成果報告書詳細
管理番号20130000000198
タイトル*平成24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)」
公開日2014/6/14
報告書年度2012 - 2012
委託先名国立大学法人東京工業大学 学校法人慶應義塾 学校法人埼玉医科大学 日本電気株式会社
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
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-FY2014) FY2012 Annual Report
(Tokyo Institute of Technology, Keio University, Saitama Medical School, NEC Corporation)

1. Collecting the pathology information necessary to early cancer diagnosis:
We evaluated the result of discrimination analysis of hepatocellular carcinoma (HCC), and reexamined false negative and false positive cases. We assessed the algorithm and candidate features for raising diagnostic accuracy, in particular for well-differentiated HCC, which was difficult for pathologists to make a diagnosis.
Regarding the liver fibrosis assessment, the main image analysis and measurement module was almost finished, and we started the design of prototype system for demonstration.
2. Development of a diagnostic marker quantification method:
The expression of the known molecules was assessed by immunohistochemical staining used for DAB and immunofluorescent quantification digital slide (IQD), and the relation to clinical information was investigated. For DAB staining specimens of novel molecular markers, which were difficult to quantify, we attempted to quantify it by image processing.
3. Development of image recognition and quantification techniques:
For the quantification of architectural features of trabecular pattern, a new method for extracting the region of cell layers was developed. It was shown that the number and the irregularity of the cell layers were effective for discriminating well-differentiated HCC. Also the technique for HCC grading using multifractal features was further improved to generate the heat map of histological grades. Detail technical meetings were dedicated to share the newly developed techniques for creating the well-differentiated HCC detection module and the heat map of histological grades.
As conventional HCC histological grading does not reflect the clinical malignancy level, we challenged for creating new criteria of companion diagnosis showing clinical malignancy level. We analyzed the nuclei features statistically about the portal vein invasion and/or intrahepatic metastasis, early recurrent cases, etc. And more using molecular X, we continued the challenge for creating clinically useful criteria.
4. Accuracy enhancement in pathology imaging:
The accuracy of color correction was improved by using standard-stained mouse embryo tissue section in addition to color chart slide. Moreover, in the color correction for liver fibrosis assessment, a new method was developed to automatically select color samples from the tissue, and hence, all the steps for liver fibrosis assessment were automated.
We also developed a new method for image compression, in which the color space was modified depending on slides, yielding 15% improvement in the compression rate.
5. Algorithm evaluation:
In order to show the practical application of the system for histologically assessing liver fibrosis, evaluation of clinical validity was started. In our cases, the relationship between the histopathological classification and measured values was significant. We started collecting the cases whole slide image of liver biopsy specimens in other institutions.
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