| 電子情報通信学会 研究会発表申込システム 講演論文 詳細 |
技報閲覧サービス [ログイン] 技報アーカイブ |
| トップに戻る | 前のページに戻る | [Japanese] / [English] |
| 講演抄録／キーワード | ||
| 講演名 | 2025-12-05 15:00 Residual Tumor Cellularity Assessment in Breast Cancer Using a Vision Transformer with Graph-Based Embedding -- Tumor Cellularity Assessment in Breast Cancer Using a Vision Transformer with Graph-Based Embedding -- ○Hossain Md Shakhawat（Kochi-Tech）・Md. Sazib Ahmed（AIM Lab）  MICT2025-48 MI2025-56 |
|
| 抄録 | （和） | Residual Tumor Cellularity (RTC) assessment is performed to quantify the amount of viable tumor remaining after a therapy, providing a critical indication of treatment effectiveness and patient prognosis. Traditionally, this evaluation is performed manually by pathologists through microscopic inspection and counting tumor cells. This process is time-consuming and subject to intra- and inter-observer variability, particularly when there is a worldwide shortage of pathologists. To address these limitations, we propose an automated framework for RTC classification into low, mid, and high categories. The method utilizes hematoxylin and eosin (H&E)-stained biopsy images. It employs a Vision Transformer (DeiT-B) to extract contextual image embeddings, which are further refined using a GraphSAGE-based neural network to model inter-patch relationships and spatial dependencies within the tissue architecture. Experimental results demonstrate that our approach achieves an accuracy of 93.1%, highlighting its potential as a robust tool to assist pathologists in automated RTC assessment. |
| （英） | Residual Tumor Cellularity (RTC) assessment is performed to quantify the amount of viable tumor remaining after a therapy, providing a critical indication of treatment effectiveness and patient prognosis. Traditionally, this evaluation is performed manually by pathologists through microscopic inspection and counting tumor cells. This process is time-consuming and subject to intra- and inter-observer variability, particularly when there is a worldwide shortage of pathologists. To address these limitations, we propose an automated framework for RTC classification into low, mid, and high categories. The method utilizes hematoxylin and eosin (H&E)-stained biopsy images. It employs a Vision Transformer (DeiT-B) to extract contextual image embeddings, which are further refined using a GraphSAGE-based neural network to model inter-patch relationships and spatial dependencies within the tissue architecture. Experimental results demonstrate that our approach achieves an accuracy of 93.1%, highlighting its potential as a robust tool to assist pathologists in automated RTC assessment. | |
| キーワード | （和） | Whole slide image / Digital pathology / Breast cancer / Tumor cellularity / / / / |
| （英） | Whole slide image / Digital pathology / Breast cancer / Tumor cellularity / / / / | |
| 文献情報 | 信学技報, vol. 125, no. 270, MI2025-56, pp. 66-69, 2025年12月. | |
| 資料番号 | MI2025-56 | |
| 発行日 | 2025-11-27 (MICT, MI) | |
| ISSN | Online edition: ISSN 2432-6380 | |
| 著作権に ついて |
技術研究報告に掲載された論文の著作権は電子情報通信学会に帰属します．(許諾番号：10GA0019/12GB0052/13GB0056/17GB0034/18GB0034) | |
| PDFダウンロード | MICT2025-48 MI2025-56 |
|
| 研究会情報 | |
| 研究会 | MICT MI |
| 開催期間 | 2025-12-04 - 2025-12-05 |
| 開催地（和） | 岡山コンベンションセンター |
| 開催地（英） | Okayama Convention Center |
| テーマ（和） | 医用画像工学一般，ヘルスケア・医療情報通信技術，一般 |
| テーマ（英） | Medical imaging technology, healthcare and medical information communication technology, etc. |
| 講演論文情報の詳細 | |
| 申込み研究会 | MI |
| 会議コード | 2025-12-MICT-MI |
| 本文の言語 | 英語 |
| タイトル（和） | |
| サブタイトル（和） | |
| タイトル（英） | Residual Tumor Cellularity Assessment in Breast Cancer Using a Vision Transformer with Graph-Based Embedding |
| サブタイトル（英） | Tumor Cellularity Assessment in Breast Cancer Using a Vision Transformer with Graph-Based Embedding |
| キーワード(1)（和/英） | Whole slide image / Whole slide image |
| キーワード(2)（和/英） | Digital pathology / Digital pathology |
| キーワード(3)（和/英） | Breast cancer / Breast cancer |
| キーワード(4)（和/英） | Tumor cellularity / Tumor cellularity |
| キーワード(5)（和/英） | / |
| キーワード(6)（和/英） | / |
| キーワード(7)（和/英） | / |
| キーワード(8)（和/英） | / |
| 第1著者 氏名（和/英/ヨミ） | Hossain Md Shakhawat / Hossain Md Shakhawat / |
| 第1著者 所属（和/英） | School of Informatics, Kochi University of Technology (略称： Kochi-Tech) School of Informatics, Kochi University of Technology (略称： Kochi-Tech) |
| 第2著者 氏名（和/英/ヨミ） | Md. Sazib Ahmed / Md. Sazib Ahmed / |
| 第2著者 所属（和/英） | AIM Lab, Kochi University of Technology (略称： AIM Lab) AIM Lab, Kochi University of Technology (略称： AIM Lab) |
| 講演者 | 第1著者 |
| 発表日時 | 2025-12-05 15:00:00 |
| 発表時間 | 20分 |
| 申込先研究会 | MI |
| 資料番号 | MICT2025-48, MI2025-56 |
| 巻番号（vol） | vol.125 |
| 号番号（no） | no.269(MICT), no.270(MI) |
| ページ範囲 | pp.66-69 |
| ページ数 | 4 |
| 発行日 | 2025-11-27 (MICT, MI) |
[研究会発表申込システムのトップページに戻る]