A Foundation Model for Cell Segmentation

• DOI: 10.48550/arxiv.2311.11004
• Publication Date: 2023-01-01
• Authors: Uriah Israel; Markus Marks; Rohit Dilip; Qilin Li; Morgan Schwartz; Elora Pradhan; Edward Pao; Shenyi Li; Alexander Pearson...; Pietro Perona; Georgia Gkioxari; Ross Barnowski; Yisong Yue; David Van Valen
• Keywords: Market Segmentation

Cells are the fundamental unit of biological organization, and identifying them in imaging data - cell segmentation is a critical task for various cellular experiments. While deep learning methods have led to substantial progress on this problem, models that seen wide use specialist work well specific domains. Methods learned general notion "what cell" can identify across different domains proven elusive. In work, we present CellSAM, foundation model generalizes diverse data. CellSAM builds top Segment Anything Model (SAM) by developing prompt engineering approach mask generation. We train an object detector, CellFinder, automatically detect cells SAM generate segmentations. show allows single achieve state-of-the-art performance segmenting images mammalian (in tissues culture), yeast, bacteria collected with modalities. To enable accessibility, integrate into DeepCell Label further accelerate human-in-the-loop labeling strategies A deployed version available at https://label-dev.deepcell.org/.