WEKO3
アイテム
統計的形状モデルの構築法とその医用画像上の臓器領域のセグメンテーションへの応用に関する研究
https://doi.org/10.18997/00004077
https://doi.org/10.18997/0000407792048305-2e2b-45f7-9d0c-cd00bc94e4fe
| 名前 / ファイル | ライセンス | アクション |
|---|---|---|
D-212_kou_k_353.pdf (7.8 MB)
|
|
| Item type | 学位論文 = Thesis or Dissertation(1) | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| 公開日 | 2014-07-30 | |||||||||
| 資源タイプ | ||||||||||
| 資源タイプ識別子 | http://purl.org/coar/resource_type/c_db06 | |||||||||
| 資源タイプ | doctoral thesis | |||||||||
| タイトル | ||||||||||
| タイトル | Study on the Method of Constructing a Statistical Shape Model and Its Application to the Segmentation of Internal Organs in Medical Images | |||||||||
| 言語 | en | |||||||||
| タイトル | ||||||||||
| タイトル | 統計的形状モデルの構築法とその医用画像上の臓器領域のセグメンテーションへの応用に関する研究 | |||||||||
| 言語 | ja | |||||||||
| 言語 | ||||||||||
| 言語 | eng | |||||||||
| 著者 |
李, 光旭
× 李, 光旭
|
|||||||||
| 抄録 | ||||||||||
| 内容記述タイプ | Abstract | |||||||||
| 内容記述 | In image processing, segmentation is one of the critical tasks for diagnostic analysis and image interpretation. In the following thesis, we describe the investigation of three problems related to the segmentation algorithms for medical images: Active shape model algorithm, 3-dimensional (3-D) statistical shape model building and organic segmentation experiments. For the development of Active shape models, the constraints of statistical model reduced this algorithm to be difficult for various biological shapes. To overcome the coupling of parameters in the original algorithm, in this thesis, the genetic algorithm is introduced to relax the shape limitation. How to construct a robust and effective 3-D point model is still a key step in statistical shape models. Generally the shape information is obtained from manually segmented voxel data. In this thesis, a two-step procedure for generating these models was designed. After transformed the voxel data to triangular polygonal data, in the first step, attitudes of these interesting objects are aligned according their surface features. We propose to reflect the surface orientations by means of their Gauss maps. As well the Gauss maps are mapped to a complex plane using stereographic projection approach. The experiment was run to align a set of left lung models. The second step is identifying the positions of landmarks on polygonal surfaces. This is solved by surface parameterization method. We proposed two simplex methods to correspond the landmarks. A semi-automatic method attempts to “copy” the phasic positions of pre-placed landmarks to all the surfaces, which have been mapped to the same parameterization domain. Another automatic corresponding method attempts to place the landmarks equidistantly. Finally, the goodness experiments were performed to measure the difference to manually corresponded results. And we also compared the affection to correspondence when using different surface mapping methods. The third part of this thesis is applying the segmentation algorithms to solve clinical problems. We did not stick to the model-based methods but choose the suitable one or their complex according to the objects. In the experiment of lung regions segmentation which includes pulmonary nodules, we propose a complementary region growing method to deal with the unpredictable variation of image densities of lesion regions. In the experiments of liver regions, instead of using region growing method in 3-D style, we turn into a slice-by-slice style in order to reduce the overflows. The image intensity of cardiac regions is distinguishable from lung regions in CT image. But as to the adjacent zone of heart and liver boundary are generally blurry. We utilized a shape model guided method to refine the segmentation results.3-D segmentation techniques have been applied widely not only in medical imaging fields, but also in machine vision, computer graphic. At the last part of this thesis, we resume some interesting topics such as 3-D visualization for medical interpretation, human face recognition and object grasping robot etc. | |||||||||
| 目次 | ||||||||||
| 内容記述タイプ | TableOfContents | |||||||||
| 内容記述 | Chapter 1: Introduction||Chapter 2: Framework of Medical Image Segmentation||Chapter 3: 2-D Organic Regions Using Active Shape Model and Genetic Algorithm||Chapter 4: Alignment of 3-D Models||Chapter 5: Corespondence of 3-D Models||Chapter 6:Experiments of Organic Segmentation||Chapter 7: Visualization Technology and Its Applications||Chapter 8: Conclusions and Future Works | |||||||||
| 備考 | ||||||||||
| 内容記述タイプ | Other | |||||||||
| 内容記述 | 九州工業大学博士学位論文 学位記番号:工博甲第353号 学位授与年月日:平成25年9月27日 | |||||||||
| キーワード | ||||||||||
| 主題Scheme | Other | |||||||||
| 主題 | Point Distribution Model | |||||||||
| キーワード | ||||||||||
| 主題Scheme | Other | |||||||||
| 主題 | Statistical Shape Models | |||||||||
| キーワード | ||||||||||
| 主題Scheme | Other | |||||||||
| 主題 | segmentation of internal organs | |||||||||
| キーワード | ||||||||||
| 主題Scheme | Other | |||||||||
| 主題 | Computer Aided Diagnosis | |||||||||
| キーワード | ||||||||||
| 主題Scheme | Other | |||||||||
| 主題 | image segmentation | |||||||||
| キーワード | ||||||||||
| 主題Scheme | Other | |||||||||
| 主題 | medical image processiong | |||||||||
| アドバイザー | ||||||||||
| 金, 亨燮 | ||||||||||
| 学位授与番号 | ||||||||||
| 学位授与番号 | 甲第353号 | |||||||||
| 学位名 | ||||||||||
| 学位名 | 博士（工学） | |||||||||
| 学位授与年月日 | ||||||||||
| 学位授与年月日 | 2013-09-27 | |||||||||
| 学位授与機関 | ||||||||||
| 学位授与機関識別子Scheme | kakenhi | |||||||||
| 学位授与機関識別子 | 17104 | |||||||||
| 学位授与機関名 | 九州工業大学 | |||||||||
| 学位授与年度 | ||||||||||
| 内容記述タイプ | Other | |||||||||
| 内容記述 | 平成25年度 | |||||||||
| 出版タイプ | ||||||||||
| 出版タイプ | VoR | |||||||||
| 出版タイプResource | http://purl.org/coar/version/c_970fb48d4fbd8a85 | |||||||||
| アクセス権 | ||||||||||
| アクセス権 | open access | |||||||||
| アクセス権URI | http://purl.org/coar/access_right/c_abf2 | |||||||||
| ID登録 | ||||||||||
| ID登録 | 10.18997/00004077 | |||||||||
| ID登録タイプ | JaLC | |||||||||
