Improving 3D Imaging with Pre-Trained Perpendicular 2D Diffusion Models
About
Diffusion models have become a popular approach for image generation and reconstruction due to their numerous advantages. However, most diffusion-based inverse problem-solving methods only deal with 2D images, and even recently published 3D methods do not fully exploit the 3D distribution prior. To address this, we propose a novel approach using two perpendicular pre-trained 2D diffusion models to solve the 3D inverse problem. By modeling the 3D data distribution as a product of 2D distributions sliced in different directions, our method effectively addresses the curse of dimensionality. Our experimental results demonstrate that our method is highly effective for 3D medical image reconstruction tasks, including MRI Z-axis super-resolution, compressed sensing MRI, and sparse-view CT. Our method can generate high-quality voxel volumes suitable for medical applications.
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| Sparse-View CT Reconstruction | LIDC Coronal View (test) | PSNR28.12 | 24 | |
| Sparse-View CT Reconstruction | LIDC Sagittal View (test) | PSNR27.66 | 24 | |
| Sparse-View CT Reconstruction | LIDC | PSNR27.51 | 24 | |
| Sparse-View Computed Tomography (SVCT) | AAPM low-dose CT 2016 (Patient L506) | Axial PSNR37.59 | 11 | |
| Limited-Angle CT Reconstruction | LIDC Axial | PSNR14.44 | 8 | |
| Limited-Angle CT Reconstruction | LIDC Sagittal | PSNR14.06 | 8 | |
| Limited-Angle CT Reconstruction | LIDC Coronal | PSNR14.54 | 8 | |
| LACT reconstruction | LACT Subject L506 1mm 1.0 (test) | PSNR (Axial)30.95 | 4 |