The aim of the project is to investigate how to come up with an ensemble of convolutional neural networks for heart segmentation. This is our submission to 2018 International Workshop on Statistical Atlases and Computational Modeling of the Heart.
Please visit the organizer's website (http://stacom2018.cardiacatlas.org/) for the raw data, which contain 100 3D late gadolinium-enhanced (LGE)-MRIs with a spatial resolution of 0.625mm x 0.625mm x 0.625mm from patients with atrial fibrillation.
Anaconda Python 3.5
GPU with Kepler generation of architecture or above
GPU memory 3GB or above
It is recommended to install all the prerequisites and have suitable hardware in place.
- Wilson Fok
- Big thanks must be given to the organizer of the challenge.