BRAIN TUMOR DETECTION USING IMAGE SEGMENTATION

  • : Ms Word, Ms Word Format
  • : 76 Pages
  • : ₦5,000
  • : 1-5 Chapters
  •  
  • Click to DOWNLOAD Materials

BRAIN TUMOR DETECTION USING IMAGE SEGMENTATION

Abstract

Image thresholding is one of the techniques that are used for image segmentation. Threshold techniques divide the image into two main regions, these are: Foreground and Background. The output of the thresholding process is a binary image with only two regions that are formed by the highest possible contrast that could be found in the image. Entropies are information gain approaches that have been used for image thresholding with various application and image modalities. However, the accuracy of the existing entropies for image thresholding has been studied in general domain (e.g.: natural images) that teams from the regular medical images and images that form in the ordinary image is a reflectionoflightobjects,Whilemedicalimages.Takenbymagneticresonanceimaging, for example,A strong magnetic field is used with radio frequencies andcomputer to produce automatic selection of the best result. It produces the results withthe highest accuracy. detailed images of organs and soft tissues, bones and other internalparts of the body. and were not compared thoroughly. In this work, the accuracy of theimage segmentation approaches and their combination in brain tumor detectionframeworkisinvestigated.Forthispurpose,aframeworkforbraintumorsegmentation is developed. The developed framework is made simple and has the core process of the image thresholding, in order to evaluate the accuracy of the entropies. Five entropies,namely,Reniyh,Maximum,Minimum,TsallisandKapurareevaluated.Theaggregationofentropieswasimplementedandevaluated.The results show that the maximum trophy is the best for brain tumor detection.

 CHAPTER ONE

 INTRODUCTION

  • Background of Study

Captured images, over decades, have helped in solving many of the problems that were difficult to resolve using the traditional ways in many fields, such as: earth science, astronomy, biology, industry, etc. Images have also contributed to the development of the most important field, the medical, which helps in the survival of the human being. With the ever increase in the value of images; there is a demand for automatic analysis, processing and recognition of these images. The processing demand is emerged by the fact that it might be difficult to re-capture the images as the phenomena cannot be brought back to an earlier time or it is too expensive to capture the same image again and again. The solution to such atomization is the digital image processing.

Digital image processing is a branch of computer science that concerns about the automatic handling of the images in term of saving, improving, analysis and information extraction. Image segmentation is an important phase in digital image processing. Image segmentation divides the image into coherent and homogeneous regions according to specific criteria, such as: region color, region shape, or region boundary. The union of the segmented regions should result in reconfiguring the original image. Image segmentation allows the extraction of valuable information from the image as it provides a high-level description of each region individually, and allows for the linkage of neighboring regions in the image. An example of a segmented image is illustrated in Figure1.1 (Gonzales& Woods, 2002).

 

cerebrum
cerebellum
Corpus collosum
Brain stem
hypothalamus

Figure 1.1: Segmentation of brain image (Myron et al., 1970)

 Tumor Detection

 Tumor is the abnormal growth of cells to form abnormal fraction that has different characteristics from the normal cells. Tumor is classified into a benign tumor, pre- malignant tumor and malignant. Benign tumor is the one that does not grow suddenly and has no effect on tissue, example of this class of tumor is moles. Pre-malignant is the class that if it is not treated quickly, it becomes a malignant tumor.Malignant tumor grows rapidly and affects the neighboring tissue and, with time, it affects human life and leads to the death. Tumor detection is an important part in the treatment process. Thus, tumor detection techniques have concerned researchers in computer fields, especially, image processing (Wu, & Chang, 2007).

  • (b)

Figure 1.2: Brain tumor detection (a) input image and (b) detected tumorregion (Wu& Chang, 2007)

Automatic tumor detection in early stage is critical task that were addressed by many existing approaches. One of the most important stage in tumor detection is image segmentation, in which tumor is being isolated from other healthy tissues. By isolating the tumor then determines its stage, the treatment becomes easier (Marcel ,2004).

 

Image segmentation

Image thresholding is one of the techniques that are used for image segmentation. Threshold techniques divide the image into two main regions, these are: Foreground and Background. The output of the thresholding process is a binary image with only two regions that formed by the highest possible contrast that could be found in the image (Abu-Shareha et. al., 2008). This type of thresholding, which produce two regions, is called global thresholding. The other type of thresholding is called multi-thresholding. Multi-thresholding, in general, is implemented by segmenting an image into multiple objects and background, as illustrated in Figure 1.2.

What’s happening, during the application of the thresholding? First, the value of the threshold is determined. Then, all pixels with values that are greater than the threshold considered in one object and all the pixels with values that are less than the threshold value is considered as a background and vice versa (Prasanna&Arora, 2006).

Figure 1.3: Example segmented image (a) an image of three objects and (b) the result of image segmentation.

The main assumption of global thresholding is that the object and background can be distinguished by searching the gray-level value that divides the image into two distinguished parts. Threshold mathematically easy and required less time compared to the other approaches of image segmentations (El-Sayed et al., 2014).

In order to determine the value of the threshold, several approaches have been developed and used. Entropy is one of these approaches that aim sat finding a threshold value that facilitates maximum information extraction from the image. Entropy has been emerged in Information Theory to extract the amount of information expressed by a piece of data (El-Sayed, 2014).

Entropy is a Greece word, which means “if any system has many point of information’s, the entropy is incense until arrive to equal distribution for this information”. This technique helps to get a good threshold for the regions in the image. Entropies not only used in computer sciences; it is used in many different fields, such as: physics, biology, astronomy, etc. Entropy in image processing measures the amount of information that can be obtained from the image, either in its original form or after some processing. There are several ways to use entropy, as well as several equations to be used as the entropy basis(Abu-Shareha et al., 2008).

The entropies that are used for thresholding, are many, each of them has different aim, such as: reducing error, increase efficiency and remove noise. Some kinds of entropy are: Renyih, maximum, Tsallis and minimum cross.

In this work, the accuracy of image thresholding, as the most important factor in tumor detection, is evaluated.

 

1.2             Problem Statement

The accuracy of the existing entropies for image thresholding has been studied in general domain (e.g.: natural images). However, natural image are different from medical images by all means (e.g.: the contrast, colors, etc.). Moreover, medical images differ from each other by the means of organ, modality and equitation parameters such as chosen thresholding value, priorities value and possibilities value. Subsequently, there is a need to evaluate the existing entropies for medical image segmentation.

This problem can be further divided into the following sub-problems:

 

  1. How to develop a tumor detection framework that depends on image

 

  1. How to use entropy based thresholding in the developed tumor detection

 

  1. How to combine more than single entropy to produce a single segmented image by merging and selection.

 

  1. How to compare between different image segmentation in the developed tumor detection framework and different combinations.

 

1.3   Goal and Objectives

 The goal of this work is to evaluate the accuracy of the image segmentation approaches and their combination in brain tumor detection framework. The objectives of this research are as follows:

  1. To develop a tumor detection framework that takes a brain image and produces a segmented image with a detected tumor if the tumor is present.

 

  1. To use different entropy based thresholding in the developed tumor detection

 

  1. To combine multiple thresholding approaches by applying logical operators (AND and OR) on the thresholding output and acquires an automatic selection of their outputs to get the best

 

  1. To evaluate and compare the entropies results and their different combinations in the developed tumor detection framework.

 

1.4             Significance of study

A human life is the most important thing in the globe; medical researcher tries to make human life comfortable by defeating and curing diseases that may decimate health. This work is motivated by both the crucial need for technology-based applications in the field of tumor detection, and also the significant amount of time and effort to be saved by involving machine learning techniques in this field. More specifically, this work is devoted to brain tumors that are not easy to be understood as it comes in images with different shapes and intensities. Currently, as the detection process is still immature, it is not really used for treatment and diagnosis, it is used for indexing and retrieval of images in teaching of medicine by example.

1.5             Research Methodology

The proposed work is implemented in various phases as given in Figure 1.4, these are:

Figure 1.4: Research Methodology

Building a Segmentation Framework

First, a segmentation framework, in which the entropies will be employed, is constructed. Simply, this framework reads the input image, applies the thresholding and report the results.

The proposed framework deals with medical image; subject matter is gray-level images. The difference between the gray-level images and color images is that each pixel in gray- level images is represented by a single value, usually 0-255, while each pixel in color images represented by more than one value (e.g.: 3 values for RGB images) (Mohamed and Clausi, 2001).

Building a Classification Mechanism

  The images, before they undergo to image segmentation for the purpose of tumor detection, they undergo classification process, which classifies the images based on the presence and absence of tumor. The classification is implemented based on the images as a whole.

Building an Analysis Mechanism

The outputs of different entropies are collected and analyzed and combined using different logical operators.

Evaluation

  The evaluation of the proposed framework is carried on based on a set of syntactic data.

 

1.6     Scope

 The research conducted in this thesis evaluates the accuracy of the image segmentation in tumor detection framework, the following summarizes the scope of the conducted research:

  • Images used in this research are synthetic images provided by a well-known trusted Obtaining Images of real tumor patients is not easy as this would involve privacy and data protection issues. However, what is applied on synthetic images can be applied on real images as they are identical by all the means.

 

  • The processing framework deals with individuals 2D 3D volume processing is outside the scope of this research.

 

  • This thesis focus on the original and mostly-utilized entropies. Other entropies that were developed by extended original one is outside the scope of this

 

1.7     Study Outlines

In this chapter, Chapter One, a brief introduction to the problem that will be investigates in this thesis is given. Moreover, the problem statement, goal and objectives and the proposed framework is given. Chapter Two, discusses the related work in the field of image segmentation and tumor detection. Chapter Three, presents and discusses the proposed work for evaluating the existing image segmentation in tumor detection framework. Chapter Four, present the experimental results and discusses he findings. Chapter Five presents a brief summary of the thesis findings and the future direction

BRAIN TUMOR DETECTION USING IMAGE SEGMENTATION

Sharing is caring!

Leave a Reply