Challenge Results¶
Thank you to everyone who participated in the AI4Life Calcium Imaging Denoising Challenge 2025 (AI4Life-CIDC25).
This challenge focused on unsupervised denoising of calcium imaging microscopy data. Calcium imaging data are spatially and temporally structured, so denoising methods need to preserve both image content and temporal activity dynamics.
The results below summarise the best published submissions for each leaderboard. The official ranking score is stSNR, the spatio-temporal signal-to-noise ratio, averaged across the files in each leaderboard dataset.
Only the best published result for each participant is listed, showing the top-3 scoring algorithms for each leaderboard.
Final Submission Phase: Content Generalisation¶
| Rank | Name | Affiliation | Algorithm | Result stSNR | Link to code |
|---|---|---|---|---|---|
| 1 | edoardogiacomello | Human Technopole | N2V 3D | 22.14 | GitHub |
| 2 | chhayanshporwal | Rajasthan Technical University Kota | Noise2Void 3D | 21.18 | GitHub |
| 3 | aagamsheth | Ahmedabad University | 3D N2V U Net DVT Inspired | 20.85 | GitHub |
Final Submission Phase: Noise Level Generalisation¶
| Rank | Name | Affiliation | Algorithm | Result stSNR | Link to code |
|---|---|---|---|---|---|
| 1 | edoardogiacomello | Human Technopole | N2V 3D | 16.75 | GitHub |
| 2 | chhayanshporwal | Rajasthan Technical University Kota | Noise2Void 3D | 15.59 | GitHub |
| 3 | mcroft | Human Technopole | Noise2Noise 2D Windowed | 15.10 |