Gland-Level Annotation Dataset

Gland-Level Annotation Dataset是由杨百翰大学物理与天文学系创建的高质量前列腺组织病理图像数据集，包含81张经过精心标注的全切片图像。该数据集由经过严格培训的预医学学生使用QuPath数字病理平台进行标注，每张图像均在腺体级别上进行了边界划定和Gleason模式及组织边界标签的分配，并由资深学生进行初步质量审查，最终由主要病理学家进行验证。此数据集旨在通过捕捉专家意见的多样性，训练能够预测Gleason模式分布而非单一真实标签的AI模型，从而提高前列腺癌检测和分级的自动化临床相关性和可推广性。

The Gland-Level Annotation Dataset is a high-quality prostate histopathological image dataset developed by the Department of Physics and Astronomy at Brigham Young University, comprising 81 meticulously annotated whole-slide images. This dataset was annotated by rigorously trained premedical students using the QuPath digital pathology platform. For each image, boundary delineation, Gleason pattern assignment, and tissue boundary labeling were performed at the glandular level. The annotations underwent preliminary quality reviews by senior students, followed by final validation by a lead pathologist. This dataset is designed to train AI models capable of predicting the distribution of Gleason patterns rather than a single ground-truth label by capturing the diversity of expert annotations, thereby enhancing the clinical relevance and generalizability of automated prostate cancer detection and grading workflows.