3D-Cine

This repository contains code accompanying the 3D Cine paper by Mark Wrobel.

The pipeline is divided into three main sections:

---

Code Structure

1. Training Data Pre-processing

The code uses two publicly available datasets:

• MMWHS
• HVSMR

After downloading, place the data in the corresponding empty folders within the repository.
Note: MMWHS requires separate folders for images and segmentations.

To pre-process the data:

1. Run HVSMR_pre_processing.ipynb
2. Run MMWHS_pre_processing.ipynb
3. Run training_data_preprocessing.ipynb

This will prepare all training data required for model training.

---

2. Model Training

Once the data is pre-processed, train the deep learning models by running the following scripts:

• 3D_debanding_train.py
• 3D_respcor_train.py
• 3D_E2E_train.py
• 3D_seg_train.py

Make sure to update the mmwhs_number and hvsmr_number variables to reflect the number of processed datasets.

---

3. Inference and Post-processing

Run 3D_cine_post_processing.ipynb.

This step can be run independently of the training pipeline.
An example healthy volunteer dataset is located in the raw_data folder and is processed using trained models located in models_final/.

Output 3D Cine data and segmentations are saved as .npy arrays in the processed_data folder.

---

Docker Support

A Dockerfile is included for creating a reproducible environment.