International Journal of Computer Science Engineering and Information Technology Research (IJCSEITR) ISSN 2249-6831 Vol. 3, Issue 1, Mar 2013, 61-66 © TJPRC Pvt. Ltd.

A COMPUTER AIDED DIAGNOSTIC SYSTEM FOR CLASSIFICATION OF BRAIN TUMORS USING TEXTURE FEATURES AND PROBABILISTIC NEURAL NETWORK SONALI PATIL1 & V. R. UDUPI2 1

Ph. D. Scholar, Shivaji University, Kolhapur, Associate Professor, Department of IT, KJSCE, Mumbai, India 2

Professor, Department of Electronics and Communication Engineering, GIT, Belgaum, India

ABSTRACT This research aims at detection of tumor blocks and classifying the type of tumor using Probabilistic Neural Network (PNN) in MR images of different patients with Astrocytoma type of Brain Tumor. The proposed technique consists of different stages, namely, preprocessing, segmentation, feature extraction and classification. The image processing techniques such as histogram equalization, thresholding, square based segmentation, component labeling and feature extraction have been developed for detection of brain tumor in the MRI images of cancer affected patients. The GLCM features are extracted from the detected tumor. These features are compared with stored features in knowledge base. Finally, a probabilistic Neural Network has been developed to classify the tumor. The developed system classifies the image into a Grade of tumor for Astrocytoma type of Brain Cancer. The system is found efficient in classification of these samples and responds on any abnormality noticed.

KEYWORDS: MR Images, Astrocytoma, Square Based Segmentation Gray Level Co-Occurrence Matrix, Probabilistic Neural Networks

INTRODUCTION Cancer is considered as life threatening disease. Cancer known medically as malignant neoplasm, is characterized by abnormal cell growth. The uncontrollable division of cells led to formation of lumps or masses of tissue called tumors. However, all tumors are not cancerous. Cancerous tumors in any organ can be primary or secondary. The tumor in any organ is said to be primary if it is the origin of the cancer in the body. The secondary tumors or metastases are result of spread of cancer into different organs of the body other than where it has originated. Brain cancer is the cancer that develops in the brain. Patients diagnosed with brain tumor/mass should not immediately conclude that they are suffering from brain cancer. The brain tumors can be benign (non-cancerous) or malignant (cancerous). The benign and malignant brain tumors can be further classified into many subtypes. Benign tumors also need to be treated. They can cause harm to neighboring tissues hence damaging the functioning of the corresponding vital organs. The malignant tumors are graded as Grade I to Grade IV. Grade I is characterized by less spread while Grade IV is characterized as maximum spread. The treatment for brain tumor is effective if the diagnosis is at the early stage. The Magnetic Resonance Imaging (MRI) is used as tool for diagnosis of brain abnormalities. The radiologist after seeing the MRI suggests biopsy of the suspected tissue. The report of biopsy will accurately diagnose the type and grade of tumor. The aim of Computer Aided Diagnostic (CAD) system designed for detection and classification of tumor is to separate the tumor from rest of the organ, extract image features from the separated tumor and classify the tumor using some classifier/inference system. This will eliminate the need for carrying out biopsy for the purpose of classification.

62

Sonali Patil & V. R. Udupi

RELATED WORK In [1], Pan Lin et al. developed an efficient automatic framework for Segmentation of MRI Brain Image. Abhishek Raj et al. [2] have demonstrated an improved framework for computer aided detection of brain tumor. They used de-noising in wavelet domain followed by enhancement using a non-linear enhancement function. Segmentation of the brain tumor was obtained by employing large sized structuring elements along with thresholding. Dipali M. Joshi et al [3] have designed a system that uses computer based procedures to detect tumor blocks or lesions and classify the type of tumor using Artificial Neural Network in MRI images of different patients with Astrocytoma type of brain tumors. They have used a Neuro Fuzzy Classifier. K. Selvanayaki et.al [4] have summarized and compared the methods of automatic detection of brain tumor through Magnetic Resonance Image (MRI) used in different stages of Computer Aided Detection System (CAD). Mohd Fauzi Othmani et al. [5] have proposed a scheme that uses feature extraction using the principal component analysis and the classification using Probabilistic Neural Network (PNN). They concluded that the Probabilistic Neural Network gives fast and accurate classification and is a promising tool for classification of the tumors. Rajesh Garg, et al [6], have done comparative analysis of different enhancement techniques. This comparison is done on the basis of subjective and objective parameters. AmirEhsan Lashkari [7] has introduced a novel automatic brain tumor detection method that uses T1, T2_weighted and PD, MR images to determine any abnormality in brain tissues. They used Gabor wavelets, energy, entropy, contrast and some other statistic features. The neural network was used to do the classification. Neelam Marshkole, et al [8] have tested features based on shape and texture of image for analysis and classification of brain tumors. After feature extraction, linear vector quantization is used to classify brain tumor in to malignant & benign types. Evangelia I.Zacharaki et al [9] used pattern classification methods for distinguishing different types of brain tumors, such as primary gliomas from metastases, and also for grading of gliomas. V.P.Gladis, et al. [10] have presented a novel method of feature selection and extraction. Their approach combined the Intensity, Texture, shape based features and classifies the tumor as white matter, Gray matter, CSF, abnormal and normal area. PCA and Linear Discriminant Analysis (LDA) were applied on the training sets. The Support Vector Machine (SVM) classifier served as a comparison of nonlinear techniques Vs linear ones. G Vijay Kumar , et al [11] have discussed an approach in computer-aided diagnosis for early prediction of brain cancer using Texture features and neuro classification logic. A neuro fuzzy approach is used for the recognition of the extracted region. The implementation is observed on various types of MRI images with different types of cancer regions.

PROPOSED WORK In this paper we proposed a Brain Cancer Detection and Classification System. The system uses computer based procedures to detect tumor blocks or lesions and classify the type of tumor using Artificial Neural Network in MRI images of different patients with Astrocytoma type of brain tumors. The important steps in the implementation of the system are as follows: 1.

Image Acquisition (Digital MR images)

2.

Image Preprocessing (Conversion of RGB image to Gray scale image, resizing, edge detection using Sobel, Binary Dilation, Histogram equalization and Thresholding)

3.

Image Segmentation (Square Based Segmentation and Component Labeling)

4.

Feature Extraction (Texture features using Gray Level Co-occurrence Matrix GLCM)

5.

Classification (Probabilistic Neural Network)

A Computer Aided Diagnostic System for Classification of Brain Tumors using Texture Features and Probabilistic Neural Network

63

The implementation of the project is carried out in MATLAB R2008b with Image Processing Toolbox and Neural Network Toolbox. In this proposed work, we have made used of Normal Brain MRI images and Astrocytoma type of brain tumor MRI images, which consists of four different grades of tumors as Grade I, Grade II, Grade III and Grade IV types. In square based segmentation method, each histogram equalized image is subdivided into a 40 x 40 square. Each 40 x 40 square, in the whole image is examined for the Region Of Interest (ROI). The ROI here in this work is, the tumor part. This segmentation should be stopped when tumor is able to be detected. In our work, we suggest a set of 11 GLCM based textural features which can be extracted from each of the gray tone spatial-dependence matrices. For classification we used PNN with 11 inputs, 6 hidden layers and 1 output layer. The 11 inputs to the neural networks are the 11 texture features that are extracted. The training in PNN is of supervised mode. The training is done for total of 39 images, out of which 5 are of normal images, 7 images of Grade I type, 5 images of Grade II type, 12 images of Grade III type and 12 images of Grade IV type. In the testing phase, a total of 15 images (3 x 5=15, 3 images from each set) are given as input to the PNN.

RESULTS AND DISCUSSIONS Fig. 1 Detection of tumor from the Brain MR Image. The result of individual step are shown as fig. a to fig. j where (a)The original image (b) Binary image filled with holes(mask) (c) Enhanced Image (d) Histogram Equalization (e) Image after border removal (f) Image after thresholding (g) Image after segmentation to locate ROI (h)Image with ROI found (i) Image with ROI darkened (j) Pure ROI of the image located. Fig. 2 shows the extracted features from all the MR Images used for training. Fig. 3 shows the result of classification in testing phase.

64

Sonali Patil & V. R. Udupi

The developed system efficiently classifies the input MRI image of Brain Cancer affected patients into a Grade of Astrocytoma type of Tumor. The MRI images of patients affected by Brain Cancer are used during Recognition/Testing phase. For the input image used for testing, the system shows the Tumor Region Extracted from the outer skull of brain. The features extracted from this region are compared with stored features in Knowledge base. The developed system then classifies the image into a Grade of the tumor for Astrocytoma type of Brain Cancer. The overall accuracy of the system is found to 94.8718%. Grade I, Grade II and Grade IV tumors have an accuracy of 100% , which means that all the input images are correctly being predicted by the developed system. And Grade III tumors have an accuracy of 91.66%. The Table 1 gives system accuracy rate.

A Computer Aided Diagnostic System for Classification of Brain Tumors using Texture Features and Probabilistic Neural Network

65

Table 2: Accuracy Rate for All the Tumor Grades and Normal Images Number of Input Images Grade I 7 Grade II 5 Grade III 12 Grade IV 10 Normal 5 Total Accuracy = 94.8718% Tumor Type

Number of Correctly Predicted Images 7 5 11 10 4

Accuracy Rate 100% 100% 91.66% 100% 80%

CONCLUSIONS •

The designed Brain Cancer Detection and Classification System use conceptually simple classification method using the probabilistic neural logic. The identification and identification of tumors is in automated fashion.

•

From the results of proposed work, it can be concluded that, The designed and implemented system provides precision detection and real time tracking by classifying the unknown sample Image into appropriate Astrocytoma type of Cancer, thus do not involve any pathological testing. This system provides precision Detection and Classification of Astrocytoma type of cancer.

•

The system has been tested with the Astrocytoma type of brain cancer Images only. The system can be designed to classify other types of cancers as well with few modifications. Also, large patient data can be used to improve accuracy. More features can be added and the most discriminating features can be selected for training to increase the accuracy and to make the system robust .

ACKNOWLEDGEMENTS Authors are grateful to Akshay Diagnostic Center, Sangli and Radiology Department of K.L.E.S Dr Prabhakar Hospital, Belgaum for providing MR images.

REFERENCES 1.

Pan Lin ,Yong Yang, Chong-Xun Zheng, Jian-Wen Gu,.An Efficient Automatic Framework for Segmentation of MRI Brain Image, Proceedings of the Fourth International Conference on Computer and Information Technology(CIT’04),IEEE,2004.

2.

Abhishek Raj, Alankrita, Akansha Srivastava, and Vikrant Bhateja, “Computer Aided

Detection of Brain Tumor

in Magnetic Resonance Images”, IACSIT International Journal of Engineering and Technology, Vol. 3, No. 5, October 2011. 3.

Dipali M. Joshi, Dr.N. K. Rana, V. M. Misra, “Classification of Brain Cancer Using Artificial Neural Network”, 2nd International Conference on Electronic Computer Technology (ICECT 2010),IEEE 2010.

4.

K.Selvanayaki, Dr. M. Karnan, CAD System for Automatic Detection of Brain Tumor through Magnetic Resonance Image-A Review, International Journal of Engineering Science and Technology, 2(10), 2010, 58905901.

5.

Mohd Fauzi Othman and Mohd Ariffanan Mohd Basri, “Probabilistic Neural Network For Brain Tumor Classification”, Second International Conference on Intelligent Systems, Modelling and Simulation, IEEE 2011.

6.

Rajesh Garg, Bhawna Mittal, Sheetal Garg, “Histogram Equalization Techniques For Image Enhancement”, International Journal of Electronics & Communication Technology, Vol. 2, Issue 1, March 2011.

66

Sonali Patil & V. R. Udupi

7.

AmirEhsan Lashkari, “A Neural Network based Method for Brain Abnormality Detection in MR Images Using Gabor Wavelets”, International Journal of Computer Applications (0975 – 8887),Volume 4 – No.7, July 2010

8.

Neelam Marshkole, Bikesh Kumar Singh, A.S Thoke, “Texture and Shape based Classification of Brain Tumors using Linear Vector Quantization”, International Journal of Computer Applications (0975 – 8887) ,Volume 30– No.11, September 2011.

9.

Evangelia I. Zacharaki, Sumei Wang, Sanjeev Chawla, Dong Soo Yoo, Ronald Wolf, Elias

R. Melhem, and

Christos Davatzikos, “Classification of Brain Tumor Type and Grade Using MRI Texture and Shape in a Machine Learning Scheme”, Magnetic Resonance in Medicine 62:1609–1618 (2009), 2009 Wiley-Liss. 10. V.P.Gladis Pushpa Rathi and Dr.S.Palani, “A Novel Approach For Feature Extraction And

Selection On

MRI Images For Brain Tumor Classification”, David C. Wyld, et al. (Eds): CCSEA, SEA, CLOUD, DKMP, CS & IT 05, pp. 225–234, 2012. 11. G Vijay Kumar and Dr GV Raju, “Biological Early Brain Cancer Detection Using Artificial Neural Network”, International Journal on Computer Science and Engineering Vol. 02, No. 08, 2721-2725, 2010. 12. Rafael C. Gonzalez and Richard E. Woods. ‘Digital Image Processing using MATLAB’, 2nd edition. Prentice Hall, 2002. ISBN 0-201-18075-8. 13. Jaceck Zurada, “Introduction to Artficial neural systems,” Jaico Publishing pages 790. 14. S. N. Sivanandam, S. Sumathi, S. N. Deepa, “Introduction to Neural Networks using Matlab 6.0”. Tata McGraw Hill Company Ltd, New Delhi, June 2005.