CLEM-Reg, Airyscan fluorescence to volume electron microscopy
One cell specimen imaged twice, an Airyscan fluorescence volume (moving) and a volume electron microscopy (fixed), with manually placed BigWarp correspondences.
Read before comparing (5)
- Images. The moving image is the CZI's MitoTracker channel, identified from its metadata. The fixed image is the author-aligned, cropped and downscaled EM TIFF that matches the released landmarks, not a composite holding warped fluorescence. Both keep their native intensities, voxels and grids.
- Units. Calibration stays in micrometres, including non-rounded TIFF resolution and slice spacing. The landmarks are in ImageJ image coordinates starting at zero, as BigWarp uses them, not CZI tile offsets: S-BSST1075 starts at storage tile (12,12), recorded in provenance.
- Alignment. Each pair gets one start from the intensities alone, an elastix translation (elastix-translation) on mutual information inside each image's body, written in the moving image's header and points. Every active landmark pair is scored.
- Baseline. With no registration the points lie 0.56 µm apart on S-BSST707 (145 points) and 9.10 µm on S-BSST1075 (42 points), where the cell's fluorescence lies beside its EM outline. The difference in scale and shape between the two volumes is left to the methods.
- Caveats. Only two cellular specimens, author-selected reference landmarks and mixed or unspecified source licences: keep results apart.
No method has a result on every pair of this dataset yet.
| Accuracy | Smoothness | Cost | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| # | Method | TRE µm | Change | rTRE | SDlogJ | Folds % | Time s | RAM GB | GPU GB | Images |
| No registration | 4.83 ±54.27 | 0.0663 ±0.6897 | 0.000 | 0.00 | ||||||
TRE: distance between matching landmarks. rTRE: the TRE over the length of the image's diagonal, to compare datasets of other sizes and units. SDlogJ: how irregular the deformation is, 0 is perfectly smooth. Folds: share of the image where the deformation folds over itself.
Pair by pair
Click a score to look at that result: the fixed image, and over it the moving image after registration, with the annotations. Underlined scores have their image; for the others you see the pair without the result. A green cell gained against no registration on that pair, a red one lost.
About the data
- Test pairs
- The two released specimens are the test cases. No independent development pair is available: a shared cell culture and donor do not make another pair independent. S-BSST1175/EMPIAR-11666 is excluded pending coordinate-frame clarification: its active moving landmarks reach y = 68.59 µm, outside the CZI's 63.50 µm height, and no axis swap, rotation, rescaling or point deletion is inferred to force agreement.
- Truth
- landmarks by experts 145 and 42 active source correspondences for S-BSST707 and S-BSST1075, all scored. Inactive source rows are excluded.
- Source
- www.ebi.ac.uk/biostudies/bioimages/studies/S-BSST707
www.ebi.ac.uk/biostudies/bioimages/studies/S-BSST1075
www.ebi.ac.uk/empiar/EMPIAR-10819
www.ebi.ac.uk/empiar/EMPIAR-11537
doi.org/10.1038/s41592-025-02794-0 - Licence
- EMPIAR volumes are CC0; no explicit reusable image licence was identified for these two BioStudies light-microscopy records. Local evaluation does not establish redistribution permission.