Conformalized Performance Range Prediction Code

Code for "Conformal Performance Range Prediction for Segmentation Output Quality Control" accepted to MICCAI UNSURE 2024.

Virtual Environment Setup

The code is implemented in Python 3.11.2. One way of getting all the requirements is using virtualenv and the requirements.txt file.

Set up a virtual environment (e.g. conda or virtualenv) with Python 3.11.2 Install as follows: pip install -r requirements.txt

Data

First download the official FIVES dataset from here. Then preprocess and save to H5 using the FIVES_toh5.py script.

Model Training

Train all models (included in src) using the train.py script.

Performance Range

Perform conformalized performance range prediction on the test set using PerformancePrediction_tta.py for TTA and PerformancePrediction.py for all other models.