This repository contains the code for reproducing the image analysis pipeline described in our manuscript [1]. The pipeline uses publicly available data hosted on AWS to perform segmentation and 3D mesh generation. The outputs of this analysis are further utilized by the companion repository, EMT_data_analysis, to generate the plots and visualizations presented in the manuscript.
-
CytoGFP Groundtruth Segmentation
- This module generates ground truth segmentations for CytoGFP-tagged images.
- It is used to create accurate masks for training and validating segmentation models.
- Detailed Instructions
-
All Cells Mask Model Training and Inference
- This component trains a deep learning model to predict all cell masks from label-free images.
- It also supports inference on new datasets to generate all cell masks.
- Detailed Instructions
-
H2B and EOMES Nuclear Segmentations
- Performs instance segmentation of H2B and EOMES nuclear markers.
- This step is critical for identifying individual nuclei.
- Detailed Instructions
-
CollagenIV Segmentation
- Segments CollagenIV structures from the provided images.
- This step is essential for analyzing extracellular matrix organization.
- Detailed Instructions
-
CollagenIV Segmentation Mesh Generation
- Converts CollagenIV segmentations into 3D meshes for downstream analysis.
- These meshes are used to study the structural properties of the extracellular matrix.
- Detailed Instructions
If you have questions about this code, please reach out to us at [email protected].
All code in this repository is provided under the Allen Institute Software License.