T-REX

This repository contains source code to generate the T-REX visualizations described in Maloca et al., Unraveling the deep learning gearbox in optical coherence tomography image segmentation towards explainable artificial intelligence. This is, Multidimensional Scaling (MDS) plots and heatmap plots based on Hamming distances that measure the distance between two pixel-wise labellings.

Setup

• Create a virtual environment with Python 3.8 64-bit with venv. See https://packaging.python.org/guides/installing-using-pip-and-virtual-environments/
• Install dependencies with pip

  pip install -r requirements.txt
  

Visualizations

• Set PLOTS_OUT_DIR in definitions.py
• Create PLOTS_OUT_DIR
• Run unit tests in plots/test_make_plots.py to generate T-REX visualizations
• T-REX visualizations are stored in PLOTS_OUT_DIR

U-Net

The U-Net model used in this study can be found in u_net/model.py.