Segmentation of 3D Tubular Tree Structures in Medical Images ...

More documents

Recommendations

Info

10 Chapter 1. Introduction 1.1.2 Discussion The above presented techniques address the segmentation of vascular/airway structures in general, and not necessarily whole tubular tree structures. Considering these techniques with respect to the requirements and problems we aim at addressing in this work (Section 1.0.1), one gains insight into the challenges of this problem. Identifying an appropriate vessel/airway model that is valid in all possible cases is a non-trivial task, in particular when one considers pathology like stonsis/aneurysms/calcifications or simple bifurcations where the appearance and/or geometry of the tubular structures is disturbed. Also cases of local disturbances such as imaging/motion artifacts or partly overlapping image structures with the same gray-value (e.g. adjacent tumors or multiple partly overlapping vessel trees) are difficult to model. For these reasons, it seems illusory to identify a model that is valid in all cases. This becomes a problem when combined with the discussed extraction schemes. While the pre-processing methods are typically applied voxel-wise, an inadequate model leads to fragmented outputs, with false positive and false negative detections. The presented extraction schemes (region-growing, active-contour, and centerline-tracking methods) all require appropriate initializations and basically merge neighboring voxels or extend a known centerline according to the model. In case of conservative model assumptions this easily leads to premature termination of the (iterative) extension process, while in case of weak model assumptions the methods are prone to leakage. Methods based on minimal path-techniques show a higher robustness against such local failures of an inadequate model. However, they require appropriate start- and end-points for each vessel. For heavily branched tubular tree structures this is a practically infeasible task. In case of failure, the segmentation errors (under/over segmentation) that result from local model-inadequateness may have a strong influence on the resulting structure/topology of the segmented tubular trees. This is problematic, since the structural information is of vital interest for several applications (Section 1.0.1). 1.1.3 Vessel Mining To address some of the problems mentioned above, Beichel et al. [10] proposed a concept to enhance the robustness of vessel segmentation approaches against local disturbances. The idea behind the concept that they termed “vessel mining” is: “first identify all potential tubular structures in the search volume and then reconnect the found tube structures based on defined criteria like for example gray-value evidence” [10]. This approach was
1.2. A General Approach for Segmentation of Branched Tubular Networks 11 designed to avoid leakage into other structures and increase the robustness in cases of noise or anisotropic voxel size, compared to simpler methods like region growing. Beichel et al. presented a system based on this concept for liver portal vein tree segmentation that has been applied to a phantom dataset and to one routinely acquired liver CT dataset. A comparison to a simple region grower showed the increased robustness of their system. In later work of the same group [113, 114], the authors focused on refining parts of the approach for their main application – liver portal vein tree segmentation – where they improved the methods ability to identify vascular structures and where they increased the accuracy of the segmented vessel walls by utilizing the identified vascular structures as initialization of an intrinsic segmentation. The general approach for segmentation of branched tubular networks as described in the next section is based on these seminal ideas of “vessel mining” as proposed by Beichel et al. [10, 113]. 1.2 A General Approach for Segmentation of Branched Tubular Networks 3D tubular tree structures form sets of tubular structures that are connected with each other. Each of the tubular parts has its structure and surface delineating it from the background. To segment all kinds of branched tubular networks, the following general approach can be use (Fig. 1.4): 1. Extraction of tubular structures: Identify all tubular structures in a volume of interest and generate a structural representation for all of them. The resulting centerline-based representations describe already major parts of the branched tubular network, but they describe only the parts following the tube model, while in the branched tubular network parts may still be missing where the appearance strongly deviates from a typical tubular shape. This may be in case of branching areas, due to other cases where the shape of the vessels is severely perturbed such as stenosis, aneurysms, or tumors, due to other structures with similar gray in the image domain, or due to imaging artifacts. 2. Grouping and linkage into connected networks: Group the tubular structures belonging to the same tubular tree and link them together recovering the structures of the different branched tubular networks. The individual tubular structures of
Page 1: Graz University of Technology Insti
Page 5 and 6: Abstract The segmentation of tubula
Page 7 and 8: Kurzfassung Die Segmentierung tubul
Page 9: Acknowledgements I am deeply gratef
Page 13 and 14: Contents 1 Introduction 1 1.0.1 Req
Page 15: CONTENTS xv 8 Conclusion and Outloo
Page 18 and 19: xviii LIST OF FIGURES 4.2 Shape pri
Page 21: List of Tables 3.1 Average centerli
Page 24 and 25: 2 Chapter 1. Introduction (a) Porta
Page 26 and 27: 4 Chapter 1. Introduction (a) Volum
Page 28 and 29: 6 Chapter 1. Introduction (a) Inacc
Page 30 and 31: 8 Chapter 1. Introduction features
Page 34 and 35: 12 Chapter 1. Introduction Input vo
Page 36 and 37: 14 Chapter 1. Introduction part: Me
Page 38 and 39: 16 Chapter 1. Introduction in the l
Page 40 and 41: 18 Chapter 2. Extraction of Tubular
Page 70 and 71: 48 Chapter 3. Grouping and Linkage
Page 82 and 83:
60 Chapter 3. Grouping and Linkage
Page 84 and 85:
Page 86 and 87:
Page 88 and 89:
66 Chapter 4. Tube Segmentation Onl
Page 90 and 91:
68 Chapter 4. Tube Segmentation obt
Page 92 and 93:
70 Chapter 4. Tube Segmentation ass
Page 94 and 95:
72 Chapter 4. Tube Segmentation (a)
Page 96 and 97:
74 Chapter 4. Tube Segmentation Tab
Page 99 and 100:
Chapter 5 Liver Vascular Tree Segme
Page 101 and 102:
5.2. Method 79 (a) MIP of the datas
Page 103 and 104:
5.3. Evaluation and Results 81 5.3.
Page 105 and 106:
5.3. Evaluation and Results 83 cent
Page 107 and 108:
5.3. Evaluation and Results 85 (a)
Page 109 and 110:
Page 111 and 112:
Page 113 and 114:
Page 115 and 116:
5.4. Discussion 93 results. Selle
Page 117:
5.5. Conclusion 95 a prior to const
Page 120 and 121:
98 Chapter 6. Coronary Artery Tree
Page 122 and 123:
Page 124 and 125:
Page 126 and 127:
Page 128 and 129:
Page 131 and 132:
Chapter 7 Airway Tree Segmentation
Page 133 and 134:
7.2. Methods 111 shown in Fig. 7.1.
Page 135 and 136:
7.2. Methods 113 Additionally, the
Page 137 and 138:
7.3. Evaluation and Results 115 Gro
Page 139 and 140:
7.3. Evaluation and Results 117 Tab
Page 141 and 142:
7.4. Discussion 119 Figure 7.5: “
Page 143 and 144:
7.5. Conclusion 121 Comparison to o
Page 145 and 146:
7.5. Conclusion 123 Figure 7.6: Exa
Page 147 and 148:
Chapter 8 Conclusion and Outlook Co
Page 149 and 150:
8.1. Conclusion 127 arteries - info
Page 151:
8.2. Directions for Future Work 129
Page 154 and 155:
132 Chapter A. List of Publications
Page 157 and 158:
BIBLIOGRAPHY 135 Bibliography [1] A
Page 159 and 160:
BIBLIOGRAPHY 137 [21] Choyke, P., Y
Page 161 and 162:
BIBLIOGRAPHY 139 [41] Florin, C., P
Page 163 and 164:
BIBLIOGRAPHY 141 [63] Kitslaar, P.
Page 165 and 166:
BIBLIOGRAPHY 143 [85] Lindeberg, T.
Page 167 and 168:
BIBLIOGRAPHY 145 [108] O’Donnell,
Page 169 and 170:
BIBLIOGRAPHY 147 [129] Schmugge, S.
Page 171 and 172:
BIBLIOGRAPHY 149 [150] van Bemmel,
Page 173:
BIBLIOGRAPHY 151 [171] Yi, J. and R
show all

Segmentation of 3D Tubular Tree Structures in Medical Images ...

You also want an ePaper? Increase the reach of your titles

Delete template?

Save as template?