APOL7C_3D: A High-Resolution 3D Electron Microscopy Dataset for Automated Multiclass Segmentation

APOL7C_3D: Automating Multiclass Segmentation of 3D EM Images Overview The APOL7C_3D dataset is designed for training and evaluating deep learning models for automated multiclass segmentation of 3D electron microscopy (EM) images. This dataset is curated for DANDIset, providing high-resolution volumetric EM data with corresponding segmentation masks for various subcellular structures. Original Dataset Specifications Dimensions: 3088 × 2156 × 500 (XYZ) Channels: Channel 1 (Mask): Annotated segmentation mask with five classes (0–4). Channel 2 (Image): High-resolution grayscale EM image (256 intensity levels). Segmentation Classes: Class 0: Background Class 1: Phagosome Membrane Class 2: Phagosome Class 3: Mitochondria Class 4: Endoplasmic Reticulum Annotated Region (Training Subset) Cropped Region: 1052 × 924 × 50 Classes Present: 0 (Background), 1 (Phagosome Membrane), 2 (Phagosome), 3 (Mitochondria), 4 (Endoplasmic Reticulum) Purpose: Provides a smaller, curated training volume optimized for deep learning model development. Applications This dataset facilitates the development of deep learning-based segmentation models for biomedical image analysis, particularly for understanding subcellular structures in electron microscopy data. It is useful for biomedical AI research, neuroscience, cell biology, and computational microscopy.