Research map / Detection and segmentation / CNN features
A Segmentation-aware Deep Fusion Network for Compressed Sensing MRI
Zhiwen Fan, Liyan Sun, Xinghao Ding, Yue Huang, Congbo Cai, John Paisley
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 segmentation-network features via multilayer aggregation and fusion (versus same architecture using only reconstruction features) (ablation) | SADFN5 · MRBrainS CS-MRI, 30% 1D Cartesian | PSNR and SSIM: around 1dB in PSNR and 0.03 in SSIM | Improved |
| add segmentation fusion (SADFN5 vs SADFN5-WOS) (ablation) | SADFN5 · MRBrainS GM segmentation of reconstructions | Dice (%): 83.85 -> 85.76 | Improved |
| add segmentation fusion (SADFN5 vs SADFN5-WOS) (ablation) | SADFN5 · MRBrainS WM segmentation of reconstructions | Dice (%): 75.84 -> 81.29 | Improved |
| add segmentation fusion (SADFN5 vs SADFN5-WOS) (ablation) | SADFN5 · MRBrainS CSF segmentation of reconstructions | Dice (%): 79.25 -> 80.08 | Improved |
| fine-tune with fused segmentation features rather than cascading segmentation loss (Liu) | SADFN5 · MRBrainS WM segmentation of reconstructions | Dice (%): 76.30 -> 81.29 | Improved |
| cascade pre-trained segmentation network after reconstruction (Liu) instead of plain Pre-RecNet5 | Pre-RecNet5 · MRBrainS CS-MRI reconstruction | PSNR and SSIM | No clear effect |
| enlarge network without segmentation information (SADFN5-WOS vs Pre-RecNet5) (ablation) | Pre-RecNet5 · MRBrainS CS-MRI reconstruction | PSNR/SSIM | No clear effect |
| add segmentation features (SADFN5 vs SADFN5-WOS) (ablation) | SADFN5 · MRBrainS CS-MRI, 1 to 5 blocks | PSNR/SSIM | Improved |
| use 20% random under-sampling mask | SADFN5 · MRBrainS CS-MRI, 20% Random mask | PSNR/SSIM | Improved |
| add segmentation fusion (SADFN5 vs Pre-RecNet5) | SADFN5 · MRBrainS GM segmentation of reconstructions | Hausdorff distance (lower better): 1.795 -> 1.690 | Improved |
| add segmentation fusion (SADFN5 vs Pre-RecNet5) | SADFN5 · MRBrainS WM segmentation of reconstructions | Dice (%): 75.16 -> 81.29 | Improved |
| adding segmentation fusion increases runtime | SADFN5 · MRBrainS runtime | seconds: 0.03 -> 0.06 | Hurt |
Closest papers by meaning
- Compressed Sensing MRI Using a Recursive Dilated Network (AAAI 2018 · similarity 0.88)
- A Unified Model for Compressed Sensing MRI Across Undersampling Patterns (CVPR 2025 · similarity 0.86)
- UFC-Net: Unrolling Fixed-point Continuous Network for Deep Compressive Sensing (CVPR 2024 · similarity 0.85)
- Multi-Cross Sampling and Frequency-Division Reconstruction for Image Compressed Sensing (AAAI 2024 · similarity 0.84)
- CPP-Net: Embracing Multi-Scale Feature Fusion into Deep Unfolding CP-PPA Network for Compressive Sensing (CVPR 2024 · similarity 0.84)
- Efficient Noise Calculation in Deep Learning-based MRI Reconstructions (ICML 2025 · similarity 0.83)
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.
