NV-Generate-MR-Brain is a three-dimensional (3D) latent diffusion model designed to generate high-quality synthetic brain magnetic resonance imaging (MRI) images, achieving the highest resolution and best FID scores among comparable models. This model is specialized for brain MRI with support for multiple modalities: T1, Fluid Attenuated Inversion Recovery (FLAIR), T2, and Susceptibility Weighted Imaging (SWI).
Compared to the previous NV-Generate-MR release, the key differences are:
The model excels at data augmentation and at generating realistic medical imaging data to supplement datasets limited by privacy concerns or the rarity of certain conditions. It can also significantly enhance the performance of other medical imaging AI models by generating diverse, realistic training data.
This model is ready for commercial use.
You can try this model via NVIDIA MAISI demo:
👉 [https://huggingface.co/spaces/nvidia/nv-generate)
The current demo showcases only a subset of the model’s capabilities. For full functionality and advanced usage, please refer to the GitHub repository.
Training and inference code are in: https://github.com/NVIDIA-Medtech/NV-Generate-CTMR/tree/main.
Use of this model is governed by the NVIDIA Open Model License. Additional Information: Apache 2.0 License.
This project was conducted by NVIDIA in collaboration with the University of Zurich, Istanbul Medipol University, and Forithmus.
Global
Medical researchers, AI developers, and healthcare institutions would be expected to use this model for:
It is not a clinically validated medical device and should not be used for clinical diagnostic purposes.
Huggingface: 03/16/2026 (GTC San Jose 2026) via https://huggingface.co/nvidia/NV-Generate-MR-brain
[1] MAISI V2: https://arxiv.org/abs/2508.05772
[2] Guo, Pengfei, et al. "MAISI: Medical AI for Synthetic Imaging." arXiv preprint arXiv:2409.11169. 2024. https://arxiv.org/abs/2409.11169
[3] Rombach, Robin, et al. "High-resolution image synthesis with latent diffusion models." Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2022. https://openaccess.thecvf.com/content/CVPR2022/papers/Rombach_High-Resolution_Image_Synthesis_With_Latent_Diffusion_Models_CVPR_2022_paper.pdf
[4] Lvmin Zhang, Anyi Rao, Maneesh Agrawala; "Adding Conditional Control to Text-to-Image Diffusion Models." Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), 2023, pp. 3836-3847. https://openaccess.thecvf.com/content/ICCV2023/papers/Zhang_Adding_Conditional_Control_to_Text-to-Image_Diffusion_Models_ICCV_2023_paper.pdf
Architecture Type: Diffusion Model
Network Architecture: 3D UNet + attention blocks (latent diffusion)
Task: Generation (Synthetic MRI Image)
Number of model parameters: 240M
Cumulative Compute: 4.375 x 10^22
Estimated Energy and Emissions for Model Training: 24,085kWh
Input Type(s): Integer, Array
Input Format(s): Integer values, Float arrays
Input Parameters: Number of Samples (1D), Modality (1D), Output Size (1D), and Spacing (1D)
Other Properties Related to Input: Supports controllable synthetic brain MRI generation with modality selection, customizable output dimensions, and configurable voxel spacing.
Output Type(s): Image
Output Format: Neuroimaging Informatics Technology Initiative (NIfTI)
Output Parameters: Three-Dimensional (3D)
Other Properties Related to Output: Synthetic brain MRI images in the specified modality (T1, FLAIR, T2, or SWI). Output dimensions and spacing are configurable within supported ranges.
Our AI models are designed and/or optimized to run on NVIDIA GPU-accelerated systems. By leveraging NVIDIA's hardware (e.g. GPU cores) and software frameworks (e.g., CUDA libraries), the model achieves faster training and inference times compared to CPU-only solutions.
Runtime Engine(s):
Supported Hardware Microarchitecture Compatibility:
Supported Operating System(s):
The integration of foundation and fine-tuned models into AI systems requires additional testing using use-case-specific data to ensure safe and effective deployment. Following the V-model methodology, iterative testing and validation at both unit and system levels are essential to mitigate risks, meet technical and functional requirements, and ensure compliance with safety and ethical standards before deployment.
0.1 - Initial release version for synthetic brain MRI image generation
The model was trained on the MR-Rate brain MRI dataset covering the supported modalities (T1, FLAIR, T2, and SWI). Data from multiple scanner types were processed to create high-quality 3D MRI volumes with corresponding anatomical annotations. The data processing pipeline ensured consistent voxel spacing, standardized orientations, and validated anatomical segmentations.
Data Modality:
Image Training Data Size:
Data Collection Method by dataset:
Labeling Method by dataset:
Properties: Approximately 28,000 MRI scans of various types (T1, FLAIR, T2, and SWI).
Data Modality:
Image Training Data Size:
Data Collection Method by dataset:
Labeling Method by dataset:
Properties: Approximately 8,000 MRI scans of various types (T1, FLAIR, T2, and SWI).
Data Modality:
Image Training Data Size:
Data Collection Method by dataset:
Labeling Method by dataset:
Properties: Approximately 4,000 MRI scans of various types (T1, FLAIR, T2, and SWI).
Acceleration Engine: PyTorch
Test Hardware:
NVIDIA believes Trustworthy AI is a shared responsibility and we have established policies and practices to enable development for a wide array of AI applications. When downloaded or used in accordance with our terms of service, developers should work with their internal model team to ensure this model meets requirements for the relevant industry and use case and addresses unforeseen product misuse.
For more detailed information on ethical considerations for this model, please see the Model Card++ Bias, Explainability, Safety & Security, and Privacy Subcards.
Please make sure you have proper rights and permissions for all input image and video content; if image or video includes people, personal health information, or intellectual property, the image or video generated will not blur or maintain proportions of image subjects included.
Please report model quality, risk, security vulnerabilities or concerns here.
NV-Generate-MR-Brain is a three-dimensional (3D) latent diffusion model designed to generate high-quality synthetic brain magnetic resonance imaging (MRI) images, achieving the highest resolution and best FID scores among comparable models. This model is specialized for brain MRI with support for multiple modalities: T1, Fluid Attenuated Inversion Recovery (FLAIR), T2, and Susceptibility Weighted Imaging (SWI).
Compared to the previous NV-Generate-MR release, the key differences are:
The model excels at data augmentation and at generating realistic medical imaging data to supplement datasets limited by privacy concerns or the rarity of certain conditions. It can also significantly enhance the performance of other medical imaging AI models by generating diverse, realistic training data.
This model is ready for commercial use.
You can try this model via NVIDIA MAISI demo:
👉 [https://huggingface.co/spaces/nvidia/nv-generate)
The current demo showcases only a subset of the model’s capabilities. For full functionality and advanced usage, please refer to the GitHub repository.
Training and inference code are in: https://github.com/NVIDIA-Medtech/NV-Generate-CTMR/tree/main.
Use of this model is governed by the NVIDIA Open Model License. Additional Information: Apache 2.0 License.
This project was conducted by NVIDIA in collaboration with the University of Zurich, Istanbul Medipol University, and Forithmus.
Global
Medical researchers, AI developers, and healthcare institutions would be expected to use this model for:
It is not a clinically validated medical device and should not be used for clinical diagnostic purposes.
Huggingface: 03/16/2026 (GTC San Jose 2026) via https://huggingface.co/nvidia/NV-Generate-MR-brain
[1] MAISI V2: https://arxiv.org/abs/2508.05772
[2] Guo, Pengfei, et al. "MAISI: Medical AI for Synthetic Imaging." arXiv preprint arXiv:2409.11169. 2024. https://arxiv.org/abs/2409.11169
[3] Rombach, Robin, et al. "High-resolution image synthesis with latent diffusion models." Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2022. https://openaccess.thecvf.com/content/CVPR2022/papers/Rombach_High-Resolution_Image_Synthesis_With_Latent_Diffusion_Models_CVPR_2022_paper.pdf
[4] Lvmin Zhang, Anyi Rao, Maneesh Agrawala; "Adding Conditional Control to Text-to-Image Diffusion Models." Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), 2023, pp. 3836-3847. https://openaccess.thecvf.com/content/ICCV2023/papers/Zhang_Adding_Conditional_Control_to_Text-to-Image_Diffusion_Models_ICCV_2023_paper.pdf
Architecture Type: Diffusion Model
Network Architecture: 3D UNet + attention blocks (latent diffusion)
Task: Generation (Synthetic MRI Image)
Number of model parameters: 240M
Cumulative Compute: 4.375 x 10^22
Estimated Energy and Emissions for Model Training: 24,085kWh
Input Type(s): Integer, Array
Input Format(s): Integer values, Float arrays
Input Parameters: Number of Samples (1D), Modality (1D), Output Size (1D), and Spacing (1D)
Other Properties Related to Input: Supports controllable synthetic brain MRI generation with modality selection, customizable output dimensions, and configurable voxel spacing.
Output Type(s): Image
Output Format: Neuroimaging Informatics Technology Initiative (NIfTI)
Output Parameters: Three-Dimensional (3D)
Other Properties Related to Output: Synthetic brain MRI images in the specified modality (T1, FLAIR, T2, or SWI). Output dimensions and spacing are configurable within supported ranges.
Our AI models are designed and/or optimized to run on NVIDIA GPU-accelerated systems. By leveraging NVIDIA's hardware (e.g. GPU cores) and software frameworks (e.g., CUDA libraries), the model achieves faster training and inference times compared to CPU-only solutions.
Runtime Engine(s):
Supported Hardware Microarchitecture Compatibility:
Supported Operating System(s):
The integration of foundation and fine-tuned models into AI systems requires additional testing using use-case-specific data to ensure safe and effective deployment. Following the V-model methodology, iterative testing and validation at both unit and system levels are essential to mitigate risks, meet technical and functional requirements, and ensure compliance with safety and ethical standards before deployment.
0.1 - Initial release version for synthetic brain MRI image generation
The model was trained on the MR-Rate brain MRI dataset covering the supported modalities (T1, FLAIR, T2, and SWI). Data from multiple scanner types were processed to create high-quality 3D MRI volumes with corresponding anatomical annotations. The data processing pipeline ensured consistent voxel spacing, standardized orientations, and validated anatomical segmentations.
Data Modality:
Image Training Data Size:
Data Collection Method by dataset:
Labeling Method by dataset:
Properties: Approximately 28,000 MRI scans of various types (T1, FLAIR, T2, and SWI).
Data Modality:
Image Training Data Size:
Data Collection Method by dataset:
Labeling Method by dataset:
Properties: Approximately 8,000 MRI scans of various types (T1, FLAIR, T2, and SWI).
Data Modality:
Image Training Data Size:
Data Collection Method by dataset:
Labeling Method by dataset:
Properties: Approximately 4,000 MRI scans of various types (T1, FLAIR, T2, and SWI).
Acceleration Engine: PyTorch
Test Hardware:
NVIDIA believes Trustworthy AI is a shared responsibility and we have established policies and practices to enable development for a wide array of AI applications. When downloaded or used in accordance with our terms of service, developers should work with their internal model team to ensure this model meets requirements for the relevant industry and use case and addresses unforeseen product misuse.
For more detailed information on ethical considerations for this model, please see the Model Card++ Bias, Explainability, Safety & Security, and Privacy Subcards.
Please make sure you have proper rights and permissions for all input image and video content; if image or video includes people, personal health information, or intellectual property, the image or video generated will not blur or maintain proportions of image subjects included.
Please report model quality, risk, security vulnerabilities or concerns here.
