Вы находитесь на странице: 1из 8

ACKNOWLEDGEMENT

First of all we thank “GOD ALMIGHTY” for his blessings that

gave us courage and mental ability to present this project and make it a

great success.

We are thankful to the CHAIRMAN of our college for his

encouragement and support.

We are bound to record our heartfelt thanks to our beloved

PRINCIPAL for his inspiration and encouragement he has provided us

for the successful completion of our project.

We express our sense of gratitude to our respected HEAD OF

THE DEPARTMENT for her much valuable support, unfledged

attention and direction, which kept this project on track.

We have immense pleasure in thanking our project CO-

ORDINATOR for her active involvement, encouragement,

stimulating suggestion and guidance which have helped us all the

time.

We are expressing our sincere thanks to our project

SUPERVISOR for her guidance which helped us to improve our

project. We are also thankful to all the staff members of our

department and friends who were involved directly or indirectly for

the successful completion of our project.

iii
ABSTRACT

Automated melanoma recognition in dermoscopy images is a

very challenging task due to the low contrast of skin lesions, the

huge intraclass variation of melanomas, the high degree of visual

similarity between melanoma and non-melanoma lesions, and the

existence of many artifacts in the image. In order to meet these

challenges, we propose a novel method for melanoma recognition by

leveraging very deep convolutional neural networks (CNNs). This

technique can ensure that our networks benefit from the performance

gains achieved by increasing network depth. Then, we construct a

fully convolutional residual network for accurate skin lesion

segmentation. The proposed segmentation technique and the

classification method to form a two-stage framework. This

framework enables the classification network to extract more

representative and specific features based on segmented results

instead of the whole dermoscopy images, further alleviating the

insufficiency of training data.

Keywords: melanoma, very deep convolution neural network,

residual learning, fully convolutional neural network, skin lesion

analysis.

iv
TABLE OF CONTENTS

CHAPTER NO TITLE PAGE NO

ABSTRACT iv

TABLE OF CONTENTS v

LIST OF TABLE ix

LIST OF FIGURES x

LIST OF ABBREVIATIONS xi

1 INTRODUCTION 1

1.1 SKIN CANCER 1

1.2 OBJECTIVE 1

1.3 TYPES OF SKIN CANCER 1

1.3.1 Basal skin cancer 2

1.3.2 Squamous cell skin cancer 2

1.3.3 Melanoma skin cancer 2

1.4 MELANOMA 2

1.5 CAUSES 3

v
1.6 TREATMENT 3

1.6.1 Immunotherapy 3

1.7 WHAT IS DERMOSCOPY 4

1.8 CONVOLUTIONAL NEURAL NETWORK 5

2 LITERATURE SURVEY 7

3 PROPOSED SYSTEM 13

4 SYSTEM ANALYSIS 16

5 PRINCIPLE AND APPLICATION OF IMAGE 28

5.1 IMAGE PROCESSING TECHNIQUE 28

5.2 IMAGE PROCESSING 28

5.3 TYPES IMAGE PROCESSING 30

5.3.1 Optical processing 30

5.3.2 Electronic processing 30

5.4 APPLICATIONS 31

5.5 STAGES IN IMAGE PROCESSING 32

5.5.1 Image aqusition 32

5.5.2 Recognition and interpretation 32

5.5.3 Segmentation 32

5.5.4 Represent and description 32

5.5.5 Knowledge base 32

5.6 THRESHOLDING 33

vi
5.7 CONNECTIVITY 33

5.8 NOISE REDUCTION 33

5.9 CONVOLUTION MASKS 34

5.10 EDGE DETECTION 34

5.11 REAL TIME IMAGE PROCESSING 35

5.12 IMAGE PROCESSING TOOLBOX 35

5.13 IMAGE TYPES 35

5.14 HISTOGRAM EQUALIZATION 36

5.15 SIMPLE THRESHOLD OF SEGMENTATION 37

5.16 MEDIAN FILTER 38

6 SOFTWARE DESCRIPTION 39

7 EXPERIMENTAL RESULT AND ANALYSIS 44

7.1 TEST IMAGES 44

7.2 SIMULATION RESULT 45

8 CONCLUSION 51

APPENDIX 52

REFERENCE 64

vii
LIST OF TABLE

FIGURE NO. TITLE PAGE NO.

4.4 List of statistics 21

viii
LIST OF FIGURES

FIGURE NO. TITLE PAGE NO.

3.1 Block diagram of existing system 14

3.2 Block diagram for the proposed system 15

4.1 Flow diagram of computerized melanoma 17

Skin cancer detection

4.2 Sample acquired image 18

4.3 HSV color space 21

7.1 Sample database images 48

7.2 Test image 49

7.3 Gray scale image 50

7.4 Intensity adjusted image 50

7.5 Binary image 51

7.6 Binary image without small object 51

7.7 Imfill image 52

7.8 Boundary image 52

7.9 ROI image 53

7.10 Classified output 54

ix
LIST OF ABBREVIATIONS

CADs - Computer Aided Detection System

CDF - Cumulative Distribution Function

CNN - Convolutional Neural Network

EISPACK - Eigen System Package

FCRN - Fully Convolutional Residual Network

GLCM - Gray level co-occurrence Matrix

GVF - Gradient Vector Flow

LINPACK - Linear System Package

MATLAB - MATrix LABoratory

ROI - Region Of Interest

Вам также может понравиться