成果報告書詳細
管理番号20110000001121
タイトル*平成22年度中間年報 「がん超早期診断・治療機器の総合研究開発/超早期高精度診断システムの研究開発:病理画像等認識技術の研究開発 病理画像等認識基礎技術の研究開発(定量的病理診断を可能とする病理画像認識技術)」(埼玉医科大学)
公開日2011/11/23
報告書年度2010 - 2010
委託先名埼玉医科大学
プロジェクト番号P10003
部署名バイオテクノロジー・医療技術部
和文要約和文要約等以下本編抜粋:1. 研究開発の内容及び成果等 (1) 画像認識・数量化技術の研究開発: 多次元学習を用いた画像認識・数量化の研究開発 肝臓癌などの病理組織診断において病理画像を用いた支援装置の開発が求められている.本プロジェクトは,がん組織の定量的解析可能な「定量的病理診断」技術も確立を目的としている.初期病変の病理診断には,僅かな異型度の評価,少量の病変での評価などが必要なため,複雑な特徴を有する組織画像全体を画一的に扱うのは困難である.そこで,病理画像全体を特徴ごとに分割した後で各領域に適した解析を行い,数量化することが有効であると考える.本年度は,1.癌による染色性の違いによる癌の可能性が高い領域の判別可能性の検討,2.多次元情報から特徴を抽出するための学習アルゴリズムとして自己組織化マップの設計と基本プログラムの開発を行った.
英文要約Title: Research and Development Project for pathological image recognition technology, Research and Development Project to develop basic technologies for recognizing pathology images, Pathological image analysis technology to enable a quantitative pathological diagnosis, (FY2010-FY2012) FY2010 Annual Report
(1) Research and Development of Image Recognition and Quantification Technology for Pathological Images: The development of the computer-aided system for a pathological image-reading is needed in order to support pathologist. The goal of our project is to establish the quantification method for liver cancer cells. Liver cancer cells have many features in multiple image resolution levels. Thus the area segmentation preprocessing is effective to analyze images in an appropriate image resolution for each part. In this study, segmentation method of cancer area by using difference of stainability due to cancer is proposed and tested with one case. The results show that rough segmentation of H&E stained pathological image is possible with further investigations by considering different staining conditions. Also we developed a clustering program based on a self-organizing map to extract the features from multi-dimensional information such as stainability, ballooning and fibrosis etc. (2) Development of High Quality Image Compression Technology for Pathological Images: Image compression is necessary for pathological images because the raw images of 1G bytes are too large to storage and to analyze. As pathological image compression needs to satisfy the requirements for handling multiple resolution images and multi-spectrum images, we propose to develop modified JPEG2000. In this study, we clarify the new requirements of the image compression for analysis of pathological image of liver cancer. Scalable image compression technology should be introduces to satisfy to analyze multi features of images that depend on the objects such as nuclei, cytoplasm and cell construction. To evaluate proposed image compression method, we develop the modified JPEG2000 software that is able to handle multi-spectrum images and high quantized images.
ダウンロード成果報告書データベース(ユーザ登録必須)から、ダウンロードしてください。

▲トップに戻る