Tissue Images Dataset (tissue_images_dataset)

Overview

This DeepTrackAI repository provides a copy of the Segmented anisotropic ssTEM dataset of neural tissue,
a dataset of serial section Transmission Electron Microscopy (ssTEM) prepared for training and evaluating image segmentation methods.

The content mirrors the public dataset by Gerhard et al., available from both the Original GitHub repository and figshare (DOI: 10.6084/m9.figshare.856713).

The dataset contains 2 stacks of ssTEM images of the Drosophila melanogaster third instar larva ventral nerve cord. Each stack consists of 20 serial sections. The imaged volume measures approximately 4.7 × 4.7 × 1 μm³, with a voxel resolution of 4.6 × 4.6 × 45–50 nm³.

Summary

• Number of stacks: 2 (1 training, 1 test)
• Sections per stack: 20 (40 total)
• Image size: 1024 × 1024 pixels
• Stacks:
  • stack1/ — labeled stack (training)
  • stack2/ — raw stack (test)
• Format: 8-bit grayscale (raw images)
• Labels: For each class, binary images; for multi-class segmentation, 8-bit images with pixel values encoding structure type (see table below).

---

Original Source

• Title: Segmented anisotropic ssTEM dataset of neural tissue
• Authors: Stephan Gerhard, Jan Funke, Julien Martel, Albert Cardona, Richard Fetter
• Source: Original GitHub repository and figshare (DOI: 10.6084/m9.figshare.856713)
• License: Creative Commons Attribution-NonCommercial-ShareAlike 3.0 Unported

If you use this dataset in your research, please follow the licensing terms of the original source and provide proper attribution.

---

Dataset Structure

/tissue_images_dataset
├── stack1/                     # Ground-truth stack (training)
│   ├── raw/                    # Raw 8-bit grayscale ssTEM images
│   ├── membranes/              # Binary membrane segmentations 
│   ├── mitochondria/           # Binary mitochondria segmentations
│   ├── synapses/               # Binary synapse segmentations
│   └── labels/                 # Combined / multi-class labels
└── stack2/                     # Test stack 
    └── raw/                    # Raw 8-bit grayscale ssTEM images

---

Label Encoding

In the labels/ directory, each pixel is assigned one of the following values:

Value	Structure
0	Membrane (0°)
32	Membrane (45°)
64	Membrane (90°)
96	Membrane (135°)
128	Membrane junction
159	Glia / Extracellular
191	Mitochondria
223	Synapse
255	Intracellular

---

How to Access the Data

Clone the Repository

git clone https://github.com/DeepTrackAI/tissue_images_dataset
cd tissue_images_dataset

---

Attribution

This replication dataset is based on the original Segmented anisotropic ssTEM dataset of neural tissue dataset. When using this replication, please cite the original dataset.

Cite the dataset:

Gerhard S, Funke J, Martel J, Cardona A, Fetter R. Segmented anisotropic ssTEM dataset of neural tissue. figshare. DOI: 10.6084/m9.figshare.856713

@misc{gerhard2013sstem,
  title        = {Segmented anisotropic ssTEM dataset of neural tissue},
  author       = {Gerhard, Stephan and Funke, Jan and Martel, Julien and Cardona, Albert and Fetter, Richard},
  howpublished = {figshare},
  year         = {2013},
  doi          = {10.6084/m9.figshare.856713}
}

---

License

This replication dataset is shared under the Creative Commons Attribution-NonCommercial-ShareAlike 3.0 Unported License, following the original licensing terms.