pycroglia 1.0.0

Software for isolating and analyzing microglial morphology.

Pycroglia

A Python-based toolkit for quantitative 3D morphology analysis of cells from fluorescence microscopy images.

Originally based on the MATLAB tool CellSelect-3DMorph, Pycroglia reconstructs individual cells voxel-by-voxel and extracts quantitative morphological descriptors. It is built with a robust, extensible Python architecture and supports both a PyQt6 graphical interface and a library mode for automated/scripted workflows.

---

Features

Morphological Analysis

• Cell volume – Voxel-accurate 3D cell volume computation
• Territorial volume – Convex-hull based territorial volume and coverage statistics
• Ramification index – Complexity ratio comparing cell volume to territorial volume
• Branch analysis – Branch length (with linear, PCHIP, and spline interpolation), branch count, branch order classification (primary to quaternary), end point detection
• Centroid analysis – 3D centroid coordinates and mean pairwise distances in physical units
• Convex hull metrics – Hull volume, surface area, solidity, and compactness ratios

Image Processing Pipeline

• Multi-channel reader – TIFF and LSM file format support with automatic channel extraction
• Otsu thresholding – Per-slice MATLAB-compatible Otsu thresholding with interactive adjustment
• Small object removal – Minimum-size filtering to eliminate noise artifacts
• Morphological erosion – Configurable 2D/3D structuring elements (disk, diamond, ball, octahedron, rectangle)
• GMM clustering – Gaussian Mixture Model-based nucleus separation for splitting merged cells
• Interactive segmentation – Per-cell segmentation with undo/rollback support

3D Skeletonization

• Lee-Kashyap-Chu algorithm – Full implementation of the 3D thinning algorithm with based Skeleton3D method used by CellSelect-3DMorph.
• Scikit-image fallback – Alternative skeletonization backend via skimage.morphology.skeletonize
• Skeleton-to-graph conversion – Convert binary skeletons to graph representations to prune short branches and back.

Visualization & Export

• 3D rendering – PyVista-powered visualizations: original cell surfaces, skeleton branch order coloring, end points, branch points, convex hulls
• Geometry export – Export skeleton surfaces, cell mask surfaces, and boolean mask volumes in OBJ, PLY, VTP, VTK, VTI formats
• Results export – Structured output to Excel (XLSX) and JSON
• Branch lengths export – Specialized branch-lengths XLSX output for downstream analysis

Workflow & Performance

• Wizard-based GUI – Step-by-step workflow: file selection → filtering → segmentation → cell selection → analysis dashboard
• Library mode – Import pycroglia.core directly for automated batch processing
• Parallel computation – Qt thread pool and multiprocessing backends via a unified Pool facade; metrics computed concurrently via a DAG-based orchestrator
• Interactive dashboards – Per-file tabs throughout the pipeline, summary tables, per-cell viewers, and graph preview before export

---

Installation

Option 1

Download executable files already packages.

Option 2 — Run using uvx

Prerequisites

• uv

Install/Update and Run

uvx pycroglia

Option 3 — Install with pip

• Python 3.11 or later
• pip

pip install pycroglia
pycroglia

Option 4 — From source

git clone https://github.com/CGK-Laboratory/pycroglia.git
cd pycroglia
uv run python -m pycroglia

Run the test suite:

uv run pytest

Jupyter Notebook

Launch a notebook server connected to the project environment:

uv run --with jupyter jupyter lab

---

Usage

GUI Mode

Launch the application and follow the wizard:

1. File Selection – Select TIFF or LSM image files
2. Filter Editor – Adjust Otsu threshold per slice, remove small objects, configure erosion
3. Segmentation Editor – View labeled cells by size, split merged cells via GMM
4. Cell Selection – Filter cells by size, remove border cells, preview selections
5. Results Dashboard – Set physical scale, compute all metrics, preview 3D visualizations, export results

---

Documentation

Full documentation is available at CGK-Laboratory.github.io/pycroglia (or locally via docs/).

---

Contributing

We welcome contributions! Please see CONTRIBUTING.md for guidelines.

---

License

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