Confocal Lacunar Canalicular Network Segmentation Data

Overview This dataset consists of confocal image stacks of the lacunar canalicular network (LCN) in cortical rat bone and their corresponding segmentations. For each sample, an image was acquired at two anatomical locations: the endosteum and in the mid-cortex of cortical bone. Segmentation was performed using a semantic segmentation model with a U-net architecture (Ronneberger et al., 2017), followed by post-processing correction. The resulting segmentations were used for downstream analysis to characterize the LCN's structure and morphology. Data Acquisition The samples were imaged using a Leica SP8 White light Laser Confocal microscope with a 63x objective and a photomultiplier tube detector (PMT) using a zoom of 1. Images were acquired at the Cell Imaging Core at the University of Utah Segmentation Labels Each voxel in the segmentation files is labeled according to the following schema: Label Structure 0 Background 1 Canaliculi 2 Lacunae 3 Canals (blood vessels) Image Resolution and Spacing Each confocal image stack is a 3D volume with anisotropic voxel spacing. Voxel dimensions are provided below to allow for accurate morphological analysis: In-plane resolution (X, Y): 0.18 µm × 0.18 µm Axial resolution (Z): 0.3 µm File Naming Convention The filenames in this dataset have the following format: [SampleID]-[Location]_[Suffix].tif Where SampleID: A unique identifier composed of a two-letter group code followed by a two-digit sample number (e.g., CT13). Location: The anatomical region where the image was acquired. endosteum - near the inner surface of the bone midcortex - in the middle of the cortical cross-section of the bone Suffix: img - for the raw confocal image seg - for the corresponding segmentation mask