Generating synthetic medical images with AI traditionally requires programming knowledge, installing specialized software, and access to powerful GPU hardware.
With Nalvera.AI, you can generate synthetic 3D CT and brain MRI images directly in your browser, without coding, installing software, or owning a local GPU.
Using MAISI-v2, you can create synthetic CT volumes with optional anatomical segmentation masks or generate brain MRI images in different contrasts.
And unlike many medical imaging AI workflows, you don't need to upload an existing scan to get started.
What Is Synthetic Medical Image Generation?
Synthetic medical images are artificially generated images that resemble medical scans but are created by AI rather than acquired directly from a patient.
Here, medical image generation means a 3D diffusion model, NVIDIA's MAISI-v2, making a whole CT or brain MRI volume from scratch.
Researchers can explore synthetic imaging for various purposes, including:
- Dataset augmentation: Exploring additional image variation for research datasets.
- Software testing: Testing medical imaging viewers, processing pipelines, and analysis tools.
- Algorithm development: Experimenting with synthetic images and their associated anatomical masks.
- Education and demonstrations: Demonstrating medical imaging workflows without using actual patient scans.
Step 1: Choose the MAISI-v2 Model
Sign in to Nalvera.AI, open Submit Task, go to Generation and select MAISI-v2 from the available AI models.
Step 2: Choose Between Synthetic CT and Brain MRI Generation
MAISI-v2 supports two main generation tasks.
Synthetic CT Generation
Generate synthetic 3D CT images for research purposes. You can choose between:
- Image only: Generate a synthetic CT volume.
- Image + mask: Generate a synthetic CT volume together with a corresponding anatomical segmentation mask.
Depending on the selected settings, you can configure the body region, image dimensions, and voxel spacing, and with Image + mask, optionally a tumour and how prominent it appears.
The optional segmentation mask labels the anatomical structures in the generated CT volume. This can be useful when experimenting with segmentation algorithms or testing workflows that require both an image and its corresponding labels.
Synthetic CT with Segmentation Masks
With Image + mask, the mask comes from a real scan of the body region you choose, and MAISI-v2 generates the CT to match it. One run makes up to 64 volumes, and its start seed makes the run repeatable.
Synthetic Brain MRI Generation
MAISI-v2 can also generate synthetic 3D brain MRI images. Supported MRI contrasts include:
- T1-weighted
- T2-weighted
- FLAIR
- SWI
You can generate an image in the contrast you select, or use cross-contrast generation to create several contrasts in one task; each contrast is a separate synthetic brain, so the images do not line up with each other.
Step 3: Configure and Generate Your Medical Image
Set the output image dimensions and voxel spacing, review the displayed processing cost, and click Generate Volume. You can follow the processing status in the Queue.
Step 4: Explore and Download Your Synthetic Images
Once generation is complete, open Results in Nalvera.AI.
You can inspect the synthetic medical image volumes directly in the browser using the viewer. For synthetic CT generation with masks, you can also review the corresponding anatomical labels.
The generated images can be downloaded in, e.g., NIfTI (.nii.gz) format for further research or use in compatible medical imaging software.
Important Considerations for Research
Synthetic medical images can be valuable for research, but they also have limitations.
The quality of generated images depends on the model and selected parameters. In particular, unusual image dimensions, voxel spacing, or fields of view can produce unrealistic results. Nalvera.AI guides you toward recommended settings: CT starts from a preset for each body region, settings the model does not support are disabled, and the field of view is checked against what the model was trained on.
When you share or publish synthetic images, label them as AI-generated.
Frequently Asked Questions About Synthetic Medical Image Generation
Do I need coding skills or a GPU to generate synthetic medical images?
No. Nalvera.AI runs MAISI-v2 in your browser on managed cloud GPUs, so you don't need to write code, install the model, or own an NVIDIA GPU. A laptop, desktop computer, or even a smartphone with an internet connection is enough.
Do I need to upload a CT or MRI scan first?
No. MAISI-v2 generates synthetic medical images without requiring an existing input scan. To run AI models on a scan of your own instead, see How to Run Medical Imaging AI Models Without Coding or a Local GPU.
Can I generate segmentation masks together with synthetic CT images?
Yes. MAISI-v2 supports synthetic CT generation with an optional paired anatomical segmentation mask.
Can I generate different MRI contrasts?
Yes. MAISI-v2 supports synthetic brain MRI generation in T1-weighted, T2-weighted, FLAIR, and SWI contrasts, one at a time or several in one task with cross-contrast generation, each a separate synthetic brain.
Can I download the generated medical images?
Yes. Generated volumes can be downloaded in NIfTI format for further research and analysis.
Get Started With Synthetic Medical Image Generation
You don't need to build a generative AI pipeline, configure coding environments, or own a powerful GPU to explore synthetic medical image generation.
With Nalvera.AI, you can select MAISI-v2, configure your synthetic CT or brain MRI, generate the image using cloud-based AI, and explore the results directly in your browser.
Explore MAISI-v2 on Nalvera.AIResearch use only. Nalvera.AI is not a medical device, and AI-generated outputs are not intended for clinical diagnosis or treatment.