Structure Tensor Validation: Functions for validating the accuracy of structure tensor analysis for estimating orientations in 2D and 3D images.

Features

• Generate 2D and 3D phantoms simulating axons in brain microscopy with parallel and crossing patterns.
• Perform structure tensor analysis to obtain orientations in 2D and 3D images.
• Visualize image orientations and their distributions.
• Estimate k-means of image orientations.
• Test the accuracy of estimated orientations.

Installation and Setup

First create and activate a new virtual environment with the following command using venv:

python3 -m venv env
source env/bin/activate

Or with conda:

conda create -n env
conda activate env

Note: This code has been tested with python version 3.11. It may not be compatable with other versions.

Then install package requirements:

pip install -r requirements.txt

Usage

Open the Jupyter notebook run_st_analysis_validation.ipynb for an example walkthrough of our validation pipeline.

Supplementary Data

The unaltered output data from structure tensor analysis validation is located in the outputs folder. These are stored in comma separated files where each row corresponds to a different parameter configuration. The phantom and structure tensor analysis parameters are listed in the first columns and the resulting error for each setting is in the last column.

The example_microscopy_data folder contains the two example microscopy patches -- one 2D and one 3D -- that were used to illustrate the relationship between simulated phantoms and real image data.