Algorithm of extracting nuclear and membrane for diagnosing early breast cancer

DOI Web Site

Bibliographic Information

Other Title
  • 早期乳癌診断のための核及び膜抽出アルゴリズム
  • ソウキ ニュウガン シンダン ノ タメ ノ カク オヨビ マク チュウシュツ アルゴリズム

Search this article

Abstract

This paper deals with an algorithm of extracting nuclear and membrane for diagnosis using breast cancer images. It is important to extract nuclei and membranes exactly one by one from cell images, in order to classify the subtype by clinicopathological indexes. Breast cancer is classified into four disease subtypes by expression of three proteins (ER, PgR, HER2). Since ER and PgR develop in nuclei and HER2 develops in membranes, the development is important factor for diagnosis. At present, the proteinic expressional level of specimen materials is visually judged by cytotechnologists. Therefore, we propose the methods of extracting nucleus by applying contour tracking and using membrane, and membrane by top-hat filter and canny filter. We use H-stained images, DAB-stained images and multiple immunostaining cell images. By our proposed methods, we can extract nuclei about 80% and membranes 60% both of two images.

Journal

Keywords

Details 詳細情報について

Report a problem

Back to top