Petrie Tracy, Samatham Ravikant, Webster Dan E, Leachman Sancy A
Oregon Health & Science University, Portland, USA.
AbbVie Inc., Chicago, USA.
Sci Data. 2025 Aug 2;12(1):1346. doi: 10.1038/s41597-025-05552-1.
We announce the release of the OHSU MoleMapper Smartphone Skin Images dataset which contains over six years of new data acquired from the Oregon Health & Science University's (OHSU) MoleMapper study. This released dataset includes 27,499 mole images curated to exclude images with protected health identifiers, 7,305 images of skin patches near the mole images, 1,000 contextual images, and basic metadata from the participants. This data is available to qualified researchers on Sage Bionetwork's Synapse platform under Synapse ID syn51520810 and represents the largest publicly available dataset of consumer-collected smartphone images of pigmented skin lesions. We release these data to the biomedical research community to enable quantitative analysis of images of non-clinician smartphone photography of skin lesions as well as to better understand what lesions appear concerning to the public. These data are unlabelled but in a machine learning context can be used to pre-train networks using self-supervised learning techniques or to quantitatively analyze the image quality of consumer-collected skin images.
我们宣布发布俄勒冈健康与科学大学（OHSU）痣映射器智能手机皮肤图像数据集，其中包含从俄勒冈健康与科学大学的痣映射器研究中获取的六年多的新数据。此次发布的数据集包括经过整理以排除带有受保护健康标识符的图像后的27499张痣图像、痣图像附近的7305张皮肤斑块图像、1000张背景图像以及参与者的基本元数据。这些数据可在Sage Bionetwork的Synapse平台上，通过Synapse ID syn51520810提供给合格的研究人员，它代表了最大的公开可用的消费者收集的色素沉着性皮肤病变智能手机图像数据集。我们将这些数据发布给生物医学研究界，以便对非临床医生智能手机拍摄的皮肤病变图像进行定量分析，并更好地了解公众认为哪些病变令人担忧。这些数据未标记，但在机器学习环境中，可用于使用自监督学习技术对网络进行预训练，或定量分析消费者收集的皮肤图像的质量。
