SpineDL dataset: Spinal Cord Cross-Sections from C57BL/6J Mice

Training, validation, and test images of intact (Uninjured/Sham) and moderately contused thoracic spinal cord cross-sections of adult female C57BL/6J mice. Tissue sections are immunostained for NeuN (Rbfox3) and DAPI. This dataset is released in SpineDL paper and it is composed by two parts: Structure identification: semantic segmentation of spinal cord regions. Classes: Background (0), White matter (1), Ependyma (2), Gray matter (3), Lesioned / damaged tissue (4). Neuron identification: instance-level identification of neuronal somas. Each instance has a unique integer value. For more detailed information regarding the images, sectioning protocols, and individual animal data, please refer to the attached Excel file: “Image information.xlsx”. The original raw confocal images, along with more extensive documentation, are available on the Open Science Framework (OSF). Data type: Paired fluorescence microscopy and segmented mask images for spine structures and neurons Microscopy data type: 2D widefield images (fluorescence) and masks obtained via manual segmentation (see https://osf.io/n32z9 for details on segmentation). Microscope: Leica high-speed confocal systems (High-Speed Confocal Scanner and Leica TCS SP5) with 20X objective. Tissue type: female mouse thoracic spinal cord from injured and uninjured female C57BL6/ mice File format: .tif (8 and 16 bits) and .xml for neuronal labels in test images Image size: full spinal cord sections of variable size (up to 5000x5000 pixels). Pixel size: 1.12 to 2.25 µm Image preprocessing: Original z-stack confocal images were acquired with the LAS-AF 2.3 system and processed in Fiji. Processing included removing non-DAPI/NeuN channels, generating a maximal projection, assigning NeuN to red, and standardizing channel order Funding bodies: Council of Education, Culture and Sports of the Regional Government of Castilla-La Mancha (Spain), co-funded by the European Union (FEDER) Author(s): Pablo Ruiz-Amezcua (1,5), Daniel Franco-Barranco (2,3,4), David Reigada (1), Teresa Muñoz-Galdeano (1), Rodrigo Martinez Maza (1), Manuel Nieto-Díaz (1) Contact email: [e-mail] Affiliation: 1) Hospital Nacional de Parapléjicos (SESCAM). Toledo, Spain2) MRC Laboratory of Molecular Biology, Cambridge, UK3) Department of Physiology, Development and Neuroscience, University of Cambridge, Cambridge, UK4) Donostia International Physics Center (DIPC), San Sebastián, Spain5) Department of Experimental Biology, University of Jaén, Jaén, Spain Associated publications: In progress本数据集包含成年雌性C57BL/6J小鼠完整（未损伤/假手术）及中度挫伤胸椎脊髓横切面的训练、验证与测试图像。组织切片经NeuN（Rbfox3）与DAPI免疫染色。本数据集随SpineDL论文发布，分为两个部分： 结构识别任务：脊髓区域的语义分割。类别包括：背景（0）、白质（1）、室管膜（2）、灰质（3）、损伤/受损组织（4）。 神经元识别任务：神经元胞体的实例级识别，每个实例对应唯一整数值。 如需获取图像、切片方案及单只动物实验数据的详细信息，请参阅附件Excel文件《Image information.xlsx》。原始共聚焦显微图像及完整配套文档可在开放科学框架（Open Science Framework, OSF）平台获取。 数据类型：用于脊髓结构与神经元识别的配对荧光显微图像及分割掩码图像。 显微数据类型：2D宽场荧光图像，以及通过手动分割生成的掩码图像（分割详情请见https://osf.io/n32z9）。 显微镜系统：徕卡高速共聚焦系统（高速共聚焦扫描仪及徕卡TCS SP5），搭配20倍物镜。 组织样本来源：来自损伤与未损伤雌性C57BL/6J小鼠的胸椎脊髓组织。 文件格式：.tif（8位与16位），以及测试图像中神经元标注所用的.xml格式文件。 图像尺寸：尺寸可变的完整脊髓横切面（最大分辨率可达5000×5000像素）。像素尺寸：1.12至2.25 µm。 图像预处理：原始z-stack共聚焦图像通过LAS-AF 2.3系统采集，并在Fiji软件中完成后处理。处理流程包括移除非DAPI/NeuN通道、生成最大强度投影图、将NeuN通道映射为红色，以及标准化通道顺序。 资助方：西班牙卡斯蒂利亚-拉曼恰自治区教育、文化与体育部，由欧盟（欧洲区域发展基金，FEDER）联合资助。 作者：Pablo Ruiz-Amezcua(1,5)、Daniel Franco-Barranco(2,3,4)、David Reigada(1)、Teresa Muñoz-Galdeano(1)、Rodrigo Martinez Maza(1)、Manuel Nieto-Díaz(1) 联系邮箱：[e-mail] 所属机构： 1) 西班牙托莱多国家截瘫医院（SESCAM） 2) 英国剑桥MRC分子生物学实验室 3) 英国剑桥大学生理学、发育与神经科学系 4) 西班牙圣塞瓦斯蒂安国际物理中心（DIPC） 5) 西班牙哈恩大学实验生物学系 相关论文：待发表