Quantitative analysis of lung morphology and function in computed tomographic images [Elektronische Ressource] / von Jan-Martin Kuhnigk

Jan-Martin Kuhnigk

Sujets

• Gesundheit

Informations

• Publié par: universitat_bremen
• Publié le: 01 janvier 2008
• Nombre de lectures: 26
• Langue: English
• Poids de l'ouvrage: 20 Mo

Extrait

Quantitative Analysis of Lung Morphology and Function
in Computed Tomographic Images
JAN-MARTIN KUHNIGK
MeVis Research
Center for Medical Image Computing
2008Datum des Promotionskolloquiums: 7. Juli 2008
Gutachter: Prof. Dr. rer. nat. Heinz-Otto Peitgen, MeVis Research, Bremen
Prof. Dr. rer. nat. Frieder Nake, Universitaet BremenQuantitative Analysis of Lung Morphology and Function
in Computed Tomographic Images
von Dipl.-Inf. Jan-Martin Kuhnigk
DISSERTATION
zur Erlangung des Grades eines Doktors der Ingenieurwissenschaften
– Dr.-Ing. –
Vorgelegt im Fachbereich 3 (Mathematik und Informatik)
der Universität Bremen
im Mai 2008c 2008 Jan-Martin Kuhnigk. All rights reserved.
The work was conducted at MeVis Research,
Center for Medical Image Computing.
Jan-Martin Kuhnigk
MeVis Research, Universitaetsallee 29, 28359 Bremen, Germany
Email: [e-mail]
Introduction 7
Overview.............................................. 7
Motivation ............................................. 7
Objectives 12
I Background 15
1 Technical Background ................................... 16
1.1 3d Image Acquisition ............................... 16
1.1.1 Computed Tomography (CT) ..................... 16
1.1.2 Other 3d Imaging Techniques 20
1.1.3 Conclusion ................................ 2
1.2 Digital Image Analysis 2
1.3 Development Platform 23
2 Medical Background .................................... 24
2.1 Lung Anatomy ................................... 24
2.2 Lung Diseases 26
2.2.1 Parenchymal Diseases.......................... 26
2.2.2 Primary Lung Cancer 28
2.2.3 Lung Metastasis ............................. 29
2.3 Clinical Applications Covered by the Thesis .................. 30
2.3.1 Diagnosis and Monitoring of Parenchymal Disease ......... 30
2.3.2 D of Primary Lung Cancer 31
2.3.3 Chemotherapy Follow-Up ....................... 32
2.3.4 Risk Analysis for Lung Resection Surgery .............. 34
3 Challenges in Pulmonary Image Analysis ........................ 35
3.1 Evaluation Issues .................................. 35
3.2 Volume Averaging ................................. 38
3.3 From2dto3dAnalysis .............................. 39
3.4 Verification and Interactive Refinement in Clinical Applications ....... 41
II Lung Segmentation 43
1Introduction ......................................... 45
1.1 Medical Background ................................ 45
1.2 Segmentation of the Lungs in Thoracic CT Scans ................ 45
1.3 Chapter Overview ................................. 47
2 Methods ........................................... 47
2.1 Related Work .................................... 47
2.2 Method Description 494T ABLE OF CONTENTS
2.2.1 Pulmonary Airspace Segmentation .................. 49
2.2.2 Bronchi Segmentation .......................... 52
2.2.3 Lung Separation ............................. 54
2.2.4 Lung Mask Closing ........................... 54
2.3 Implementation and Optimization ........................ 58
2.3.1 Confinement to Regions of Interest 58
2.3.2 Downscaling ............................... 58
2.3.3 Downsampling.............................. 59
2.4 Limitations ..................................... 61
3 Experiments and Results .................................. 63
3.1 General Experiment Setup 63
3.1.1 Evaluation Data ............................. 63
3.1.2 Image Data Processing ......................... 64
3.1.3 Study Overview 65
3.2 Performance Evaluation 65
3.3 Quality Evaluation ................................. 65
3.3.1 Populations................................ 65
3.3.2 Evaluation Criteria ........................... 6
3.3.3 Segmentation Quality Scoring ..................... 67
3.3.4 Results .................................. 72
3.4 Reproducibility Evaluation ............................ 76
3.4.1 Population, Limitations, and Evaluation Parameters ........ 76
3.4.2 Results 7
4 Discussion .......................................... 78
4.1 Segmentation Quality ............................... 79
4.2 Inter-Scan Reproducibility............................. 80
4.3 Computational Performance ........................... 80
4.4 Further Applications ................................ 81
4.5 Future Work ..................................... 81
5 Summary and Conclusion ................................. 82
III Lung Lobe Segmentation 85
1Introduction ......................................... 87
1.1 Medical Background 87
1.2 Segmentation of the Lobar Fissures in CT Data ................. 87
1.3 Chapter Overview 89
2 Methods ........................................... 89
2.1 Related Work .................................... 89
2.2 Method Description ................................ 91
2.2.1 Vascular Segmentation ......................... 93
2.2.2 V Distance Computation .................... 94
2.2.3 Combination of Vascular and Fissure Information.......... 94
2.2.4 Bronchi Analysis............................. 94
2.2.5 Lobe Segmentation by Marker-Based Watershed Transform .... 96
2.3 Implementation and Optimization ........................ 97
2.3.1 Performance ............................... 97
2.3.2 Robustness 98
3 Experiments and Results .................................. 98
3.1 General Experiment Setup 98
3.1.1 Evaluation Data 98TABLE OF CONTENTS 5
3.1.2 Image Data Processing ......................... 9
3.1.3 Study Overview ............................. 10
3.2 Segmentation Quality ............................... 10
3.2.1 Populations................................ 101
3.2.2 Evaluation Criteria ........................... 102
3.2.3 Segmentation Quality Scoring ..................... 103
3.2.4 Results .................................. 106
3.3 Interobserver Agreement 10
3.3.1 Experiment Setup ............................ 10
3.3.2 Results 12
3.4 Performance..................................... 12
3.4.1 Setup ................................... 112
3.4.2 Results 113
4 Discussion .......................................... 13
4.1 Segmentation Quality ............................... 14
4.2 Inter-Observer Agreement............................. 17
4.3 Computational Performance ........................... 17
4.4 Future Work 18
5 Summary and Conclusion ................................. 120
IV Segmentation and Volumetry of Pulmonary Nodules 121
1Introduction ......................................... 123
1.1 Medical Background ................................ 123
1.2 Segmentation of Pulmonary Lesions ....................... 124
1.3 Volumetry of Pulmonary Lesions......................... 125
1.4 Chapter Overview 126
2 Segmentation Method ................................... 126
2.1 Related Work .................................... 126
2.2 Method Description 127
2.2.1 Initialization ............................... 128
2.2.2 Separation from the Chest Wall .................... 128
2.2.3 S from Attached Vasculature ................ 129
2.2.4 Boundary Refinement .......................... 135
2.2.5 Extensions ................................ 135
3 Volumetry Method ..................................... 137
3.1 Introduction and Related Work 137
3.2 Segmentation-based Partial Volume Analysis (SPVA) ............. 139
4 Experiments and Results .................................. 141
4.1 Accuracy Evaluation: Phantom Studies ..................... 142
4.1.1 Software Phantom Study ........................ 142
4.1.2 Physical Phantom Study 143
4.2 Inter-Observer and Inter-Scan Reproducibility: In Vivo Study ........ 14
4.3 Reproducibility Regarding Imaging Protocol and Segmentation Threshold
Variations: Phantom Study ............................ 145
4.3.1 Sensitivity to Threshold Variations .................. 147
4.3.2 Reproducibility Regarding Scanning and Reconstruction Parame-
ter Variations ............................... 148
5 Discussion .......................................... 149
5.1 Accuracy....................................... 150
5.2 In Vivo Inter-Scan Reproducibility ........................ 1526T ABLE OF CONTENTS
5.3 Benefits of Segmentation-based Partial Volume Analysis ........... 153
5.4 Method Generalization .............................. 154
6 Summary and Conclusion ................................. 15
V Applications and Clinical Evaluation 157
3D1TheMEVISPULMO Application ............................ 158
1.1 Application Framework 159
3D1.2 The MEVISPULMO Startup Screen ...................... 159
1.3 Lung Tumor Analysis (PULMOTREAT) ..................... 160
1.3.1 Medical Background .......................... 160
1.3.2 Application Overview 162
1.4 Quantitative Parenchyma Analysis (PULMOQUANT)............. 163
1.4.1 Medical Background 163
1.4.2 Application Overview 164
1.4.3 Verification and Interactive Refinement Policy............ 164
1.4.4 Morphological Lung Analysis ..................... 16
1.4.5 Quantitative Analysis and Result Examination ........... 169
1.5 Visualization for Lung Intervention Planning (INTERVENTIONPLANNER) . 172
2 Clinical Evaluation ..................................... 173
2.1 Tumor Volumetry Evaluation ........................... 173
2.2 Quantitative Parenchyma Analysis........................ 175
3 Discussion and Conclusi