The abstract is read at the publisher; the summary is JClub's.
SAM2, a microscopy foundation model for joint 2D and 3D segmentation, is trained on 23 million+ cells and 1.6 million structures and achieves state-of-the-art volumetric segmentation.
- Why it matters: Existing segmentation models are largely 2D-trained and struggle with complex volumetric structures, while covering only limited microscopy analysis tasks.
- What they did: SAM2 is adapted for light and electron microscopy using large-scale annotations, then evaluated on automatic and interactive segmentation plus classification and cell tracking in 2D/3D time-lapse data.
- The result: It outperforms generalist and specialist methods on volumetric data and, with dedicated models, enables accurate mitochondrion and nucleus segmentation in electron microscopy.
The findingWhy it mattersWhat they didThe result
AI summary of the abstract · may contain errors · not medical advice · read the paper.