Back

High-fidelity bioimage restoration via adversarial learning

Munoz-Barrutia, A.; Lachowski, D.; Rey-Paniagua, G.

2026-01-24 bioengineering
10.64898/2026.01.22.701051 bioRxiv
Show abstract

Live-cell microscopy restoration is constrained by a trade-off between inference latency and texture preservation. While diffusion models provide high textural fidelity, the computational cost of iterative sampling currently limits their use in low-latency instrument feedback loops. Here, we present NAFNet GAN, a restoration framework that couples an activation-free backbone with a perceptual adversarial objective to enable high-throughput analysis. Unlike diffusion architectures, NAFNet GAN achieves an inference latency of[~] 110 ms for 1024 x 1024 inputs, potentially suitable for real-time instrument feedback loops. Across eight datasets ranging from STED nanoscopy to histopathology, the method achieves the lowest Learned Perceptual Image Patch Similarity (LPIPS) scores in 7 of 8 benchmarks while preserving structural coherence (e.g., MS-SSIM > 0.968 in Cryo-EM), which facilitates reliable downstream analysis. Supported by performance benchmarks in the AI4Life Denoising Challenge, NAFNet GAN restores structural features from low-photon-budget acquisitions, maintaining the temporal resolution required for dynamic live-cell workflows.

Matching journals

The top 2 journals account for 50% of the predicted probability mass.

1
Nature Methods
336 papers in training set
Top 0.1%
43.8%
2
Nature Communications
4913 papers in training set
Top 24%
7.6%
50% of probability mass above
3
Advanced Science
249 papers in training set
Top 3%
5.1%
4
ACS Photonics
13 papers in training set
Top 0.1%
3.9%
5
Science
429 papers in training set
Top 8%
3.8%
6
Cell Systems
167 papers in training set
Top 5%
2.9%
7
Nature Biotechnology
147 papers in training set
Top 3%
2.9%
8
Light: Science & Applications
16 papers in training set
Top 0.2%
2.5%
9
Nature Medicine
117 papers in training set
Top 1%
2.5%
10
ACS Nano
99 papers in training set
Top 2%
2.5%
11
Nature Biomedical Engineering
42 papers in training set
Top 0.7%
1.9%
12
Proceedings of the National Academy of Sciences
2130 papers in training set
Top 31%
1.8%
13
Science Advances
1098 papers in training set
Top 20%
1.6%
14
Nature Machine Intelligence
61 papers in training set
Top 2%
1.6%
15
Communications Biology
886 papers in training set
Top 10%
1.6%
16
Nature Cell Biology
99 papers in training set
Top 4%
1.2%
17
Nature Neuroscience
216 papers in training set
Top 5%
1.2%
18
Nano Letters
63 papers in training set
Top 3%
0.8%
19
Science Translational Medicine
111 papers in training set
Top 5%
0.8%
20
Nature
575 papers in training set
Top 16%
0.7%
21
IEEE Transactions on Medical Imaging
18 papers in training set
Top 0.6%
0.7%
22
Scientific Reports
3102 papers in training set
Top 78%
0.7%
23
Neuron
282 papers in training set
Top 10%
0.5%