A Python library for osteocyte network analysis using confocal laser scanning microscopy data segmented with the Tool for Image and Network Analysis (TINA), providing lamellar surface reconstruction and temporal quantification of cellular and dendritic densities via interpolated 3D Euclidean distance transforms (MLAEDT-3D).

This library extends the Tool for Image and Network Analysis (TINA) software by providing an interface for importing image and cell/segment data, as well as a novel set of tools to approximate the geometry of lamellar surfaces using a Multi-Label Anisotropic 3D Euclidean Distance Transform (MLAEDT-3D).

It computes cellular densities and dendritic surface densities as a function of the normalised distance from the cement line to the Haversian canal, enabling quantitative analysis of osteocyte network organisation within bone osteons.

This project was developed as part of an Undergraduate Engineering Honours thesis at the Queensland University of Technology (QUT), titled "Image Analysis of Osteocyte Networks in Bone", in 2025.

Supervisors:

• Associate Professor Pascal Buenzli
• Dr Jessica Crawshaw

Student:

• Tarang Janawalkar

It is recommended that this package is installed within a virtual environment using venv or conda. Mayavi is required for 3D visualisation, but due to VTK API compatibility issues, it is highly recommended to install Mayavi using conda-forge which provides pre-built binaries with compatible dependencies.

# Using conda
conda create -n osteonx "python<3.13"
conda activate osteonx
conda install -c conda-forge mayavi
conda remove qt-main pyqt # Conda might install multiple Qt versions
pip install git+https://github.com/tarang74/osteonx.git

# Using venv
# Mayavi 4.8.1-4.8.2 has been tested on Python 3.11
# Mayavi 4.8.3 has been tested on Python 3.12
python -m venv osteonx-env
source osteonx-env/bin/activate
pip install "mayavi==<version>"
pip install git+https://github.com/tarang74/osteonx.git

from pathlib import Path
from osteonx import io, analysis, visuals

# 1. Import 3D image stack
osteon = io.import_masks(
    path=Path("masks"),
    downsample=(2, 2, 1),
    um_per_voxel=(0.379, 0.379, 0.296),
)

# 2. Compute Euclidean distance transforms
dts = analysis.compute_edt(osteon, path=Path("arrays"))

# 3. Interpolate surfaces
t, phi = analysis.interpolate_surfaces(osteon, dts, tsamples=20)

# 4. Visualise
visuals.plot_surfaces_2d(phi, layers=(phi.shape[2]//2,), out="figures/lamellar_growth_2d.png")

Three example scripts are provided in the examples/ directory:

1. lamellar_growth_approximation.py - Basic surface interpolation

   python examples/lamellar_growth_approximation.py --help

   

2. cell_density_analysis.py - Osteocyte density analysis

   python examples/cell_density_analysis.py --help

   

3. segment_surface_density_analysis.py - Dendritic surface density analysis

   python examples/segment_surface_density_analysis.py --help

   

4. derivative_methods_comparison.py - Comparison of derivative methods for density estimation

   python examples/derivative_methods_comparison.py --help

   

This package uses pytest to test the analysis methods on synthetic data. Synthetic data is generated inside the generators module, which includes functions to create spherical, cylindrical, and prismatic geometries. It also generates synthetic cell and segment data arranged within a lattice or uniformly distributed within the specified volume.

These tests can be run using:

python -m pytest tests/

Test output figures have been compiled in tests/README.md.