Cell Segmentation and Co-localization Analysis Pipeline

A comprehensive Python pipeline for automated cell counting and co-localization analysis in fluorescence microscopy images. This toolkit processes multi-channel confocal microscopy data (.czi files) to quantify cell populations and analyze marker co-expression patterns.

🔬 Overview

This pipeline consists of two main components:

1. Image Preprocessing (01_image_preprocessing.ipynb): Extracts individual channels from multi-channel microscopy files and performs automated cell segmentation
2. Cell Counting Analysis (02_cell_counting_analysis.ipynb): Counts cells in each channel and analyzes co-localization patterns between fluorescence markers

🚀 Features

Image Preprocessing

• Multi-channel extraction: Splits .czi files into individual channel/Z-plane images
• Automated segmentation: Uses Cellpose (cpsam model) for accurate cell boundary detection
• Parameter optimization: Tools to find optimal segmentation parameters for your dataset
• Batch processing: Handle multiple files efficiently
• Quality control: Progress reporting and error handling

Cell Counting Analysis

• Multi-marker analysis: Supports 3-channel analysis (GFP, c-fos, NeuN)
• ROI-based counting: Process user-defined regions of interest
• Co-localization detection: Identifies cells expressing multiple markers
• 3D analysis: Handles multiple Z-planes independently
• Export capabilities: Generates CSV reports and ImageJ-compatible ROI files
• Boundary exclusion: Removes edge artifacts for accurate counting

📋 Prerequisites

• Python 3.8+
• CUDA-compatible GPU (optional, for faster Cellpose segmentation)
• Sufficient RAM (8GB+ recommended for large images)

🛠 Installation

1. Clone the repository:

git clone https://github.com/yourusername/cell-segmentation-pipeline.git
cd cell-segmentation-pipeline

2. Create a virtual environment (recommended):

python -m venv cell_analysis_env
source cell_analysis_env/bin/activate  # On Windows: cell_analysis_env\Scripts\activate

3. Install dependencies:

pip install -r requirements.txt

4. GPU Setup (optional but recommended):
   • For CUDA 11.x: Uncomment cupy-cuda11x>=9.0.0 in requirements.txt
   • For CUDA 12.x: Uncomment cupy-cuda12x>=12.0.0 in requirements.txt
   • Then run: pip install -r requirements.txt

📖 Usage

Step 1: Image Preprocessing

1. Open 01_image_preprocessing.ipynb in Jupyter Lab/Notebook
2. Update file paths in Cell 1 to point to your .czi files
3. Run all cells to:
   • Extract individual channels
   • Perform automated segmentation
   • Generate training data for analysis

Optional: Parameter Optimization

• Uncomment and run Cell 4 to test different Cellpose parameters
• Choose optimal settings based on your cell types and imaging conditions

Step 2: Cell Counting Analysis

1. Open 02_cell_counting_analysis.ipynb
2. Update the sample configuration in Cell 2:

   sample_list = [
       ("path/to/your/file.czi", "path/to/roi_file.zip", "path/to/cellpose_output/"),
       # Add more samples as needed
   ]

3. Run all cells to:
   • Process all ROIs and Z-planes
   • Count cells in each fluorescence channel
   • Analyze co-localization patterns
   • Generate comprehensive reports

📊 Output Files

From Image Preprocessing:

• filename_C{channel}_Z{z-plane}.ome.tif: Individual channel images
• filename_C{channel}_Z{z-plane}_mask.tiff: Segmentation masks
• filename_C{channel}_Z{z-plane}.zip: ROI coordinates for detected cells

From Cell Counting Analysis:

• cell_counting_results.csv: Comprehensive cell count table
• filename_ROI{X}_final.zip: Co-localized cell coordinates for ImageJ

📁 File Structure

cell-segmentation-pipeline/
│
├── 01_image_preprocessing.ipynb       # Step 1: Channel extraction & segmentation
├── 02_cell_counting_analysis.ipynb   # Step 2: Cell counting & co-localization
├── requirements.txt                   # Python dependencies
├── README.md                         # This file
│
└── example_data/                     # (Optional) Example datasets
    ├── sample.czi                   # Example microscopy file
    └── sample_rois.zip              # Example ROI file

🔧 Configuration

Channel Mapping

By default, the pipeline expects:

• Channel 0 (C0): GFP - Specific cell population marker
• Channel 1 (C1): c-fos - Neuronal activation marker
• Channel 3 (C3): NeuN - General neuronal marker

Modify the channel assignments in read_channels_z() function if your setup differs.

Segmentation Parameters

Key Cellpose parameters you can adjust:

• diameter: Expected cell diameter in pixels (None = auto-detect)
• flow_threshold: Quality threshold for flow (0.4 default, range 0-1)
• cellprob_threshold: Cell probability threshold (0.0 default, range -6 to 6)

Co-localization Analysis

• threshold: IoU overlap threshold for co-localization (0.8 default)
• Adjust in count_overlaps() function calls

🧪 Example Workflow

1. Prepare your data:

   • Multi-channel .czi files from confocal microscopy
   • Manually defined ROIs saved as .zip files (using ImageJ/FIJI)

2. Run preprocessing:

   jupyter lab 01_image_preprocessing.ipynb

3. Analyze cell counts:

   jupyter lab 02_cell_counting_analysis.ipynb

4. Review results:

   • Check cell_counting_results.csv for quantitative data
   • Import ROI files into ImageJ for visual validation

📈 Results Interpretation

The analysis generates a CSV table with columns:

• Z: Z-plane number
• ROI: Region of interest identifier
• GFP: Count of GFP+ cells
• c-fos: Count of c-fos+ cells
• NeuN: Count of NeuN+ cells
• GFP+ NEUN+: Count of cells expressing both GFP and NeuN
• FOS+ NEUN+: Count of cells expressing both c-fos and NeuN
• GFP+ FOS+ NEUN+: Count of cells expressing all three markers

🤝 Contributing

Contributions are welcome! Please feel free to submit a Pull Request. For major changes, please open an issue first to discuss what you would like to change.

📝 License

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

🙏 Acknowledgments

• Built with Cellpose for cell segmentation
• Uses aicspylibczi for .czi file handling
• Inspired by the need for reproducible microscopy analysis workflows

🐛 Issues and Support

If you encounter any problems or have questions:

1. Check the Issues page
2. Create a new issue with detailed description of the problem
3. Include sample data and error messages when possible

📧 Contact

For questions or collaborations, please contact: your.email@institution.edu

---

This pipeline was developed to standardize and automate fluorescence microscopy analysis workflows, making quantitative cell biology more accessible and reproducible.