Research map / 3D vision / Depth and stereo
SparseSSP: 3D Subcellular Structure Prediction from Sparse-View Transmitted Light Images
Jintu Zheng, Yi Ding, Qizhe Liu, Yuehui Chen, Yi Cao, Ying Hu, Zenan Wang
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 |
|---|---|---|---|
| use prefix interpolation instead of postfix (ablation) | RepMode 3-to-2D · AllenCell (r=2) | R^2: .4589 -> .4616 | Improved |
| use prefix interpolation instead of postfix on 3D topology (ablation) | RepMode 3D · AllenCell (r=2) | R^2: .4193 -> .4453 | Improved |
| use prefix interpolation instead of none (direct 2D output channels) (ablation) | RepMode 3-to-2D · AllenCell (r=2) | R^2: .4032 -> .4616 | Improved |
| use hybrid 3-to-2D topology instead of pure 3D (ablation) | Tgnet · AllenCell (r=2) | R^2: .4459 -> .4693 | Improved |
| use pure 2D topology instead of hybrid (ablation) | Tgnet · AllenCell (r=2) | R^2: .4693 -> .4034 | Hurt |
| project features in 3D space instead of 2D before channel arrangement (ablation) | DoDNet 3-to-2D · AllenCell (r=8) | R^2: .3969 -> .4236 | Improved |
| train pure 2D topology twice as long (7000 to 14000 epochs) (ablation) | DoDNet 2D · AllenCell (r=2) | R^2: .3883 -> .4176 | Improved |
| train 3-to-2D and 3D topologies beyond default epochs (ablation) | 3D and 3-to-2D topologies · AllenCell | R^2 | No clear effect |
| replace ResNet18 with Swin/ViT-Large backbone in 2-to-3D (ablation) | TSNs 2-to-3D · AllenCell (r=2) | R^2: .4323 -> .4559 | Improved |
| replace ResNet18 with DINOv2-Base backbone in 2-to-3D (ablation) | TSNs 2-to-3D · AllenCell (r=2) | R^2: .4323 -> .4124 | Hurt |
| use 3-to-2D hybrid instead of 3D for compute | RepMode · AllenCell (r=2) | MACs: 66.29G -> 43.47G | Improved |
| increase sparsity ratio to 8 (87.5% fewer imaging slices) | Tgnet-3-to-2D · AllenCell vs dense Multi-Net | R^2 relative improvement (%): 8.2084 (r=2) -> -1.037 (r=8) | Mixed |
Closest papers by meaning
- SR-CACO-2: A Dataset for Confocal Fluorescence Microscopy Image Super-Resolution (NeurIPS 2024 · similarity 0.85)
- scSplit: Bringing Severity Cognizance to Image Decomposition in Fluorescence Microscopy (NeurIPS 2025 · similarity 0.85)
- From Pixels to Views: Learning Angular-Aware and Physics-Consistent Representations for Light Field Microscopy (NeurIPS 2025 · similarity 0.84)
- Volume Tells: Dual Cycle-Consistent Diffusion for 3D Fluorescence Microscopy De-noising and Super-Resolution (CVPR 2025 · similarity 0.84)
- V2V3D: View-to-View Denoised 3D Reconstruction for Light Field Microscopy (CVPR 2025 · similarity 0.84)
- SCCS: Deep Neural Spectral Clustering for Self-Supervised Subcellular Structure Segmentation (AAAI 2025 · 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.
