CellSAM
A foundation model for Cell Segmentation
Segment Anything Model - Vision Transformer
cellsam.deepcell.org
Outperforming CellPose
Similar to Human
Single-cell spatial transcriptomics
Live-cell tracking
3D cell segmentation
@natmethods.nature.com 2025
www.nature.com/articles/s41...
A foundation model for Cell Segmentation
Segment Anything Model - Vision Transformer
cellsam.deepcell.org
Outperforming CellPose
Similar to Human
Single-cell spatial transcriptomics
Live-cell tracking
3D cell segmentation
@natmethods.nature.com 2025
www.nature.com/articles/s41...
December 11, 2025 at 12:36 PM
CellSAM
A foundation model for Cell Segmentation
Segment Anything Model - Vision Transformer
cellsam.deepcell.org
Outperforming CellPose
Similar to Human
Single-cell spatial transcriptomics
Live-cell tracking
3D cell segmentation
@natmethods.nature.com 2025
www.nature.com/articles/s41...
A foundation model for Cell Segmentation
Segment Anything Model - Vision Transformer
cellsam.deepcell.org
Outperforming CellPose
Similar to Human
Single-cell spatial transcriptomics
Live-cell tracking
3D cell segmentation
@natmethods.nature.com 2025
www.nature.com/articles/s41...
Despite great progress in image analyses, analyzing live-cell images remains highly challenging, which requires further improvement of the already state-of-the-art algorithms such as Cellpose developed for snapshot images.
Check out our preprint: www.biorxiv.org/content/10.1...
Check out our preprint: www.biorxiv.org/content/10.1...
LivecellX: Corrective Deep Learning for Object-Oriented Single-Cell Analysis in Live-Cell Imaging
Live cell imaging uniquely captures single-cell dynamics in space and time, but robust analysis is limited by segmentation and tracking errors that accumulate across frames. We present LivecellX, a de...
www.biorxiv.org
December 3, 2025 at 11:32 PM
Despite great progress in image analyses, analyzing live-cell images remains highly challenging, which requires further improvement of the already state-of-the-art algorithms such as Cellpose developed for snapshot images.
Check out our preprint: www.biorxiv.org/content/10.1...
Check out our preprint: www.biorxiv.org/content/10.1...
- new Cellpose 3 integration
Cellpose 3 is wonderful, easier to install and supports GPUs everywhere.
Cellpose 3 is wonderful, easier to install and supports GPUs everywhere.
November 20, 2025 at 11:42 AM
- new Cellpose 3 integration
Cellpose 3 is wonderful, easier to install and supports GPUs everywhere.
Cellpose 3 is wonderful, easier to install and supports GPUs everywhere.
- Cellpose-SAM integration
There is a dedicated detector for Cellpose SAM. It takes into account the must lower number of parameters. I have included one: the possibility to pass to the model one channel or all channels. This might impact the segmentation accuracy based on your input.
There is a dedicated detector for Cellpose SAM. It takes into account the must lower number of parameters. I have included one: the possibility to pass to the model one channel or all channels. This might impact the segmentation accuracy based on your input.
November 20, 2025 at 11:42 AM
- Cellpose-SAM integration
There is a dedicated detector for Cellpose SAM. It takes into account the must lower number of parameters. I have included one: the possibility to pass to the model one channel or all channels. This might impact the segmentation accuracy based on your input.
There is a dedicated detector for Cellpose SAM. It takes into account the must lower number of parameters. I have included one: the possibility to pass to the model one channel or all channels. This might impact the segmentation accuracy based on your input.
Israel A. Huaman, Fares D. E. Ghorabe, Sofya S. Chumakova, Alexandra A. Pisarenko, Alexey E. Dudaev, Tatiana G. Volova, Galina A. Ryltseva, Sviatlana A. Ulasevich, Eka...
Cellpose+, a morphological analysis tool for feature extraction of stained cell images
https://arxiv.org/abs/2410.18738
Cellpose+, a morphological analysis tool for feature extraction of stained cell images
https://arxiv.org/abs/2410.18738
October 25, 2024 at 8:32 AM
Israel A. Huaman, Fares D. E. Ghorabe, Sofya S. Chumakova, Alexandra A. Pisarenko, Alexey E. Dudaev, Tatiana G. Volova, Galina A. Ryltseva, Sviatlana A. Ulasevich, Eka...
Cellpose+, a morphological analysis tool for feature extraction of stained cell images
https://arxiv.org/abs/2410.18738
Cellpose+, a morphological analysis tool for feature extraction of stained cell images
https://arxiv.org/abs/2410.18738
More details can be found in this discussion. I have a few real images with structure at different scales for which I am going to run cellpose-sam vs cellpose-cyto3 tests on shortly. forum.image.sc/t/cellpose4-...
Cellpose4 (Cellpose-SAM) tests
Below is cellpose3-cyto3 compared to cellpose4-sam on a resolution test image which consists of circles of different sizes. This is based on a test you can find here which shows how to make the spher...
forum.image.sc
May 6, 2025 at 1:15 PM
More details can be found in this discussion. I have a few real images with structure at different scales for which I am going to run cellpose-sam vs cellpose-cyto3 tests on shortly. forum.image.sc/t/cellpose4-...
Shuo Zhao, Jianxu Chen
Data Efficiency and Transfer Robustness in Biomedical Image Segmentation: A Study of Redundancy and Forgetting with Cellpose
https://arxiv.org/abs/2511.04803
Data Efficiency and Transfer Robustness in Biomedical Image Segmentation: A Study of Redundancy and Forgetting with Cellpose
https://arxiv.org/abs/2511.04803
November 10, 2025 at 8:44 AM
Shuo Zhao, Jianxu Chen
Data Efficiency and Transfer Robustness in Biomedical Image Segmentation: A Study of Redundancy and Forgetting with Cellpose
https://arxiv.org/abs/2511.04803
Data Efficiency and Transfer Robustness in Biomedical Image Segmentation: A Study of Redundancy and Forgetting with Cellpose
https://arxiv.org/abs/2511.04803
Ignacio Arganda-Carreras gave us a lecture on the history of Deep Learning in bioimage analysis, stardist cellpose, and how the postprocessing of those outputs are designed.
January 28, 2025 at 7:35 PM
Ignacio Arganda-Carreras gave us a lecture on the history of Deep Learning in bioimage analysis, stardist cellpose, and how the postprocessing of those outputs are designed.
Qupath + Cellpose + Stardistで細胞質と核をsegmentaionめっちゃ綺麗にできて、細胞ごとにデータ得られる！すごすぎ！
June 15, 2024 at 6:13 AM
Qupath + Cellpose + Stardistで細胞質と核をsegmentaionめっちゃ綺麗にできて、細胞ごとにデータ得られる！すごすぎ！
🔬 ML-based segmentation (CellPose, ilastik, QuPath) will replace manual ImageJ counting
Why: reproducibility, throughput, objective quantitation
Lore: Reviewers now expect automated or blinded image analysis
(9/10)
Why: reproducibility, throughput, objective quantitation
Lore: Reviewers now expect automated or blinded image analysis
(9/10)
November 9, 2025 at 4:50 PM
🔬 ML-based segmentation (CellPose, ilastik, QuPath) will replace manual ImageJ counting
Why: reproducibility, throughput, objective quantitation
Lore: Reviewers now expect automated or blinded image analysis
(9/10)
Why: reproducibility, throughput, objective quantitation
Lore: Reviewers now expect automated or blinded image analysis
(9/10)
By the way, installation in M1 mac was a bit of a challenge but summarized to the following sequence of installation commands. The problem was with `numba` installation which should happen independently before the cellpose installation.
January 28, 2025 at 7:24 PM
By the way, installation in M1 mac was a bit of a challenge but summarized to the following sequence of installation commands. The problem was with `numba` installation which should happen independently before the cellpose installation.
Difficult to generalize but Cellpose was probably simplest to install even though I ran the remaining two from Google Colab. Low-res images were easier to segment with all 3 tools (cell boundaries are clearly visible). MicroSam surprisingly good even with the confocal but still not reliable.
February 16, 2025 at 9:38 PM
Difficult to generalize but Cellpose was probably simplest to install even though I ran the remaining two from Google Colab. Low-res images were easier to segment with all 3 tools (cell boundaries are clearly visible). MicroSam surprisingly good even with the confocal but still not reliable.
Absolutely agree. The real challenge isn’t generating probability maps, but separating dense/touching cells. In my experience, postprocessing (Voronoi-Otsu, watershed, etc.) always needs dataset-specific tuning. For standard data, Cellpose seems often faster and just as accurate.
July 18, 2025 at 7:37 AM
Absolutely agree. The real challenge isn’t generating probability maps, but separating dense/touching cells. In my experience, postprocessing (Voronoi-Otsu, watershed, etc.) always needs dataset-specific tuning. For standard data, Cellpose seems often faster and just as accurate.
3D #SpatialMultiomics on the way!😍
Cycle Hybridization Chain Reaction
14 bp split L+R DNA barcodes
3 color channels
Cellpose-based 3D nucleus segmentation
120 RNA probes + 8 antibodies
~50 µm hippocampal slice
44 imaging cycles over 11 days
@science.org 2025
www.science.org/doi/10.1126/...
Cycle Hybridization Chain Reaction
14 bp split L+R DNA barcodes
3 color channels
Cellpose-based 3D nucleus segmentation
120 RNA probes + 8 antibodies
~50 µm hippocampal slice
44 imaging cycles over 11 days
@science.org 2025
www.science.org/doi/10.1126/...
March 20, 2025 at 10:44 AM
3D #SpatialMultiomics on the way!😍
Cycle Hybridization Chain Reaction
14 bp split L+R DNA barcodes
3 color channels
Cellpose-based 3D nucleus segmentation
120 RNA probes + 8 antibodies
~50 µm hippocampal slice
44 imaging cycles over 11 days
@science.org 2025
www.science.org/doi/10.1126/...
Cycle Hybridization Chain Reaction
14 bp split L+R DNA barcodes
3 color channels
Cellpose-based 3D nucleus segmentation
120 RNA probes + 8 antibodies
~50 µm hippocampal slice
44 imaging cycles over 11 days
@science.org 2025
www.science.org/doi/10.1126/...
Save the dates for the upcoming #YMIA python-based event series!
🔧Setting up Conda environments
🧠Intro to #LLMs in Bioimage Analysis
💡 Mastering Prompt Engineering
🔬 Hands-on with Cellpose for image segmentation
Perfect for bioimage enthusiasts & ML beginners! #AI #Bioinformatics #MachineLearning
🔧Setting up Conda environments
🧠Intro to #LLMs in Bioimage Analysis
💡 Mastering Prompt Engineering
🔬 Hands-on with Cellpose for image segmentation
Perfect for bioimage enthusiasts & ML beginners! #AI #Bioinformatics #MachineLearning
September 27, 2024 at 10:50 AM
Save the dates for the upcoming #YMIA python-based event series!
🔧Setting up Conda environments
🧠Intro to #LLMs in Bioimage Analysis
💡 Mastering Prompt Engineering
🔬 Hands-on with Cellpose for image segmentation
Perfect for bioimage enthusiasts & ML beginners! #AI #Bioinformatics #MachineLearning
🔧Setting up Conda environments
🧠Intro to #LLMs in Bioimage Analysis
💡 Mastering Prompt Engineering
🔬 Hands-on with Cellpose for image segmentation
Perfect for bioimage enthusiasts & ML beginners! #AI #Bioinformatics #MachineLearning
Cool, thank you. I was asking because I had problems with Conda and CellPose
March 12, 2025 at 4:22 PM
Cool, thank you. I was asking because I had problems with Conda and CellPose
Followed by a very interesting presentation from Stephan Fisher on the Bioinformatics analysis part of the spatial omics pipeline, and a great improvised session of retraining Cellpose in QuPath using SAM annotation by Thierry Pecot
May 15, 2025 at 3:53 PM
Followed by a very interesting presentation from Stephan Fisher on the Bioinformatics analysis part of the spatial omics pipeline, and a great improvised session of retraining Cellpose in QuPath using SAM annotation by Thierry Pecot
Cellpose-SAM: superhuman generalization for cellular segmentation [new]
Adapts a foundation model to Cellpose, improving generalization by increasing robustness to image distortions, surpassing inter-human agreement.
Adapts a foundation model to Cellpose, improving generalization by increasing robustness to image distortions, surpassing inter-human agreement.
May 1, 2025 at 7:05 PM
Cellpose-SAM: superhuman generalization for cellular segmentation [new]
Adapts a foundation model to Cellpose, improving generalization by increasing robustness to image distortions, surpassing inter-human agreement.
Adapts a foundation model to Cellpose, improving generalization by increasing robustness to image distortions, surpassing inter-human agreement.
Cellpose-SAM is currently available in the GUI.
May 4, 2025 at 2:47 PM
Cellpose-SAM is currently available in the GUI.
Cellpose as a reliable method for single-cell segmentation of autofluorescence microscopy images https://www.biorxiv.org/content/10.1101/2024.06.07.597994v1
Cellpose as a reliable method for single-cell segmentation of autofluorescence microscopy images https://www.biorxiv.org/content/10.1101/2024.06.07.597994v1
Autofluorescence microscopy uses intrinsic sources of molecular contrast to provide cellular-level i
www.biorxiv.org
June 10, 2024 at 4:46 PM
Cellpose as a reliable method for single-cell segmentation of autofluorescence microscopy images https://www.biorxiv.org/content/10.1101/2024.06.07.597994v1
Following general
introduction, we try to understand how the famous CellPose works.
www.nature.com/articles/s41...
introduction, we try to understand how the famous CellPose works.
www.nature.com/articles/s41...
March 20, 2025 at 2:13 PM
Following general
introduction, we try to understand how the famous CellPose works.
www.nature.com/articles/s41...
introduction, we try to understand how the famous CellPose works.
www.nature.com/articles/s41...
🧫Does anyone know of database for 2D plant anatomical data?
I’m especially thinking of datasets from: CellSeT, Icy (.xml) or .zip ImageJ ROIs (e.g., via Celer, PaCeQuant, Cellpose, etc.)
Something like what’s being done with ggPlantMap would be amazing!
#PlantScience #OpenScience
I’m especially thinking of datasets from: CellSeT, Icy (.xml) or .zip ImageJ ROIs (e.g., via Celer, PaCeQuant, Cellpose, etc.)
Something like what’s being done with ggPlantMap would be amazing!
#PlantScience #OpenScience
April 11, 2025 at 8:35 AM
🧫Does anyone know of database for 2D plant anatomical data?
I’m especially thinking of datasets from: CellSeT, Icy (.xml) or .zip ImageJ ROIs (e.g., via Celer, PaCeQuant, Cellpose, etc.)
Something like what’s being done with ggPlantMap would be amazing!
#PlantScience #OpenScience
I’m especially thinking of datasets from: CellSeT, Icy (.xml) or .zip ImageJ ROIs (e.g., via Celer, PaCeQuant, Cellpose, etc.)
Something like what’s being done with ggPlantMap would be amazing!
#PlantScience #OpenScience
Turns out Transformers do outperform Cellpose when trained by Marius and Carsen 😄 Super exiting to see this being released and can't wait to try it out on our data!
🚀🔬🦠 Releasing 🤖Cellpose-SAM🤖, a cellular segmentation algorithm with superhuman generalization 🦸♀️. Try it now on 🤗 huggingface.co/spaces/mouse...
paper: www.biorxiv.org/content/10.1...
w/ @computingnature.bsky.social 1/n
paper: www.biorxiv.org/content/10.1...
w/ @computingnature.bsky.social 1/n
May 5, 2025 at 5:33 PM
Turns out Transformers do outperform Cellpose when trained by Marius and Carsen 😄 Super exiting to see this being released and can't wait to try it out on our data!
evaluation of real-time segmentation methods and a realtime data analysis dashboard. Our autofocusing achieves a Mean Absolute Error of 0.0226\textmu m with inference times below 50~ms. Among eleven Deep Learning segmentation methods, Cellpose~3 [3/4 of https://arxiv.org/abs/2504.00047v1]
April 2, 2025 at 6:16 AM
evaluation of real-time segmentation methods and a realtime data analysis dashboard. Our autofocusing achieves a Mean Absolute Error of 0.0226\textmu m with inference times below 50~ms. Among eleven Deep Learning segmentation methods, Cellpose~3 [3/4 of https://arxiv.org/abs/2504.00047v1]