🧬 Leaderboard 3: 2D denoising (Nuclei)¶

👉 Download the training data here - Data on Zenodo

General Information¶

This leaderboard focuses on denoising 2D fluorescence microscopy images of cell nuclei, a common task in bioimage analysis.

• Guillaume Jacquemet. (2021). Noisy nuclei dataset for testing deep learning-based denoising tools. Zenodo. https://doi.org/10.5281/zenodo.5750174
• Image set BBBC006v1 from the Broad Bioimage Benchmark Collection [Ljosa et al., Nature Methods, 2012].

These include images acquired using widefield and spinning disk confocal microscopes, covering two different imaging styles and noise characteristics.

Training Dataset¶

The training dataset includes paired noisy and ground truth (GT) images and is split into two groups based on image size:

├── small_images
│   ├── noisy
│   └── gt
└── large_images
  ├── noisy
  └── gt

small_images/ contains X image pairs of shape (256, 256)

large_images/ contains Y image pairs of shape (1024, 1024)

In both folders:

• Noisy inputs are stored in the noisy/ subfolder
• Corresponding high-SNR targets are in the gt/ subfolder
• File names are aligned across folders (e.g., noisy/00001.tif↔ gt/00001.tif)

Evaluation¶

We’ll evaluate submissions on a private set of full-size nuclear images not included in the training set. These test images come from the same data sources.

Your goal is to build a model that can denoise across both widefield and spinning disk confocal images while keeping the nuclear structure intact.

👀 Check out the example submission here - Example submission on GitHub

✍️ Submit here - Submission page on GC