• January 7, 2020
  • N/A

Trained models for VollSeg

Trained models for denoising using CARE, segmentationusing UNET and StarDist to be used in the Napari plugin VollSeg

1. VollSeg models on Zenodo

The trainedmodels with hyperparameter optimized on the training data are provided and catalogued here.

1) Generic denoising model in 3D to be used for transfer learning to train CARE models for denoising
with the same model parameters as are in the json file of model parameters:

2) CARE trained denoising model for human breast carcinoma cells imaged using concfocal microscope:

3) UNET trained model for light sheet imaged Ascadian embryo cells fused from four angles :

4) UNET trained model for Xenopus tissue in 2D:

5) StarDist trained model for white blood cells in 2D:

6) StarDist trained model for human breast carcinoma cells in 3D:

7) StarDist trained model for human breast carcinoma cells in 3D:


8) Unet trained segmentation model for microtubules kymograph imaged using TIRF microscope:


9) Unet trained model for white blood cells in 2D:

10) Tissue segmentation for Xenopus cells in 3D:

11) Tissue segmentation for Xenopus cells in 3D:

12) Unet trained model for Cytoplasm segmentation in2D:

13) Unet trained model for Human lung segmentation imaged using X rays:

2. Demo videos on youtube

Using these models with VollSeg Napari plugin is explained in the demo videos hosted on Youtube and provided here.

3. Training Dataset

The training dataset that we created are catalogued here.

1.a) Raw data for Arabidopsis.

1.b) Labels for Arabidopsis.

1.c) Binary masks for Arabidopsis.

2.a) Raw data for four angle fused dataset for Ascidian embryo image via light sheet, the link to label data is also provided at the link.

3) Fully annotated Human breast carcinoma cells for 3D segmentation training.

Tags:

Leave a Reply

Your email address will not be published. Required fields are marked *