Peri-implantation mouse embryos

3D data volumes of developing mouse embryos. The dataset consists of:<br> 1. Cell nuclei data acquired with the confocal microscope. Voxel size: 0.207×0.207×1 µm or 0.23.23×0.23×1 µm<br> 2. Cell plasma membrane data acquired with the light-sheet microscope. Annotated nuclei volumes are stored in the 'Nuclei' directory. It contains two subdirectories 'train' and 'test', which were used for training a semantic segmentation network to predict the foreground pixels. There are 22 training and 13 validation stacks in the 'train' and 'test' directories respectively. Annotated plasma membrane volumes can be find inside the 'Membrane' directory. It contains two subdirectories 'train' and 'val', which were used for training a semantic segmentation network to predict the cell membranes. There are 5 training and 1 validation stacks in the 'train' and 'val' directories respectively. The data are saved as HDF5 files. Each HDF5 file contains the following datasets:<br> - `raw` - input image<br> - `label` - ground truth instance segmentation发育中小鼠胚胎的三维数据体。本数据集包含以下内容：<br>1. 共聚焦显微镜（confocal microscope）采集的细胞核数据，体素（voxel）尺寸为0.207×0.207×1 µm 或 0.23×0.23×1 µm；<br>2. 光片显微镜（light-sheet microscope）采集的细胞膜数据。已标注的细胞核体积存储于`Nuclei`（细胞核）目录下，该目录包含`train`（训练集）与`test`（测试集）两个子目录，用于训练语义分割网络（semantic segmentation network）以预测前景像素。其中`train`与`test`目录分别包含22个训练堆叠与13个验证堆叠。已标注的细胞膜体积可于`Membrane`（细胞膜）目录中找到，该目录包含`train`（训练集）与`val`（验证集）两个子目录，用于训练语义分割网络（semantic segmentation network）以预测细胞膜，其中`train`与`val`目录分别包含5个训练堆叠与1个验证堆叠。所有数据均以HDF5格式存储。每个HDF5文件包含以下数据集：<br>- `raw`：输入图像<br>- `label`：真值实例分割（instance segmentation）标签