Fluorescence Image Analysis Pipeline for Mitochondrial Z-Projections

Overview

This Python script provides an enhanced image analysis pipeline for fluorescence microscopy images. It leverages the napari viewer and the napari-sam plugin (note: the repository is named napari-sam, and the installable package is napari-sam) to facilitate interactive segmentation of cells and intracellular objects (e.g., mitochondria) using the Segment Anything Model (SAM). Following segmentation, the script extracts a comprehensive set of features for each segmented object, including advanced morphology descriptors and intensity measurements, enabling detailed quantitative analysis of cellular structures.

|napari-sam Operations | Commonly Used Mouse Keys| |---|---| | Select Intensity Points | Middle Mouse Button| | Deselect Intensity Points | Ctrl + Middle Mouse Button| |Undo Selection | Ctrl + Z |

Key Features

• Interactive Segmentation with SAM: Integrates with the napari-sam plugin for user-friendly, AI-assisted segmentation of cells and objects directly within Napari.
• Comprehensive Feature Extraction: Calculates a wide range of features for each segmented object.
• Cell-Specific Object Analysis: Optionally performs cell-specific analysis.
• Robust Error Handling and Debugging: Includes checks for image and mask validity and error handling.
• Clean Output to CSV: Exports extracted features to a well-structured CSV file (object_features_enhanced.csv).
• Napari Integration: Keeps the Napari viewer open after analysis.

Installation

Before running this script, ensure you have the following installed:

• Python: Python 3.7 or higher is recommended.

• PyTorch: Required by napari-sam. Install PyTorch following the instructions on the official PyTorch website: https://pytorch.org/get-started/locally/

  • If you have an NVIDIA GPU: Determine your CUDA version (nvidia-smi) and select the appropriate PyTorch version.
  • If you do not have an NVIDIA GPU: Select the CPU-only version.

  Example (GPU, CUDA 11.8 - adjust as needed):

  bash pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118 Example (CPU): bash pip install torch torchvision torchaudio

• Napari:

  bash pip install napari[all]

• napari-sam Plugin:

  Method 1: Using pip:

  bash pip install napari-sam

  Method 2: Using Napari GUI:

1.  Open Napari.
2.  **Plugins > Install/Uninstall Plugins...**.
3.  Search for `segment-anything` and install `napari-sam`.
4.  Restart Napari if prompted.
5.  Activate via **Plugins > Segment Anything > Segment Anything** and download a SAM model.

• Other Libraries:

  bash pip install numpy>=1.20 pandas>=1.3 scikit-image>=0.18 scipy>=1.7 napari[all] scikit-network networkx

Usage Instructions

1. Download the Script: Download the fluorescence_analyzer_enhanced.py script.

2. Open Napari via the Script:

   bash python fluorescence_analyzer_enhanced.py

3. Load Your Image: In Napari, drag and drop your image, or use File > Open File(s)....

4. Segment Cells (Optional):

*   Use `napari-sam` (**Plugins > Segment Anything > Segment Anything**).
*   Create cell segmentations.
*   **Rename the layer to `cell_masks`.**

1. Segment Objects:

*   Use `napari-sam`.
*   **Rename this layer to `object_masks`.**

1. Trigger Feature Extraction: In the terminal, press Enter.

2. Feature Extraction and Output: The script will:

*   Extract features.
*   Save to `object_features_enhanced.csv`.
*   Display a summary in the terminal.
*   Keep Napari open.

1. Analyze Results: Open object_features_enhanced.csv in your preferred data analysis tool.

Note:

• No cell_masks layer skips cell-specific analysis.
• Layer names must be exactly cell_masks and object_masks.
• Adjust image_path in the script to auto-load an image.

Features Extracted

1. Basic Region Properties

These features provide fundamental descriptors of object shape and location.

| Feature | Description | | --------------------- | ----------------------------------------------------------------------------------------------------------------------------- | | label | Object Identifier. A unique integer ID for each segmented object. | | object_area | Object Area. The number of pixels within the object's mask. | | centroid-0, centroid-1 | Object Centroid. (y, x) or (row, column) coordinates of the object's center. | | length | Major Axis Length. Length of the major axis of the best-fitting ellipse. | | width | Minor Axis Length. Length of the minor axis of the best-fitting ellipse. | | shape_eccentricity | Eccentricity. A measure of how elongated the object is (0 = circle, 1 = line). | | shape_solidity | Solidity. Ratio of the object's area to its convex hull area (closer to 1 = more convex, less concave). | | shape_extent | Extent. Ratio of the object's area to the area of its bounding box. | | shape_orientation | Orientation. Angle (in radians) of the major axis of the best-fitting ellipse, relative to the horizontal axis. | | object_perimeter | Perimeter. The total length of the object's boundary. |

2. Intensity Features

These features quantify the fluorescence intensity within each object.

| Feature | Description | | --------------------- | ------------------------------------------------------------------------------------------------- | | intensity_mean | Mean Intensity. The average pixel intensity within the object. | | intensity_integrated | Integrated Intensity. The sum of all pixel intensities within the object. | | intensity_raw_min | Minimum Intensity. The lowest pixel intensity within the object. | | intensity_raw_max | Maximum Intensity. The highest pixel intensity within the object. | | intensity_raw_std | Intensity Standard Deviation. The standard deviation of pixel intensities within the object. |

3. Enhanced Morphology Features

These features provide more sophisticated measures of object shape, derived from the basic properties.

| Feature | Description | | ------------------------ | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | aspect_ratio | Aspect Ratio. Calculated as length / width. Indicates object elongation. | | form_factor | Form Factor (Circularity). Calculated as (4 * π * object_area) / (object_perimeter²). Values closer to 1 indicate a more circular shape. | | perimeter_area_ratio | Perimeter-Area Ratio. Calculated as object_perimeter / object_area. A 2D proxy for surface area-to-volume ratio; higher values indicate more complex shapes or smaller objects of the same shape. | | convexity | Convexity. Calculated as object_area / shape_solidity. Effectively the inverse of solidity, emphasizing deviation from a perfectly convex shape. | | num_branches | Number of Branches. The total number of branches detected in the object's skeleton. | | total_branch_length | Total Branch Length. The sum of the lengths of all branches in the object's skeleton. | | avg_branch_length | Average Branch Length. The average length of the branches in the object's skeleton. | | num_junctions | Number of Junctions. The number of junction points (where branches meet) in the object's skeleton. |

4. Summary & Contextual Information

| Feature | Description | | -------------- | ----------------------------------------------------------------------------------------------------------------------------------------- | | object_count | Object Count. The total number of objects analyzed. | | cell_label | Cell Label. Indicates the cell to which an object belongs ("cell_[label]") or "outside_cell" if not assigned to a segmented cell. |

Dependencies

• Python (>=3.7)
• napari
• napari-sam
• PyTorch
• numpy (>=1.20)
• pandas (>=1.3)
• scikit-image (skimage) (>=0.18)
• scipy (>=1.7)