苏木精和伊红（H＆E）染色载玻片的图像数据集

腺体是重要的组织学结构，其作为分泌蛋白质和碳水化合物的主要机制存在于大多数器官系统中。已经表明，由腺上皮引起的恶性肿瘤，也称为腺癌，是最常见的癌症形式。病理学家常规使用腺体的形态来评估几种腺癌的恶性程度，包括前列腺癌，乳腺癌，肺癌和结肠癌。 准确分割腺体通常是获得可靠形态统计数据的关键步骤。尽管如此，由于不同组织学分级的腺体形态的巨大变化，任务本质上是非常具有挑战性的。到目前为止，大多数研究都集中在健康或良性样本中的腺体分割，但很少用于中等或高等级癌症，并且通常，它们针对特定数据集进行了优化。 在这一挑战中，鼓励参与者在苏木精和伊红（H＆E）染色载玻片的图像上运行腺体分割算法，这些载玻片由多种组织学分级组成。数据集与专家病理学家提供的地面实况注释一起提供。要求参与者在提供的训练数据集上开发和优化他们的算法，并在测试数据集上验证他们的算法。

Glands are important histological structures that exist in most organ systems as the primary mechanism for secreting proteins and carbohydrates. Malignant tumors arising from glandular epithelium, also known as adenocarcinomas, have been shown to be the most common form of cancer. Pathologists routinely use glandular morphology to assess the malignancy of several adenocarcinoma types, including prostate, breast, lung, and colon cancers. Accurate gland segmentation is often a critical step in obtaining reliable morphological statistical data. Nevertheless, the task is inherently highly challenging due to the vast variations in glandular morphology across different histological grades. To date, most studies have focused on gland segmentation in healthy or benign samples, with few targeting moderate or high-grade cancers, and they are typically optimized for specific datasets. In this challenge, participants are encouraged to run gland segmentation algorithms on images of hematoxylin and eosin (H&E)-stained slides, which encompass multiple histological grades. The dataset is provided alongside ground truth annotations provided by expert pathologists. Participants are required to develop and optimize their algorithms on the provided training dataset, and validate their algorithms on the test dataset.

搜集汇总

背景与挑战

背景概述

该数据集包含苏木精和伊红（H＆E）染色载玻片的图像，专注于腺体分割任务，用于评估腺癌的恶性程度。数据集提供了多种组织学分级的图像和专家注释，旨在支持腺体分割算法的开发和优化。