CHAPTER ONE - INTRODUCTION

1.1          Introduction

Many people both children and adults have died due to the inability of doctors to be able to detect some hidden bBrain tumors in MR images or due to the misinterpretation of these images. Tumor is a mass of tissue that grows out of control of the normal forces that regulates growth. The complex brain tumors can be separated into two general categories depending on the tumor’s origin, growth pattern and malignancy. Primary brain tumors are tumors that arise from cells in the brain or from the covering of the brain. A secondary or metastatic brain tumor occurs when cancer cells spread to the brain from a primary cancer in another part of the body. Brain tumors can have consequences that can affect their quality of life. They can affect various parts of the brain which control various functions. That means the effects of a brain tumor one may experience depends on the location, size, and aggressiveness of the tumor. Symptoms of brain cancer may be ambiguous, or precise. Cerebral cancer signs may be vague, or correct.  The presence of a brain tumor may cause harm to healthy brain tissue, affecting the area's normal function. Brain tumors can give rise to increased intracranial pressure, leading to headaches or consciousness impairment. Brain tumors have this important effect on the brain as they cause the non-tumoral brain tissue not only to adapt to the presence of a growing mass but also to the penetration of tumor cells into healthy brain tissue. Machine Vision using Artificial Neural Networks (ANNs) are mathematical analogues of biological neural systems, in the sense that they are made up of a parallel interconnected system of nodes, called neurons. The evolution of digital computers as well as the development of modern theories for learning and information processing led to the emergence of Computational Intelligence (CI) engineering. Artificial Neural Networks (ANNs), Genetic Algorithms (GAs) and Fuzzy Logic are CI non-symbolic learning approaches for solving problems. The huge mass of applications, which ANNs have been used with satisfactory results, has supported their rapid growth. Fields that ANNs were used are image processing, environmental problems, Climate study, and financial analysis.

The main goal of computerized brain tumor detection is to obtain important clinical information regarding the presence of tumor, the location, and the type of tumor in the MR image that would be fed into the ANN system. The information obtained through clinical imaging can guide and control any future interventions and thus leads to the correct diag nosis and treatment of the tumor. Computer-aided brain tumor diagnosis from MRI images consists of tumor detection, segmentation, and classification processes. Detection of brain tumors is the first step in the more ambitious goal of developing a fully integrated computer-aided diagnostic device that is quicker and more precise than any neurologist. I am aiming to explore the critical first step in this project.

1.2          Background of Study

According to the statistics of NBTF (National Brain Tumor Foundation) for research; about 29,000 people in U.S are diagnosed with primary brain tumor and among them nearly 13,000 die. About 16 out of 1,000 each cancer found in U.K are brain tumors. In India about 80,271 people are affected by tumor (Logeswari & Karnan, 2010). Therefore, Brain Tumor is one of the leading reasons for heightening the mortality among children and adults (Padma & Sukenesh, 2013).

Most Research in developed countries show that the number of people who develop brain tumors and die from them has increased perhaps as much as 300 over past three decades (Bhattacharyya & Kim 2011).

The shortcomings of conventional computer-aided mammography detection (CAD) systems, the significant importance of early breast cancer detection, and the high effect of patient false diagnosis push researchers to investigate deep learning (DL) methods for mammograms (MGs).

1.3          Statement of Problem

Brain tumors are a heterogeneous group of neoplasms of the central nervous system that occur within or adjacent to the brain. Also, the tumor's position inside the brain has a profound impact on the symptoms of the patient, the surgical treatment options, and the possibility of a conclusive diagnosis. Currently, brain tumors are only identified after the onset of neurological symptoms by MRI. There are no early warning techniques in use, except in persons considered to be at risk for types of brain tumors due to their genetics. Brain tumors are a heterogeneous group of central nervous system neoplasms that occur within or adjacent to the brain. Furthermore, the location of the tumor within the brain has a profound effect on the brain. The patient's symptoms, the treatment options for the surgery, and the likelihood of a definitive diagnosis. Also, the position of the tumor in the brain significantly changes the risk of neurological toxicity which alters the patient's quality of life. Usually, brain tumors are only detected after MRI has established neurological symptoms. There are no early warning methods in use, except for individuals who are considered at risk due to their genetics for unique forms of brain tumors.

 Figure 1.1: Brain Anatomy (Retrieved From: National Cancer Institute, US)

1.4          Aim and Objectives

1.4.1      Aim

The aim of this research is to use an Artificial Neural Network (ANN) system to detect the presence of any form of Brain Tumor from a MR Image that is being fed into the ANN system with the least possible error of detection.

1.4.2      Objectives

1.            To acquire MRI images for the test

2.            To develop the algorithm for the Artificial Neural Network

3.            To carry out segmentation, blurring, and cropping

4.            Finally, to create a neural network algorithm for the brain tumor using the VGG_16 model of the python program to determine the presence of tumor in the images or not.

 1.5         Scope of Study

This system only detects the presence of brain tumor with respect to the MR images that is being fed into the ANN system to train it, that is, the system won’t recognize a brain tumor image it has not been trained with. Note, that this system can only detect, it does not cure or eliminate the  presence of brain tumor. More so, in this project, we are utilizing the VGG_16 model of the CNN in the Python Language.