Keras implementation of modified pix2pix with 3D convolutions. Developed for CT data.
This repository contains the source code of the following paper:
Generating Synthetic Contrast Enhancement from Non-contrast Chest Computed Tomography Using a Generative Adversarial Network
Scientific Reports 2021 Oct 14;11(1):20403. doi: 10.1038/s41598-021-00058-3.
To try out your own training and inference, each case in the data set should contain a pair of stack of axial CT scans in DICOM format. Please refer to train.ipynb and inference.ipynb for details.
Although this project was developed for CT data, the pix2pix3D network can work for any type of input data, if source/data_loader.py is properly modified.