CellposeAdapt: Tailoring Models to Unique Data

This project is part of the FIP project appl-fm and provides tools for visualizing and evaluating cell segmentation results using the Cellpose model. It includes scripts for processing images, evaluating model performance, and visualizing results with Napari.

Installation

1. Clone the repository:

   git clone https://github.com/mario-koddenbrock/cellpose-adapt.git
   cd cellpose-viz

2. Install the required packages:

   pip install -r requirements.txt

Usage

Main Script

The main.py script processes images, evaluates the model, and visualizes results with Napari.

Parameters:

• image_path (str): Path to the image file.
• param_file (str): Path to the parameter YAML file.
• output_dir (str): Directory to save the output.
• cache_dir (str): Directory for cache files (default is '.cache').
• show_gt (bool): Show ground truth labels.
• show_prediction (bool): Show prediction labels.
• video_3d (bool): Export 3D video.
• show_viewer (bool): Show Napari viewer.
• export_video (bool): Export video.
• type (str): Type of segmentation ('Nuclei' or 'Membranes').

Example:

python cellpose_adapt/main.py --image_path path/to/image.tif --param_file path/to/params.yaml --output_dir Segmentation --show_gt --show_prediction --show_viewer

Visualize Best Scores

The viz.py script visualizes the best score for each image and type as a grouped bar plot.

Parameters:

• file_path (str): Path to the CSV file containing experiment results.
• metric (str): The column name of the metric to visualize (default is 'jaccard').
• output_file (str): Path to save the bar plot (default is 'best_scores_barplot.png').

Example:

python cellpose_adapt/viz.py --file_path results.csv --metric jaccard --output_file best_scores_barplot.png

License

This project is licensed under the MIT License. See the LICENSE file for details.