Mouse Cortex Axon Images with Segmentation Labels, 152 Samples

152 two-photon microscopy images of axons from the mouse somatosensory cortex, each sized 512x512 pixels. The dataset includes binary segmentation labels for the axons and is split into 132 training and 20 test images. It was compiled by researchers at Imperial College London from two published sources.

Use Cases

• Train image segmentation models based on the provided binary axon masks.
• Develop and test generative adversarial networks (GANs) for medical image synthesis based on the described image-label pairs.
• Benchmark computer vision algorithms for axon detection and analysis in two-photon microscopy images.
• Study axon morphology in the mouse somatosensory cortex based on the manually produced labels.

Strengths

• Includes 152 high-resolution (512x512) images with corresponding ground-truth segmentation labels.
• Provides a predefined train/test split (132 training, 20 test images) for model evaluation.
• Manually produced labels ensure a high-quality benchmark for segmentation tasks.

Limitations

• Dataset size is relatively small (152 images), which may limit training for data-hungry models.
• Column-level documentation is absent; field semantics must be inferred after download.
• Last update date is unknown; freshness unverified.

Provenance

Source

Combined data from Bass et al., 2017 and Canty et al., 2018 publications.

Collection Method

Images were collected using in-vivo two-photon microscopy and generated via max projection over 3D stacks.

Geography

Mouse somatosensory cortex.