InstanSeg brightfield_nuclei v0.1.1 (mirror)
Mirror of the InstanSeg brightfield_nuclei model (v0.1.1, bioimage.io package) for use by Technoplasm Vision's IHC quantification. The weights are unchanged from the upstream release: https://github.com/instanseg/instanseg/releases/tag/instanseg_models_v0.1.1.
instanseg.pt— TorchScript model. Input: RGB float32(1, 3, H, W)in 0–255 at 0.5 µm/px, normalised per channel to the 0.1–99.9 percentile range. Output: instance label map(1, 1, H, W)(0 = background).rdf.yaml— bioimage.io model description (pre/post-processing, axes, test tensors).LICENSE— Apache License 2.0 (InstanSeg).
License
The InstanSeg method, code and this model are shared under the Apache License 2.0.
The model was trained on the following datasets. Users are responsible for using the model in accordance with the licenses of the source datasets (all permit commercial use with attribution):
| dataset | license | URL |
|---|---|---|
| tnbc_2018 | CC BY 4.0 | https://zenodo.org/records/3552674 |
| lynsec | CC BY 4.0 | https://zenodo.org/records/8065174 |
| nuinsseg | CC BY 4.0 | https://zenodo.org/records/10518968 |
| ihc_tma | CC BY 4.0 | https://zenodo.org/records/7647846 |
| consep | Apache 2.0 | https://warwick.ac.uk/fac/cross_fac/tia/data/hovernet |
Citation
Goldsborough, T. et al. (2024) InstanSeg: an embedding-based instance segmentation algorithm optimized for accurate, efficient and portable cell segmentation. arXiv. https://doi.org/10.48550/arXiv.2408.15954
Goldsborough, T. et al. (2024) A novel channel invariant architecture for the segmentation of cells and nuclei in multiplexed images using InstanSeg. bioRxiv. https://doi.org/10.1101/2024.09.04.611150