Research map / Label-efficient and robust learning / Self-supervised and contrastive
Coordinate-based Speed of Sound Recovery for Aberration-Corrected Photoacoustic Computed Tomography
Tianao Li, Manxiu Cui, Cheng Ma, Emma Alexander
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 neural field SOS representation instead of APACT wavefront search (ablation) | Ours (NF) · Numerical phantoms (5 SOS maps, mean) | SOS PSNR: 17.74 -> 22.29 | Improved |
| use neural field SOS representation instead of APACT wavefront search (ablation) | Ours (NF) · Numerical phantoms (5 SOS maps, mean) | reconstruction time: 23 min -> 40.3 s | Improved |
| use neural field instead of pixel grid SOS (ablation) | Ours · Numerical phantoms (5 SOS maps, mean) | SOS PSNR: 21.26 -> 22.29 | Improved |
| use neural field instead of pixel grid SOS (ablation) | Ours · Numerical phantoms (5 SOS maps, mean) | reconstruction time: 113.3 s -> 40.3 s | Improved |
| use neural field instead of pixel grid SOS (ablation) | Ours · Numerical phantoms (5 SOS maps, mean) | SOS SSIM: 0.903 -> 0.931 | Improved |
| use ground truth SOS (oracle) instead of estimated SOS (ablation) | Ours (NF) · Numerical phantoms (5 SOS maps, mean) | IP PSNR: 25.08 -> 25.61 | Improved |
| use multi-channel deconvolution with estimated SOS instead of Dual-SOS DAS (ablation) | Ours (NF) · Numerical phantoms (5 SOS maps, mean) | IP PSNR: 24.42 -> 25.08 | Improved |
| increase number of delays up to 16 (ablation) | Ours · Numerical phantoms (5 SOS maps, mean) | IP and SOS quality: gains diminish beyond 16 delays; more compute time | Improved |
| add TV regularization to pixel grid SOS (ablation) | Ours (PG) · Numerical phantoms (5 SOS maps, mean) | SOS quality: IP quality little changed | Improved |
| increase TV weight above 1e-4 (ablation) | Ours (PG) · Numerical phantoms (5 SOS maps, mean) | IP quality: optimum lambda 1x10^-4 | Hurt |
| enlarge NF network from 0 to hidden layers / more parameters (131 to 264,707) (ablation) | Ours (NF) · Numerical phantoms (5 SOS maps, mean) | IP and SOS quality: noisier SOS, worse IP | Hurt |
Closest papers by meaning
- Reliability-Aware Restoration Framework for 4D Spectral Photoacoustic Data (TPAMI 2023 · similarity 0.82)
- UltraGauss: Ultrafast Gaussian Reconstruction of 3D Ultrasound Volumes (ICLR 2026 · similarity 0.82)
- SSC-VAE: Structured Sparse Coding Based Variational Autoencoder for Detail Preserved Image Reconstruction (AAAI 2025 · similarity 0.81)
- Efficient Noise Calculation in Deep Learning-based MRI Reconstructions (ICML 2025 · similarity 0.80)
- Reconstructing the Vocal Tract with Differentiable Acoustic Simulation (NeurIPS 2026 · similarity 0.79)
- RFWave: Multi-band Rectified Flow for Audio Waveform Reconstruction (ICLR 2025 · similarity 0.79)
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.
