Brain Tumor Detection Using Deep Learning

ABSTRACT

Brain Tumor segmentation is one of the most crucial and arduous tasks in the terrain of medical
image processing as a human-assisted manual classification can result in inaccurate prediction
and diagnosis. Moreover, it is an aggravating task when there is a large amount of data present to
be assisted. Brain tumors have high diversity in appearance and there is a similarity between
tumor and normal tissues and thus the extraction of tumor regions from images becomes
unyielding. we proposed a method to extract brain tumor from 2D Magnetic Resonance brain
Images (MRI) by Fuzzy C-Means clustering algorithm which was followed by traditional
classifiers and convolution neural network. The experimental study was carried on a real-time
dataset with diverse tumor sizes, locations, shapes, and different image intensities.
 INTRODUCTION

The early detection and treatment of brain tumour helps in early diagnosis which aids in reducing
mortality rate. Image processing has been widespread in recent years and it has been an
inevitable part in the medical field also. The abnormal growth of cells in the brain causes brain
tumour. Brain tumour is also referred to as intracranial neoplasm. The two types of tumours are
malignant and benign tumours. Standard MRI sequences are generally used to differentiate
between different types of brain tumours based on visual qualities and contrast texture analysis
of the soft tissue. More than 120 classes of brain tumours are known to be classified in four
levels according to the level malignancy by the World Health Organization (WHO) . All types of
brain tumours evoke some symptoms based on the affected region of the brain. The major
symptoms may include headaches, seizures, vision problems, vomiting, mental changes, memory
lapses, balance losing etc . Incidence of brain tumours are due to genetics, ionizing radiation
mobile phones, extremely low frequency magnetic fields, chemicals, head trauma and injury,
immune factors like viruses, allergies, infections, etc. The malignant tumours, also known as
cancerous tumours, are of two types - primary tumours, which start from the brain, and
secondary tumours, which originate somewhere and spread to the brain. The risk factors for brain
tumour are exposure to vinyl chloride, neurofibromatosis, ionising radiations and so on. The
various diagnostic methods are computed tomography, magnetic resonance imaging, tissue
biopsy etc. Better treatments are now available for brain tumours. There is a chance of focal
neurological deficits, such as motor deficit, aphasia or visual field defects in the treatment. Side
effects can be avoided by measuring tumour size and time to tumour progression (TTP) .
Estimation of density of affected areas can give a better measurement in therapy. Deep learning
is a machine learning technique that instructs computers what to do as a human think and do in a
scenario. In deep learning, a computer model is able to do classification tasks from images,
sound or text. Sometimes human level performance is being exceeded by deep learning
techniques. The aim of this project is to build a system that would help in cancer detection from
MRI images through the convolution neural network. The proposed method was tested and
compared with the existing classification techniques to determine the accuracy of the proposed
method.
 EXISTING SYSTEM

This study proposes a computer aided detection approach to diagnose brain tumor in its early
stage using Mathematical Morphological Reconstruction (MMR). Image is pre-processed to
remove noise and artefacts and then segmented to find regions of interest with probable tumor. A
large number of textural and statistical features are extracted from the segmented image to
classify whether the brain tumor in the image is benign or malignant. Experimental results show
that the segmented images have a high accuracy while substantially reducing the computation
time. The study shows that the proposed solution can be used to diagnose brain tumor in patients
with a high success rate.
A computer-based method for defining tumor region in the brain using MRI images is
presented. A classification of brain into healthy brain or a brain having a tumor is first done
which is then followed by further classification into begin or malignant tumor. The algorithm
incorporates steps for preprocessing, image segmentation, feature extraction and image
classification using neural network techniques. Finally the tumor area is specified by region of
interest technique as confirmation step. A user friendly Matlab GUI program has been
constructed to test the proposed algorithm.
DISADVANTAGES OF EXISTING SYSTEM

 Classification is not accurate

 Inefficient
 Non-Robust
 Support Small Dataset
 PROPOSED SYSTEM

Convolutional Neural Network is broadly used in the field of Medical image processing. Over
the years lots of researchers tried to build a model which can detect the tumor more efficiently.
We tried to come up with an exemplary which can accurately classify the tumor from 2D Brain
MRI images. A fully-connected neural network can detect the tumor, but because of parameter
sharing and sparsity of connection, we adopted CNN for our model.
A Five-Layer Convolutional Neural Network is introduced and implemented for tumor
detection. The aggregated model consisting of seven stages including the hidden layers provides
us with the most prominent result for the apprehension of the tumor. Following is the proposed
methodology with a brief narration

Using convolutional layer as the beginner layer, an input shape of the MRI images is
generated which is 64*64*3 converting all the images into a homogeneous dimension. After
accumulating all the images in the same aspect, we created a convolutional kernel that is
convoluted with the input layer — administering with 32 convolutional filters of size 3*3 each
with the support of 3 channels tensors. ReLU is used as an activation function so that it’s not
corroborating with the output.

In this ConvNet architecture, progressively shorten the spatial size of the depiction for
diminishing the chunk of parameters and computational time of the network. Working on the
Brain MRI image can also cost the contamination of the overfitting and for this Max Pooling
layer perfectly works for this perception. For spatial data which substantiate with our input
image, we use MaxPooling2D for the model. This convolutional layer runs on 31*31*32
dimension. Because of divide the input images in both spatial dimensions, the pool size is (2, 2)
which means a tuple of two integers by which to downscale by vertically and horizontally.

ADVANTAGES OF PROPOSED SYSTEM

 Classification is accurate
 Efficient
 Robust
 Support Large Dataset
 SYSTEM REQUIREMENTS

HARDWARE REQUIREMENTS

 Processor- Intel (R) Core (TM) i3-4200U

 CPU - 1.6GHz

 RAM:4 GB

 Hard Disk:500 GB.

SOFTWARE REQUIREMENTS

 Operating System - Windows 10

 IDE - Google Colab
 Environment - Tenserflow
 Programming Language - Python
 Libraries - Numpy, Pandas
 FEASIBILITY STUDY
An important outcome of preliminary investigation is the determination that the system request is
feasible. This is possible only if it is feasible within limited resource and time. The different
feasibilities that have to be analyzed are

 Operational Feasibility
 Economic Feasibility
 Technical Feasibility

Operational Feasibility
Operational Feasibility deals with the study of prospects of the system to be developed.
This system operationally eliminates all the tensions of the Admin and helps him in effectively
tracking the project progress. This kind of automation will surely reduce the time and energy,
which previously consumed in manual work. Based on the study, the system is proved to be
operationally feasible.

Economic Feasibility

Economic Feasibility or Cost-benefit is an assessment of the economic justification for a

computer based project. As hardware was installed from the beginning & for lots of purposes
thus the cost on project of hardware is low. Since the system is a network based, any number of
employees connected to the LAN within that organization can use this tool from at anytime. The
Virtual Private Network is to be developed using the existing resources of the organization. So
the project is economically feasible.

Technical Feasibility

According to Roger S. Pressman, Technical Feasibility is the assessment of the technical

resources of the organization. The organization needs IBM compatible machines with a graphical
web browser connected to the Internet and Intranet. The system is developed for platform
Independent environment. Python, JavaScript, HTML, SQL server and WebLogic Server are
used to develop the system. The technical feasibility has been carried out. The system is
technically feasible for development and can be developed with the existing facility.