Research map / Image generation / Super-resolution
SkelEM: Explicit Decoupling of Topology and Details for Self-supervised Axial Super-Resolution in Volume Microscopy
Bohao Chen, Yanchao Zhang, Yanan Lv, Chenxun Deng, Hua Han, Xi Chen
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add diffusion refiner on top of skeleton (ablation) | SkelEM · EPFL | LPIPS XZ: 0.5171 -> 0.3213 | Improved |
| add skeleton prior to the refiner (ablation) | SkelEM · EPFL | 3D-PSNR XY: 22.43 -> 25.56 | Improved |
| apply Ladapt to a skeleton-free refiner (ablation) | Refiner only · EPFL | 3D-PSNR XY: 22.43 -> 20.26 | Hurt |
| retrain skeleton on pseudo-HR volume instead of pretrained RIFE (ablation) | SkelEM · EPFL | 3D-PSNR XY: 25.19 -> 25.56 | Improved |
| use learned skeleton instead of bilinear interpolation (ablation) | SkelEM · EPFL | 3D-PSNR XY: 22.55 -> 25.56 | Improved |
| add bidirectional self-alignment Ladapt (ablation) | SkelEM · EPFL | LPIPS XZ: 0.3976 -> 0.3213 | Improved |
| Ladapt vs Sparse3Diff adaptation (ablation) | SkelEM · EPFL | LPIPS XZ: 0.3640 -> 0.3213 | Improved |
| inject residual prior at S=1 (lambda 0 -> 1) (ablation) | SkelEM · EPFL | 3D-PSNR XY: 23.92 -> 25.93 | Improved |
| inject residual prior at S=3 (lambda 0 -> 1) (ablation) | SkelEM · EPFL | 3D-PSNR XY: 25.51 -> 25.56 | No clear effect |
| increase steps from 3 to 5 (ablation) | SkelEM · EPFL | 3D-PSNR XY: 25.56 -> 25.31 | Hurt |
| use 3 steps instead of 1 (ablation) | SkelEM · EPFL | LPIPS XZ: 0.3969 -> 0.3213 | Improved |
| SkelEM vs D2R-AENet on fidelity | SkelEM · FIB25 | 3D-PSNR XY: 27.62 -> 26.24 | Hurt |
Closest papers by meaning
- VEMamba: Efficient Isotropic Reconstruction of Volume Electron Microscopy with Axial-Lateral Consistent Mamba (CVPR 2026 · similarity 0.87)
- EMGauss: Continuous Slice-to-3D Reconstruction via Dynamic Gaussian Modeling in Volume Electron Microscopy (CVPR 2026 · similarity 0.87)
- Volume Tells: Dual Cycle-Consistent Diffusion for 3D Fluorescence Microscopy De-noising and Super-Resolution (CVPR 2025 · similarity 0.86)
- A Gaussian Filter-Based 3D Registration Method for Series Section Electron Microscopy (AAAI 2025 · similarity 0.85)
- MicroDiffusion: Implicit Representation-Guided Diffusion for 3D Reconstruction from Limited 2D Microscopy Projections (CVPR 2024 · similarity 0.84)
- Learning Large-Factor EM Image Super-Resolution with Generative Priors (CVPR 2024 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
