CR-AI4SkIN dataset

CR-AI4SkIN is a public dataset composed of H&E patches extracted from Whole Slide Images. This dataset contains Cutaneous Spindle Cell neoplasms, and the task is to classify WSIs into benign or malignant. The main feature of this dataset is that it has been labeled by ten non-expert annotators. This repository contains the associated files to replicate the study titled "Annotation Protocol and Crowdsourcing Multiple Instance Learning Classification of Skin Histological Images: The CR-AI4SkIN Dataset", published in the Artificial Intelligence in Medicine journal. For further details on the study and the dataset, please see the published article. The zipped file 'annotations.zip' contains three .csv files with crowdsourcing annotations. These files provide the train/val/test split as well as label information. The 'GT' column stands for the expert label, 'MV' for the majority vote among non-experts, and 'Marker_X' is the label given by the X-th annotator. The zipped file 'img.zip' contains the dataset images divided into two sub-directories indicating the source hospital. Each image is associated with its own folder, which is identified using anonymized IDs. Within each folder, we include the extracted patches representing the predicted regions of interest. Note: We did not include 1 image in the training set and another in the test set because the regions of interest were not found/had troubles. Label Dictionary: 0: Benign 1: Malignant -1: Missing value (e.g., if an annotator did not label that image). Hospital Dictionary: HCUV: Hospital Clínico Universitario de Valencia (Valencia Hospital).HUSC: Hospital Universitario San Cecilio (Granada Hospital). Citation: If you use this dataset, please cite the following article: @article{del2023annotation, title={Annotation protocol and crowdsourcing multiple instance learning classification of skin histological images: The CR-AI4SkIN dataset}, author={Del Amor, Rocío and Pérez-Cano, Jose and López-Pérez, Miguel and Terradez, Liria and Aneiros-Fernandez, Jose and Morales, Sandra and Mateos, Javier and Molina, Rafael and Naranjo, Valery}, journal={Artificial Intelligence in Medicine}, volume={145}, pages={102686}, year={2023}, publisher={Elsevier} } Funding: This work has received funding from the Spanish Ministry of Economy and Competitiveness through project PID2019-105142RB (AI4SKIN) and Spanish Ministry of Science and Innovation through project PID2022-140189OB, from Horizon 2020, the European Union’s Framework Programme for Research and Innovation, under the grant agreement No. 860627 (CLARIFY), grant B-TIC-324-UGR20 funded by Consejería de Universidad, Investigación e Innovación (Junta de Andalucía) and by “ERDF A way of making Europe”, and GVA through the project INNEST/2021/321 (SAMUEL). The work of Rocío del Amor has been supported by the Spanish Ministry of Universities (FPU20/05263). The work of Miguel López Pérez has been supported by the University of Granada postdoctoral program “Contrato Puente”. The work of Sandra Morales has been co-funded by the Universitat Politècnica de València through the program PAID-10-20.CR-AI4SkIN是一款公开数据集，其数据为从全视野数字病理切片（Whole Slide Images, WSI）中提取的苏木精-伊红（H&E）染色图像块。该数据集涵盖皮肤梭形细胞肿瘤，核心任务为将全视野数字病理切片划分为良性与恶性两类。本数据集的核心特色在于，其标注由十位非专业标注人员完成。 本仓库包含复现《皮肤组织学图像的标注协议与众包多实例学习分类：CR-AI4SkIN数据集》研究所需的配套文件，该研究已发表于《医学人工智能（Artificial Intelligence in Medicine）》期刊。如需了解该研究与数据集的更多细节，请查阅已发表的学术论文。 压缩文件「annotations.zip」内含3个CSV格式文件，存储众包标注信息。这些文件同时提供了训练集、验证集与测试集的划分方案及标签信息：其中「GT」列代表专家标注标签，「MV」列代表非专业标注者的多数投票结果，「Marker_X」则代表第X位标注者给出的标注标签。 压缩文件「img.zip」内含数据集图像，其按来源医院分为两个子目录。每张图像对应一个以匿名ID命名的文件夹，文件夹内包含提取的、代表预测感兴趣区域的图像块。注意：由于未找到感兴趣区域或存在相关问题，训练集与测试集各有1张图像未被纳入本数据集。 标签对照表： 0：良性；1：恶性；-1：缺失值（例如标注者未对该图像进行标注的情况）。 医院对照表： HCUV：瓦伦西亚大学临床医院（Hospital Clínico Universitario de Valencia）；HUSC：圣塞西利奥大学医院（Hospital Universitario San Cecilio，即格拉纳达医院）。 引用规范： 若您使用本数据集，请引用以下学术论文： @article{del2023annotation, title={Annotation protocol and crowdsourcing multiple instance learning classification of skin histological images: The CR-AI4SkIN dataset}, author={Del Amor, Rocío and Pérez-Cano, Jose and López-Pérez, Miguel and Terradez, Liria and Aneiros-Fernandez, Jose and Morales, Sandra and Mateos, Javier and Molina, Rafael and Naranjo, Valery}, journal={Artificial Intelligence in Medicine}, volume={145}, pages={102686}, year={2023}, publisher={Elsevier} } 资助信息： 本研究获得西班牙经济与竞争力部通过项目PID2019-105142RB（AI4SKIN）、西班牙科学与创新部通过项目PID2022-140189OB的资助；同时获得欧盟框架计划研究与创新项目Horizon 2020下资助协议号860627（CLARIFY）的支持、安达卢西亚自治区大学、研究与创新委员会资助的项目B-TIC-324-UGR20（该项目同时获得“欧洲区域发展基金：打造欧洲之路”的配套支持），以及巴伦西亚自治区通过项目INNEST/2021/321（SAMUEL）提供的资助。 罗西奥·德尔·阿莫尔的研究工作得到西班牙大学部FPU20/05263项目的支持。米格尔·洛佩斯·佩雷斯的研究工作得到格拉纳达大学“Contrato Puente”博士后项目的支持。桑德拉·莫拉莱斯的研究工作由瓦伦西亚理工大学PAID-10-20项目联合资助。