An Improved Gaussian Filter Technique for Biomedical Image Processing: An Early Lung Cancer Detection Technique

ABSTRACT For accurate and high rate of lung cancer detection using image processing, an effective preprocessing technique is required. A quality preprocessing technique is necessary to ensure effective removal of noise that interferes with the features of the image and hence improves lung cancer detection rate and accuracy. In this research, an Improved Gaussian Filter (IGF) technique was developed for effective lung image preprocessing. For image segmentation, Otsu thresholding method was used. The binarization was used for classification and Matlab as the simulation software. The filtering performance of the developed method was compared with the filtering performance of optimized Gaussian Filter (GF), the result showed that PSNR values obtained using improved Gaussian filter has an average gain of 1.2557dB over the PSNR values obtained using the optimized Gaussian Filter (GF). The detection rate and accuracy of the output from the Improved Gaussian filter was compared to the detection rate and accuracy of the output of the Gaussian filter and the result showed an improvement in average lung cancer detection rate and accuracy of 17.5% and 2.68% respectively when Improved Gaussian filter was used.

TABLE OF CONTENT

TITLE PAGE - - - - - - - - - i

APPROVAL PAGE - - - - - - - - - - ii

CERTIFICATION - - - - - - - - iii

DEDICATION - - - - - - --- iv

ACKNOWLEDGEMENT - - - - - - - - v

ABSTRACT - - - - - - - - - vi

TABLE OF CONTENT - - - - - - - - - vii

LIST OF FIGURES - - - -- - - - xi

LIST OF TABLES - - - - - - -- xii

CHAPTER ONE: INTRODUCTION

1.1. Background of the Study - - - - - - - - - 1

1.2. Statement of the Problem - - - - - - - - - 5

1.3. Objectives - - - - - - - - - 5

1.4. Scope of the study - - - - - - - - - 5

1.5. Significance of Study - - - - - - - - - 6

1.6. Proposed Method - - - - - - - - - 6

1.7. Plan of Thesis - - - - - - - - - 6

CHAPTER TWO: LITERATURE REVIEW

2.1. Theory of digital image - - - - - - - - - 7

2.1.1. Types of Digital image - - - - - - - - 8

2.1.1.1. Grayscale image - - - - - - - - 8

2.1.1.1.1 Normalized grayscale image - - - - - - - - 9

2.1.1.1.2. Binary image - - - - - - - - 9

2.1.1.2. Colour image - - - - - - - - 10

2.1.1.3. Multispectral image - - - - - - - - 10

2.1.2. CT lung image - - - - -- - - 11

2.1.3. Generation of Digital image - - - - -- - - 11

2.1.4. Corruption of Digital image - - - - - - - - 12

viii

2.1.4.1. Noise in Digital Lung Image - - - - - - - - 13

2.2. Image Processing Performance Metrics - - - - - - - - 14

2.2.1. Preprocessing Performance Metrics - - -- - -- - 14

2.2.1.1. Peak signal-to-noise ratio - - - - - - - 15

2.2.1.2. Mean square error - - - - - - - 15

2.2.1.3. PSNR Gain - - - - - - - 16

2.2.2. Cancer detection Performance Metrics - - - - - - -- 16

2.2.2.1. True acceptance rate - - - - - - -- 16

2.2.2.2. False acceptance rate - - - - - - -- 17

2.2.2.3. Pixel error rate - - - - - - -- 17

2.2.2.4. Recognition accuracy - - - - - - -- 18

2.3. Image preprocessing - - - - - - - - - 18

2.3.1. Types of preprocessing operations - - - - -- - - 20

2.3.1.1. Global preprocessing operations - - - - - - -- 20

2.3.1.2. Local preprocessing operations - - - - - - -- 20

2.3.1.2.1. Goal based preprocessing operation - - - - - - -- 21

2.3.1.2.1.1. Local smoothening operation - - - - - - -- 21

2.3.1.2.1.2. Local gradient operation - - - -- - - - 21

2.3.1.2.2. property based preprocessing operation - - - - - - -- 22

2.3.1.2.2.1. Linear filtering operation - - - -- - --- 22

2.3.1.2.2.2. Nonlinear filtering operation - - - - - - -- 23

2.3.2. Convolution Operations - - - - -- - - 23

2.3.3. Homomorphic filtering - - - - - - -- 25

2.3.4. Suitability of preprocessing filters - - - - - - -- 25

2.3.5. Preprocessing Techniques - - - - - - -- 26

2.3.5.1. Spatial domain techniques - - - - - - -- 26

2.3.5.1.1. Gaussian filter - - - - - - -- 27

2.3.5.1.1.1. Standard Gaussian filter - - - - - - -- 28

2.3.5.1.1.1.1. 1-dimensional Gaussian filter - - - - - - 28

2.3.5.1.1.1.2. 2-dimensional Gaussian filter - - - - - - -- 35

2.3.5.1.1.2. Recursive Gaussian filter - - - - - - -- 40

ix

2.3.5.1.2. Gabor filter - - - - - - - - - 41

2.3.5.1.3. Median filter - - - - - - - - - 44

2.3.5.1.3.1. Standard median filter - - - - - - - - 45

2.3.5.1.3.2. Recursive median filter - - - - - - - - 45

2.3.5.1.4. Averaging filter - - - - - - - - - 46

2.3.5.1.4.1. Standard averaging filter - - - - - - - - 47

2.3.5.1.4.2. Recursive averaging filter - - - - - - - - 48

2.3.5.1.5. Wiener filter - - - - - - - - 50

2.3.5.2. Frequency domain filtering - - - - - - - - - 51

2.3.5.2.1. Fourier Transforms filtering - - - - - - - - - 51

2.4. Image segmentation - - - - - - - - - 54

2.4.1. Thresholding technique - - - - - - - - - 54

2.4.2. Watershed segmentation approach - - - - - - - - - 58

2.5. Image Classification - - - - - - - - - 61

2.5.1. Binarization method - - - - - - - - - 61

2.5.2. Masking method - - - - - - - - - 63

2.6. Overview of Related Works - - - - - - - - - 64

2.7. Related Works - - - - - - - - - 64

2.8. Proposed Research - - - - - - - - - 67

CHAPTER THREE: RESEARCH METHODOLOGY

3.1. Introduction - - - - - - - - - 69

3.2. Proposed model - - - - - - - - - 69

3.3. Description of the technique - - - - - - - - 73

3.4. Steps in applying the proposed algorithm - - - - - - - - 73

CHAPTER FOUR: RESULTS AND DISCUSSION

4.1. Introduction - - - - - - - - - 75

4.2. Simulation and Analysis - - - - - - - - - 77

4.2.1. Filtering performance - - - - - - - - - 77

4.2.2. Cancer detection performance - - - - - - - - - 80

4.2.2.1. Cancer detection rate - - - - - - - - - 81

4.2.2.1. Cancer detection accuracy - - - - - - - - - 82

x

4.3. Model validation - --- - -- - - 84

4.3.1. Filtering performance validation - - - - - - -- 85

4.3.2. Cancer detection performance validation - - - - - - -- 87

4.3.2.1. Cancer detection rate validation - - - - - - - - 88

4.3.3.2. Cancer detection accuracy validation - - -- - -- -- 89

4.4. Noisy lung images and filter outputs - - - - - - - - 91

CHAPTER FIVE: RECOMMENDATION AND CONCLUSION

5.1. Summary - - - - - - - - - 93

5.2. Achievement - - - - - - - - - 93

5.3. Recommendation - - - - - - - - - 93

5.4. Conclusion - - - - - - - - - 94

REFERENCES - - - - - - - - - 95

APPENDIXES

Appendix A Definition of key terms - - - -- -- - 103

Appendix B Formation of DNA adducts - - -- - - -- 105

Appendix C Representation of tracheobronchial system - - - - - - - 106

Appendix D Representation of digital images using values - - - - - -- - 107

Appendix E MatLab code for the work - - - -- - - - 108

Appendix F CT lung images used in the work - ---- -- -- 113

Appendix E Greek letters used and their meanings - - - -- - - - 114