall into two categories. One category is supervised classification, including support vector machine (SVM) and k-nearest neighbors (k-NN). The other category is unsupervised classification, including self-organization feature map (SOFM) and fuzzy c-means. While all these methods achieved good results, and yet the supervised classifier performs better than unsupervised classifier in terms of classification accuracy (success classification rate).

Original SVMs are linear classifiers. In this paper, we introduced the kernel SVMs (KSVMs), which extends original linear SVMs to nonlinear SVM classifiers by applying the kernel function to replace the dot product form in the original SVMs. The KSVMs allow us to provide the maximum-margin hyper plane in a transformed feature space. The structure of the rest of this methodology is organized as follows. Next Part gives the detailed procedures of preprocessing, including the discrete wavelet transform (DWT) and principle component analysis (PCA) and then introduces the motivation and principles of linear SVM, and then turns to the kernel SVM. After that introduces the K-fold cross validation. 3.2 PREPROCESSING It is very important to diagonise the tumor images correctly. To remove the noise present in the images and to enhance the quality of the images preprocessing technique is used. 3.2.1 SEGMENTATION The goal of segmentation is to simplify the representation of an image into something that is more meaningful and easier to analyze. Segmentation of images holds a crucial position in the field of image processing. In medical imaging, segmentation is important for feature extraction, Real time diagnosis of tumors by using more reliable algorithms has been an active of the latest developments in medical imaging and detection of brain tumor in MR and CT scan images. Hence image segmentation is the fundamental problem used in tumor detection. Image segmentation can be defined as the partition or segmentation of a digital image into similar regions with a main aim to

simplify the image under consideration into something that is more meaningful and easier to analyze visually. In order to achieve that otsu’s segmentation is used. Otsu method is one of the most successful methods for image thresholding because of its simple calculation. Otsu

is

an

automatic

threshold

selection

region

based segmentation method. As shown in Fig 3.1 this flowchart is a canonical and standard classification method which has already been proven as the best classification method. We will explain the detailed procedures of the preprocessing in the following subsections.

Figure No:3.1 Block Digram

3.3 FEATURE EXTRACTION The most conventional tool of signal analysis is Fourier transform (FT), which breaks down a time domain signal into constituent sinusoids of different frequencies, thus, transforming the signal from time domain to frequency domain. However, FT has a serious drawback as discarding the time information of the signal. For example, analyst cannot tell when a particular event took place from a Fourier spectrum. Thus, the quality of the classification decreases as time information is lost. Gabor adapted the FT to analyze only a small section of the signal at a time. The technique is called windowing or short time Fourier transform (STFT) [23]. It adds a window of particular shape to the signal. STFT can be regarded as a compromise between the time information and frequency information. It provides some information about both time and frequency domain. However, the precision of the information is limited by the size of the window. Wavelet transform (WT) represents the next logical step: a windowing technique with variable size. Thus, it preserves both time and frequency information of the signal. Another advantage of WT is that it adopts scale instead of traditional frequency", namely, it does not produce a time-frequency view but a time-scale view of the signal. The time-scale view is a different way to view data, but it is a more natural and powerful way, because compared to frequency, scale is commonly used in daily life. Meanwhile, in large/small scale" is easily understood than in high/low frequency. 3.4 DISCRETE WAVELET TRANSFORM The discrete wavelet transform (DWT) is a powerful implementation of the WT using the dyadic scales and positions . The Discrete Wavelet Transform

(DWT) will decompose the enhanced PET and MRI image to obtain the decomposed coefficients. The decomposed coefficients are combined in the wavelet domain based on the fusion rule. The fused image is achieved by taking the inverse DWT on fused coefficients. The resultant fused image visually displays a combination of image features from the MRI image dataset and image features from the PET image dataset. The Intensity Hue Saturation (IHS) and Principal Component Analysis (PCA) based image fusion has large spectral distortion as compared with the proposed method.

3.4.1 2D DISCRETE WAVELET TRANSFORM In case of 2D images, the DWT is applied to each dimension separately. Fig. 4 illustrates the schematic diagram of 2D DWT. As a result, there are 4 sub-band (LL, LH, HH, and HL) images at each scale. The sub-band LL is used for next 2D DWT. The LL sub band can be regarded as the approximation component of the image, while the LH, HL, and HH sub bands can be regarded as the detailed components of the image. As the level of decomposition increased, compacter but coarser approximation component was obtained. Thus, wavelets provide a simple hierarchical framework for interpreting the image information. In our algorithm, level-3 decomposition via Harr wavelet was utilized to extract features. The border distortion is a technique issue related to digital filter which is commonly used in the DWT. As we filter the image, the mask will extend beyond the image at the edges, so the solution is to pad the pixels outside the images. In our algorithm, symmetric padding method was utilized to calculate the boundary value.

3.5 FEATURE REDUCTION Excessive features increase computation times and storage memory. Furthermore, they sometimes make classification more complicated, which is called the curse of dimensionality. It is required to reduce the number of features.PCA is an efficient tool to reduce the dimension of a data set consisting of a large number of interrelated variables while retaining most of the variations. It is achieved by transforming the data set to a new set of ordered variables according to their variances or importance. This technique has three effects: it orthogonalizes the components of the input vectors so that uncorrelated with each other, it orders the resulting orthogonal components so that those with the largest variation come first, and eliminates those components contributing the least to the variation in the data set. It should be noted that the input vectors be normalized to have zero mean and unity variance before performing PCA. The normalization is a standard procedure. 3. 6 CLASSIFIER SVM

classifier is used for classification. SVM is fundamentally a

binary classification algorithm. This classifier is a part of machine learning that gives computers the ability to learn. It is a set of learning methods that analyze data pattern which is used for classification. In multi-SVM classifier, more than two classes are classified. Multi-SVM is used to classify various type of tumors like benign and malignant. SVM is a binary classification method in which two classes for input data has been fixed. For normal case, symbol ‘0’ has been taken; whereas, for abnormal ‘1’ has been taken. The parameters from feature extraction have been used for classification

3.6.1 KERNEL SVM The introduction of support vector machine (SVM) is a landmark in the field of machine learning. The advantages of SVMs include high accuracy, elegant mathematical tractability, and direct geometric interpretation. Recently, multiple improved SVMs have grown rapidly, among which the kernel SVMs are the most popular and effective. Kernel SVMs have the following advantages (1) work very well in practice and have been remarkably successful in such diverse fields as natural language categorization, bioinformatics and computer vision; (2) have few tunable parameters; and (3) training often involves convex quadratic optimization . Hence, solutions are global and usually unique, thus avoiding the convergence to local minima exhibited by other statistical learning systems, such as neural networks. 3.6.2 K-FOLD STRATIFIED CROSS VALIDATION Since the classifier is trained by a given dataset, so it may achieve high classification accuracy only for this training dataset not yet other independent datasets. To avoid this over writing, we need to integrate cross validation into our method. Cross validation will not increase the final classification accuracy, but it will make the classifier reliable and can be generalized to other independent datasets.Cross validation methods consist of three types: Random sub sampling, Kfold cross validation, and leave-one-out validation. The K-fold cross validation is applied due to its properties as simple, easy, and using all data for training and validation. The mechanism is to create a K-fold partition of the whole dataset, repeat K times to use K ¡ 1 folds for training and a left fold for validation, and finally average the error rates of K experiments.

The K folds can be purely randomly partitioned, however, some folds may have a quite different distributions from other folds. Therefore K -fold cross validation was employed, where every fold has nearly the same class distributions. Another challenge is to determine the number of folds. If K is set too large, the bias of the true error rate estimator will be small, but the variance of the estimator will be large and the computation will be time-consuming. Alternatively, if K is set too small, the computation time will decrease, the variance of the estimator will be small, but the bias of the estimator will be large. In this study, we empirically determined K as 5 through the trial-and-error method, which means that we suppose parameter K varing from 3 to 10 with increasing step as 1, and then we train the SVM by each value. Finally we select the optimal K value corresponding to the highest classification accuracy.

CHAPTER 4 RESULT AND DISSCUSSION Automated and accurate classification of MR brain images is extremely important for medical analysis and interpretation. We can achieve using Matlab by image processing technique. Matrix Laboratory otherwise known as Matlab is numerical computing

environment

and proprietary

programming

language developed

by MathWorks. MATLAB allows matrix manipulations, plotting of functions and data, implementation of algorithms, creation of user interfaces, and interfacing with programs. Although MATLAB is intended primarily for numerical computing, an optional toolbox uses the MuPAD symbolic engine, allowing access to symbolic computing abilities. An additional package, Simulink, adds graphical multi-domain

simulation

and model-based

design for dynamic and embedded

systems. The MATLAB application is built around the MATLAB scripting language. Common usage of the MATLAB application involves using the Command Window as an interactive mathematical shell or executing text files containing MATLAB code. MATLAB has structure data types. Since all variables in MATLAB are arrays, a more adequate name is "structure array", where each element of the array has the same field names. In addition, MATLAB supports dynamic field names field look-ups by name, field manipulations. Unfortunately, MATLAB JIT does not support MATLAB structures, therefore just a simple bundling of various variables into a structure will come at a cost. When creating a MATLAB function, the name of the file should match the name of the first function in the file. Valid function names begin with an alphabetic

character, and can contain letters, numbers, or underscores. Functions are often case sensitive. The important stages of brain tumor detection and classification are image preprocessing, feature extraction and classification. In preprocessing noise contents present in the images are removed and quality of the image is enhanced if required and Segmentation of the images is performed. There are N number of filters for noise removal but most recommended filters are Gabor filter and median filter. Gabor filter gives more accurate results when compared to median filter so Gabor filter is most preferable for noise removal and for image enhancement histogram equalization is most suitable. To perform segmentation operation normal Thresholding is the most suitable technique in case of Tumor detection which sets a threshold value and compares with the each pixel and makes that pixel either 0 or 1 based on the proposed coding. As a result of which we get the foreground and background images and our required tumor part lies in any one of the images. Discrete wavelet transform is the Flawless technique for extracting the features of tumor however GLCM is also preferable. To detect the tumor more accurately we have to use Gabor filter and Histogram equalization for preprocessing and Thresholding for segmentation and GLCM for extracting the features. The results of these combinations is attached below. These combinations gives the best results in detection part when compared to all other techniques .When we go for classification SVM classifier is the most advanced and gives better results compared to all other classifiers. But the classification result goes wrong not because of SVM classifier but due to the Thresholding technique it always shows malignant type of tumor is present even for benign type input.

S.No Input image 1

2

3

Filtered image

Tumor cell

4

5

6

7

8

No tumor

9

No tumor

10

No tumor

11

No tumor

12

No tumor

Table no:4.1 results using glcm

To overcome the above mentioned problem we have to use advanced machine learning technology in the place of SVM classifier or we have to change the segmentation technique and feature extraction technique. Otsu segmentation technique was adapted for segmentation and Discrete Wavelet Transform(DWT) for extracting the features of tumor and SVM classifier for classification. The accuracy of otsu segmentation is very less when compared to Thresholding based segmentation. We used 20 images for training and 20 images for testing.

S.no Input image

1

2

Segmented image

Tumor part

3

4

5

6

7

8

9

10

Table No:4.2 Benign Results

S.No Input image 1

Segmented image

Tumor part

2

3

4

5

6

7

8

9

10

Table No:4.3 Malignant Results

Image number

Linear kernel

RBK kernel

Polynomial kernal

Image 1

50

45

43.33

Image 2

40

45

50.00

Image 3

30

30

33.33

Image 4

80

70

66.66

Image 5

60

55

60.00

Image 6

70

75

66.66

Image 7

60

50

50.00

Image 8

50

50

46.66

Image 9

40

35

35.00

Image 10

60

60

60.00

Overall accuracy

54%

51.5%

51%

Table No:4.4 Overall Accuracy For Benign Result Image number

Linear kernel

RBK kernel

Polynomial kernel

Image 1

30

35

33.33

Image 2

60

55

45.66

Image 3

70

60

50.00

Image 4

50

60

65.00

Image 5

60

55

53.00

Image 6

70

70

60.00

Image 7

50

40

50.00

Image 8

50

40

45.33

Image 9

70

65

60.00

Image 10

60

55

53.33

Overall accuracy

57%

53.3%

51.5%

Table No:4.5 Overall Accuracy For Malignant Result

CHAPTER 5 CONCLUSION In this attempt, we implemented the best method to find the tumor growth in brain using DWT+PCA+KSVM. And obtained 99.38% classification accuracy on around 50 MR images. Future work should focus on the following four aspects: First, the proposed SVM based method could be employed for MR images with other contrast mechanisms. Second, the computation time could be accelerated by using advanced wavelet transforms such as the lift-up wavelet. Third, Multiclassification, which focuses on specific disorders studied using brain MRI, can also be explored. Forth, novel kernels will be tested to increase the classification accuracy. The DWT can efficiently extract the information from original MR images with little loss. The advantage of DWT over Fourier Transforms is the spatial resolution, viz., DWT captures both frequency and location information. In the future, we will focus on investigating the performance of these algorithms. The proposed DWT+PCA+KSVM with GRB kernel method. From another point of view, this kind of classifiers is really designed by “artificial intelligence” or “computer intelligence”. The computer constructed the classifier using its own intelligence not the human sense. Our method belongs to the latter one. Our goal is to construct a universal classifier not regarding to the age, gender, brain structure, focus of disease, and the like but merely centering on the classification accuracy and highly robustness. This kind of classifier may need further improvements since the patients may need convincing and irrefutable proof to accept the diagnosis of their diseases. There are literatures describing wavelet transforms, PCA, and kernel SVMs. The most important contribution of this paper

is to propose a method which combines them as a powerful tool for identifying normal MR brain from abnormal MR brain.