Abstract

The brain is the command center of our nervous system, which enables thoughts,
memories, movements, and emotions. In other words, it is the most important or-
gan in the human body. The human brain is very vulnerable to tumors, as merely
growing old can be the cause of a tumor. Furthermore, the effects of a tumor can
be fatal to a person because, as the tumor grows inside the brain, it can deform
the structure of the brain and cause several diseases, the most fatal being cancer
in the brain. Hence, to prevent such severe diseases, early detection of tumors is
critical for a patient’s treatment. Moreover, modern technology has emerged to
excellent heights, as MRI scans and CT scans can detect brain tumor regions.
However, to accurately detect where the tumor is situated, a team of doctors is
still needed to this day. Therefore, we have planned to use convolutional neural
Networks to de- velop a faster and inexpensive method to detect tumors from MRI
images in the early stages. Moreover, we plan to develop a system where our
proposed CNN model will be able to detect tumors as well as identify three types
of tumors, which are glioma, meningioma and pituitary tumors. Also, if there are
no tumors, the system should be able to detect them too. To develop our
proposed model, we have used data pre-processing techniques with a combination
of gray scaling, One encoding, and CLAHE. Also, we have used a dataset of 6484
MRI images, segment- ing them by testing and training. To compare and analyze
our proposed model’s performance, we have tested and trained seven pre-trained
models with the same dataset. The models are Vgg16, Vgg19, ResNet50,
InceptionV3, DenseNet-121, EfficientNetB0, MobileNet and we received the
following testing accuracy accord- ingly: 93.37%, 92.42%, 75.38%, 91.48%, 94.89%,
23.30% and 96.02%. However, the
testing accuracy of our proposed model surpassed all the other pre-trained models,
as it gained 98.11% accuracy in testing. In conclusion, we have aimed to build a
CNN model that exceeds all the other CNN models in terms of overall performance,
which is why we have integrated a sufficient number of parameters to handle any
unfavorable situations; however, the parameters are set in such a way that the over-
all system does not clutter and remains lightweight.

Keywords: Brain Tumor, CNN, accuracy, MRI, Deep learning, Machine learn-
ing, Tumor Detection, Neural Network, Gray Scaling, One hot encoding, Clahe,
Vgg16, Vgg19, Inception V3, DenseNet-121, EfficientNetB0, MobileNet
Acknowledgement
First and foremost, thanks to Allah, our thesis was completed without substantial
interruption. Second, Our preference is to thank our overseer, Dr. Rafeed Rahman,
and our co-supervisor, Dr. Iqbal. Hossain and Dibyo Fabian Dofadar, sir, for your
kind assistance and counsel in our endeavor.work They assisted us whenever we
needed it. All of the feedback they provided was beneficial to our further efforts.
Finally, without the unwavering assistance of our parents, it might not be feasi-
ble. We are almost done with our education because of their amazing prayers and
support.
Dedication
A dedication is a heartfelt expression of gratitude and connection from the author
to another individual. It serves as a poignant reminder of the bonds we share and
the appreciation we hold for those who have played a significant role in our creative
journey. Whether occupying a single line or spanning several, the dedication en-
capsulates the profound impact that individuals can have on our lives and work. It
stands as a testament to the power of human connection and the enduring value of
showing appreciation for those who inspire and support us on our creative path.This
project is dedicated to the countless individuals who have served as a wellspring of
inspiration. From mentors who shared their wisdom to friends who offered unwaver-
ing support, and from the great minds whose ideas sparked our imagination to the
everyday heroes who embody resilience and determination, it is to each and every
one of you that we extend our deepest gratitude. Your stories, your insights, and
your unwavering belief in the power of innovation have fueled our journey and made
this project possible. This dedication is a small token of our appreciation for the
inspiration you have provided, and a promise to carry forward the torch of creativity
and excellence that you have ignited in us.
Nomenclature
CNN - Convolutional Neural Network
ACNN - Artificial Convolutional Neural Network
R-CNN - Regional Convolutional Neural
Network Vgg16 - Visual Geometry Group 16
Vgg19 - Visual Geometry Group 19
ML - Machine Learning
CAD - Computer Aided Diagnosis
ReLU - Rectified Linear Unit
SGD - Stochastic Gradient Descent
PPV - Positive Predictive Value
NPV - Negative Predictive Value
MCC - Matthews Correlation Coefficient
VOTT - Visual Object Tagging Tool
ROIs - Regions of Interest
RPN - Regional Proposal Network
NMS - Non-maximum Suppression
E2E - End to End
MRI - Magnetic Resonance Image
CT - Computed Tomography
FNN - Fully Connected Layer
Table of Contents

Declaration i

Approval ii

Abstract iv

Acknowledgment v

Dedication vi

Nomenclature vii

Table of Contents viii

List of Figures x

List of Tables 1

1 Introduction 2
1.1 Contribution:.......................................................................................................3
1.2 Research Problem...............................................................................................4
1.3 Research Objective.............................................................................................5

2 Literature Review 6

3 Work Plan 14

4 Description of the Dataset 16

4.1 Proposed Methodology :.................................................................................16
4.2 Data Analysis :................................................................................................16
4.3 Input Data :.....................................................................................................17
4.4 Data Sample :..................................................................................................17
4.5 Data Preprocessing :.......................................................................................18

5 Proposed Model Implementation 19

5.1 Model Architecture Using CNN :.................................................................19
5.2 Proposed CNN Model.....................................................................................20
5.2.1 Convulation Layer................................................................................20
5.2.2 Batch Normalization :........................................................................20
5.2.3 Max Pooling Layer :...........................................................................20
5.2.4 Fully Connected Layer :.....................................................................21
 5.3 Proposed Model Summary :...........................................................................21
5.4 Pre-Trained Model...........................................................................................23
5.4.1 VGG 16:...............................................................................................23
5.4.2 VGG 19................................................................................................23
5.4.3 Resnet-50..............................................................................................23
5.4.4 MobileNet.............................................................................................24
5.4.5 InceptionV3...........................................................................................24
5.4.6 EfficientNetB0.......................................................................................25
5.4.7 DenseNet121:.......................................................................................25

Performance Analysis 26
6.1 Performance Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
6.2 Performance Parameter . . . . . . . . . . . . . . . . . . . . . . . . . . 26
6.3 Performance of Proposed Model . . . . . . . . . . . . . . . . . . . . . 27
6.3.1 Performance of proposed Model with other Data-set: . . . . 28
.
6.4 Performance of Pre-trained Model . . . . . . . . . . . . . . . . . . . . 30
6.4.1 Vgg16 : . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
.
6.4.2 Vgg19 : . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
.
6.4.3 Resent50 : . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
6.4.4 Inceptionv3: . . . . . . . . . . . . . . . . . . . . . . . . . . 33
.
6.4.5 DenseNet121: . . . . . . . . . . . . . . . . . . . . . . . . . 34
.
6.4.6 EfficientNetB0: . . . . . . . . . . . . . . . . . . . . . . . . . . 35
6.4.7 MobileNet: . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
.
6.5 Performance Matrix: . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
6.5.1 Proposed CNN Model: . . . . . . . . . . . . . . . . . . . . . . 37
6.5.2 Pre-trained Models: . . . . . . . . . . . . . . . . . . . . . . . . 42
6.6 Compare and Analysis : . . . . . . . . . . . . . . . . . . . . . . . . 49
.

Conclusion 51
7.1 Future Work: . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

Bibliography 52
List of Figures

3.1 Work Flow Diagram.........................................................................................15

4.1 Data Sample.....................................................................................................17

5.1 Working Process of CNN.................................................................................19

6.1 Graph of training and testing accuracy of proposed model........................27

6.2 Graph of training and testing loss of proposed model.................................28
6.3 Graph of training and testing accuracy of CNN on other Dataset.............29
6.4 Graph of training and testing loss of CNN on other Dataset......................29
6.5 Graph of training and testing accuracy of VGG16 model..........................30
6.6 Graph of training and testing loss of VGG16 model...................................30
6.7 Graph of training and testing accuracy of VGG19 model..........................31
6.8 Graph of training and testing loss of VGG19 model...................................31
6.9 Graph of training and testing accuracy of Resnet50 model........................32
6.10 Graph of training and testing loss of Resnet50 model................................32
6.11 Graph of training and testing accuracy of Inceptionv3 model...................33
6.12 Graph of training and testing loss of Inceptionv3 model............................33
6.13 Graph of training and testing accuracy of DenseNet-121 model................34
6.14 Graph of training and testing loss of DenseNet-121 model........................34
6.15 Graph of training and testing accuracy of EfficientNetB0 model..............35
6.16 Graph of training and testing loss of EfficientNetB0 model.......................35
6.17 Graph of training and testing accuracy of MobileNet model.....................36
6.18 Graph of training and testing loss of MobileNet model..............................36
6.19 F1 score of Proposed CNN model..................................................................38
6.20 Confusion Matrix of Proposed CNN model..................................................39
6.21 ROC Curves of Proposed CNN Model..........................................................41
6.22 F1 score of Vgg16 model.................................................................................42
6.23 Confusion Matrix of Vgg16 model.................................................................42
6.24 F1 score and Confusion Matrix of Vgg19 model..........................................43
6.25 F1 score and Confusion Matrix of Vgg19 model..........................................43
6.26 F1 score and Confusion Matrix of ResNet50 model....................................44
6.27 F1 score and Confusion Matrix of ResNet50 model....................................44
6.28 F1 score and Confusion Matrix of inceptionV3 model................................45
6.29 F1 score and Confusion Matrix of inceptionV3 model................................45
6.30 F1 score and Confusion Matrix of DenseNet-121 model.............................46
6.31 F1 score and Confusion Matrix of DenseNet-121 model.............................46
6.32 F1 score of EfficientNetB0 model..................................................................47
6.33 Confusion Matrix of EfficientNetB0 model...................................................47
6.34 F1 score of MobileNet model.........................................................................48
6.35 Confusion Matrix of MobileNet model..........................................................48
6.36 Bar chart of all model’s accuracy...................................................................50