lacss 0.14.3

Tools for cell segmentation

LACSS

LACSS is a deep-learning model for 2D/3D single-cell segmentation from microscopy images.

   pip install lacss[cuda12]

Models checkpoints

Multi-modality (2D + 3D)

name	#params	download	mAP LiveCell*	mAP Cellpose*	mAP NIPS*	ovule (3D)*	platynereis (3D)*
small	60M	model	56.3	52.0	54.2	44.4	56.7
base	152M	model	57.1	56.0	62.9	47.0	60.8
base-e	304M	model	57.4	58.3	65.7	49.8	61.9

• mAP is the average of APs at IOU threshoulds of 0.5-0.95 (10 segments). Evaluations are on either testing or validation split of the corresponding datasets.

For benchmarking (2D only)

name	#params	training data	download	AP50	AP75	mAP
small-2dL	40M	LiveCell	model	84.3	61.1	57.4
small-2dC	40M	Cellpose+Cyto2	model	87.6	62.0	56.4
small-2dN	40M	NIPS challenge	model	84.6	64.8	57.3

Deployment

You can now deploy the pretrained models as GRPC server:

   python -m lacss.deploy.remote_server --modelpath=<model_file_path>

For a GUI client see the Trackmate-Lacss project, which provides a FIJI/ImageJ plugin to perform cell segmentation/tracking in an interactive manner.

Why LACSS?

• Multi-modality: works on both 2D (multichannel) images and 3D image stacks.

• Speed: Inference time of the base model (150M parameters) is < 200 ms on GPU for an 1024x1024x3 image. We achieve this by desigining an end-to-end algorithm and aggressively eliminate CPU-dependent post-processings.

• Point-supervised training: Lacss is a multi-task model with a separate "localization" head (besides the segmentation head) predicting cell locations. This also means that you can train/fine-tune cell-segmentation using only point labels. See references for details.

Give It A Try:

Gradio Demo: try your own images (2D only)

• Demo site

Colabs

Inference

• Inference
• GRPC call

Train

• Supervised Training
• With point label only

Documentation

API documentation

References

• IEEE TMI (2023) doi:10.1109/TMI.2023.3312988
• Communications Biology 6,232 (2023)