3D Brain MRI Synthesis

Conditional Latent Diffusion Model for Synthetic 3D Brain MRI Generation to Enhance Temporal Lobe Epilepsy Detection

3D Brain MRI Synthesis using DiT-3D | Graduate Student Researcher, UCSD Health

- Developed a novel conditional latent diffusion pipeline by integrating a VAE-GAN for latent space compression with a DiT-3D based diffusion transformer, enabling the generation of high-fidelity, class-specific synthetic 3D brain MRIs
- Augmented the real brain MRI dataset with synthetic data, leading to substantial improvements in accuracy, precision, recall, F1-score, AUC-ROC, and AUCPR for a Temporal Lobe Epilepsy classifier trained on an EfficientNet-V2 backbone.

[Presentation] [Project Report]