Welcome to GAN-MAT

GAN-MAT: Generative Adversarial Network-based Microstructural profile covariance Analysis Toolbox

GAN-MAT is a comprehensive pipeline to analyze brain microstructure using only the T1-weighted MRI.

Three main features of GAN-MAT:

1. T1w to T2w MRI synthesis

GAN-MAT synthesizes 3D T2-weighted MRI from 3D T1-weighted MRI using a conditional generative adversarial network (GAN).

2. Myelin-sensitive proxy calculation

Using the synthesized T2-weighted MRI, GAN-MAT calculates the ratio between T1- and T2-weighted MRI.

3. Microstructural profile covariance and gradient generation

The ready-to-use microstructural profile covariance (MPC) matrix, microstructural gradient, and moment features are computed.

Contents

• INSTALLATION
• USAGE NOTES
• MAIN OUTPUTS
• REFERENCES & ACKNOWLEDGMENTS

Core developers

• Yeongjun Park, MIP Lab - Sungkyunkwan University

• Bo-yong Park, CAMIN Lab - Inha University

& the team

• Mi Ji Lee, Seoul National University Hospital

• Seulki Yoo, CAMIN Lab - Inha University

• Chae Yeon Kim, CAMIN Lab - Inha University

• Jong Young Namgung, CAMIN Lab - Inha University

• Yunseo Park, CAMIN Lab - Inha University

• Hyunjin Park, MIPL - Sungkyunkwan University

• Eunjung Lee, Poderosa

• Yeodong Yun, Poderosa

• Casey Paquola, Multiscale Neuroanatomy Lab - INM-1 at Forschungzentrum Juelich

• Boris Bernhardt, MICA Lab - Montreal Neurological Institute