| 講演名 | 2020-05-29 [Poster Presentation] Tumor detection from colonoscopy Whole Slice Images By Deep Learning Cherubin Mugisha(School of Computer Science and Engineering), Incheon Paik(School of Computer Science and Engineering), |
|---|---|
| PDFダウンロードページ |  PDFダウンロードページへ |
| 抄録(和) | Image semantic segmentation is a technique of segregating an image into many parts. The goal of this research was to use High-resolution whole slice images from colonoscopy exams and identify tumors and trace its borders with the highest precision of an expert. We modified the U-net architecture that we used to train high-resolution images and test our results. Having enough data is not easy in biomedical imaging, so we had to work on data augmentation to get data from 250 images up to 3200. As a result, the accuracy was 92.49% with a mean of Intersection over Union (IoU) of 0.44. this result shows that AI and Image processing can help to automate tumor assessment by coupling models with devices that produce those images as a service. |
| 抄録(英) | Image semantic segmentation is a technique of segregating an image into many parts. The goal of this research was to use High-resolution whole slice images from colonoscopy exams and identify tumors and trace its borders with the highest precision of an expert. We modified the U-net architecture that we used to train high-resolution images and test our results. Having enough data is not easy in biomedical imaging, so we had to work on data augmentation to get data from 250 images up to 3200. As a result, the accuracy was 92.49% with a mean of Intersection over Union (IoU) of 0.44. this result shows that AI and Image processing can help to automate tumor assessment by coupling models with devices that produce those images as a service. |
| キーワード(和) | Deep Learning / Biomedical Image / WSI / U-net / Tumor |
| キーワード(英) | Deep Learning / Biomedical Image / WSI / U-net / Tumor |
| 資料番号 | |
| 発行日 |
| 研究会情報 | |
| 研究会 | SC |
|---|---|
| 開催期間 | 2020/5/29(から1日開催) |
| 開催地（和） | オンライン/会津大 |
| 開催地（英） | Online/Univ of Aizu |
| テーマ（和） | サービスコンピューティング環境に向けたAIとその応用,その他 |
| テーマ（英） | AI Application for Service Computing Environment and Other Issues |
| 委員長氏名（和） | 中村 匡秀(神戸大) |
| 委員長氏名（英） | Masahide Nakamura(Kobe Univ.) |
| 副委員長氏名（和） | 菊地 伸治(物質・材料研究機構) / 山登 庸次(NTT) |
| 副委員長氏名（英） | Shinji Kikuchi(NIMS) / Yoji Yamato(NTT) |
| 幹事氏名（和） | 細野 繁(東京工科大) / 木村 功作(富士通研) |
| 幹事氏名（英） | Shigeru Hosono(Tokyo Univ. of Tech.) / Kosaku Kimura(Fujitsu Lab.) |
| 幹事補佐氏名（和） | |
| 幹事補佐氏名（英） | |
| 講演論文情報詳細 | |
| 申込み研究会 | Technical Committee on Service Computing |
|---|---|
| 本文の言語 | ENG |
| タイトル（和） | |
| サブタイトル（和） | |
| タイトル（英） | [Poster Presentation] Tumor detection from colonoscopy Whole Slice Images By Deep Learning |
| サブタイトル（和） | |
| キーワード(1)（和/英） | Deep Learning / Deep Learning |
| キーワード(2)（和/英） | Biomedical Image / Biomedical Image |
| キーワード(3)（和/英） | WSI / WSI |
| キーワード(4)（和/英） | U-net / U-net |
| キーワード(5)（和/英） | Tumor / Tumor |
| 第 1 著者 氏名（和/英） | Cherubin Mugisha / Cherubin Mugisha |
| 第 1 著者 所属（和/英） | The University of Aizu(略称：School of Computer Science and Engineering) The University of Aizu(略称：School of Computer Science and Engineering) |
| 第 2 著者 氏名（和/英） | Incheon Paik / Incheon Paik |
| 第 2 著者 所属（和/英） | The University of Aizu(略称：School of Computer Science and Engineering) The University of Aizu(略称：School of Computer Science and Engineering) |
| 発表年月日 | 2020-05-29 |
| 資料番号 | |
| 巻番号（vol） | vol.120 |
| 号番号（no） | SC-49 |
| ページ範囲 | pp.-(), |
| ページ数 | |
| 発行日 | |