fidoptBioTISR wide-field to 3D-SIM, live COS-7 cells
The wide-field fluorescence volume of one live COS-7 cell region at low excitation, its first time point, with the provider's higher-excitation 3D-SIM reconstruction as target: F-actin, microtubules or mitochondria.
Read before comparing (6)
- Source. The provider's wide-field average at excitation level 1, linearly interpolated onto the SIM grid, which samples twice as finely in plane. It keeps its camera-count offset.
- Target. The higher-excitation reconstruction, not a human annotation, normalised by its maximum with its negative values kept. The score's dynamic range is one.
- Eligibility. prepare.py --survey finds a systematic wide-field/SIM displacement in 66 of 163 regions. Only the 97 regions displaced by under one SIM pixel are eligible; the corpus registers none of the others into eligibility.
- Mask. The smoothed maximum projection of the target, thresholded, dilated by one micrometre and repeated through the slices. It is given to every method.
- Spacing. In micrometres: 0.0306 in plane and 0.135 between slices on the target grid.
- Donors. Region identifiers define the split; the release gives no higher-level donor split.
No method has a result on every case of this dataset yet.
| Accuracy | Cost | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| # | Method | MAE | Change | PSNR µm | SSIM | Time s | RAM GB | GPU GB | Images |
| Source as given | 138.4 ±10.9 | -42.86 ±0.71 | 0.000 ±0.000 | ||||||
MAE: mean absolute error of a synthetic image against its target, inside the body, in the image's units. PSNR: the same by signal-to-noise ratio, higher is better. SSIM: the same by structural similarity, 1 is perfect.
Case by case
Click a score to look at that result: the target image, and over it the method's synthetic image. Underlined scores have their image; for the others you see the case without the result.
About the data
- Test cases
- Ten test regions and ten development regions drawn with seed 2026 by structure from the 97 eligible regions (prepare.py). Only each region's first time point is used; the later ones enter no split.
- Truth
- target image as released the provider 3D-SIM reconstruction of the same region and first time point, divided by its maximum value
- Source
- zenodo.org/records/13924484
zenodo.org/records/13843673
zenodo.org/records/13932988
zenodo.org/records/13843183 - Licence
- CC BY 4.0
10 cases for tuning, never tested
actin_031, mt_009, mito_057, mito_042, actin_043, mt_020, mito_003, mito_058, actin_001, mt_012