WEKO3
アイテム
An application of cascaded 3D fully convolutional networks for medical image segmentation
http://hdl.handle.net/2237/00028579
http://hdl.handle.net/2237/00028579908f5089-907c-426b-8b3a-3c140d274e38
| 名前 / ファイル | ライセンス | アクション |
|---|---|---|
cmig2018_rothhr_cascaded_fcn_final (10.0 MB)
|
|
| Item type | 学術雑誌論文 / Journal Article(1) | |||||
|---|---|---|---|---|---|---|
| 公開日 | 2018-09-18 | |||||
| タイトル | ||||||
| タイトル | An application of cascaded 3D fully convolutional networks for medical image segmentation | |||||
| 言語 | en | |||||
| 著者 |
Roth, Holger R.
× Roth, Holger R.× Oda, Hirohisa× Zhou, Xiangrong× Shimizu, Natsuki× Yang, Ying× Hayashi, Yuichiro× Oda, Masahiro× Fujiwara, Michitaka× Misawa, Kazunari× Mori, Kensaku |
|||||
| 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 | |||||
| 権利 | ||||||
| 言語 | en | |||||
| 権利情報 | © 2018. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/ | |||||
| キーワード | ||||||
| 主題Scheme | Other | |||||
| 主題 | Fully convolutional networks | |||||
| キーワード | ||||||
| 主題Scheme | Other | |||||
| 主題 | Deep learning | |||||
| キーワード | ||||||
| 主題Scheme | Other | |||||
| 主題 | Medical imaging | |||||
| キーワード | ||||||
| 主題Scheme | Other | |||||
| 主題 | Computed tomography | |||||
| キーワード | ||||||
| 主題Scheme | Other | |||||
| 主題 | Multi-organ segmentation | |||||
| 抄録 | ||||||
| 内容記述タイプ | Abstract | |||||
| 内容記述 | Recent advances in 3D fully convolutional networks (FCN) have made it feasible to produce dense voxel-wise predictions of volumetric images. In this work, we show that a multi-class 3D FCN trained on manually labeled CT scans of several anatomical structures (ranging from the large organs to thin vessels) can achieve competitive segmentation results, while avoiding the need for handcrafting features or training class-specific models. To this end, we propose a two-stage, coarse-to-fine approach that will first use a 3D FCN to roughly define a candidate region, which will then be used as input to a second 3D FCN. This reduces the number of voxels the second FCN has to classify to ∼10% and allows it to focus on more detailed segmentation of the organs and vessels. We utilize training and validation sets consisting of 331 clinical CT images and test our models on a completely unseen data collection acquired at a different hospital that includes 150 CT scans, targeting three anatomical organs (liver, spleen, and pancreas). In challenging organs such as the pancreas, our cascaded approach improves the mean Dice score from 68.5 to 82.2%, achieving the highest reported average score on this dataset. We compare with a 2D FCN method on a separate dataset of 240 CT scans with 18 classes and achieve a significantly higher performance in small organs and vessels. Furthermore, we explore fine-tuning our models to different datasets. Our experiments illustrate the promise and robustness of current 3D FCN based semantic segmentation of medical images, achieving state-of-the-art results. | |||||
| 言語 | en | |||||
| 内容記述 | ||||||
| 内容記述タイプ | Other | |||||
| 内容記述 | ファイル公開：2019-06-01 | |||||
| 言語 | ja | |||||
| 出版者 | ||||||
| 出版者 | Elsevier | |||||
| 言語 | en | |||||
| 言語 | ||||||
| 言語 | eng | |||||
| 資源タイプ | ||||||
| 資源タイプ識別子 | http://purl.org/coar/resource_type/c_6501 | |||||
| 資源タイプ | journal article | |||||
| 出版タイプ | ||||||
| 出版タイプ | AM | |||||
| 出版タイプResource | http://purl.org/coar/version/c_ab4af688f83e57aa | |||||
| DOI | ||||||
| 関連タイプ | isVersionOf | |||||
| 識別子タイプ | DOI | |||||
| 関連識別子 | https://doi.org/10.1016/j.compmedimag.2018.03.001 | |||||
| ISSN（print） | ||||||
| 収録物識別子タイプ | PISSN | |||||
| 収録物識別子 | 0895-6111 | |||||
| 書誌情報 |
en : Computerized Medical Imaging and Graphics 巻 66, p. 90-99, 発行日 2018-06 |
|||||
| 著者版フラグ | ||||||
| 値 | author | |||||
