Learning Transferable Self-Supervised Priors for Super-Resolution Reconstruction in Structured Illumination Microscopy
Abstract
Structured illumination microscopy (SIM) extends the optical passband, and reconstruction of detail beyond it depends on prior knowledge. Hand-designed regularizers depend on how well their structural assumptions match the specimen; learned priors can be sensitive to changes in imaging conditions and specimen structure. We introduce SIMAdapter, which pretrains a network that predicts the emitter and the point-spread function (PSF) by self-supervision on 23,237 raw SIM stacks from BioSR, BioTISR, and simulations spanning different PSFs and specimen structures, then adapts it to a single unlabeled target stack. Adaptation refines the network against a differentiable image-formation model, with the light pattern calibrated from that stack. Both stages take their supervision from the raw measurements and need no paired high-resolution reference. On two held-out synthetic domains, SIMAdapter reaches a mean emitter normalized root-mean-square error of 0.156, compared with 0.403 for Sparse-SIM. The same adaptation started from a network pretrained on BioSR alone is less accurate in both domains. In three experimental case studies, adaptation reduces flanking artifacts and yields more distinct profiles across filament pairs, mitochondrial boundaries, and calibration lines. A single pretrained network can thus be reused across SIM measurements, with each reconstruction refined against its own raw data.
Related
- Structured Illumination improves Multislice Ptychography
- Coherent Structured Illumination Microscopy with Enhanced Suppression of Coherent Defocus Artifacts
- Transferable reconstruction of nonlinear network dynamics from sentinel nodes
- Probe-aware self-supervised neural network for X-ray holographic reconstruction
- Real-Time Atomic-Resolution Electron Phase Imaging without Probe Calibration via Ptychography-Supervised Learning
BibTeX
@misc{weng2026learning,
title = {{Learning Transferable Self-Supervised Priors for Super-Resolution Reconstruction in Structured Illumination Microscopy}},
author = {Tong-Tian Weng and Ze-Hao Wang and Qi Wang and Xi-Hua Wang and Xiang-Dong Chen and Fang-Wen Sun},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.16538},
eprint = {2609.16538},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.16538},
}
RIS
TY - UNPB
TI - Learning Transferable Self-Supervised Priors for Super-Resolution Reconstruction in Structured Illumination Microscopy
AU - Tong-Tian Weng
AU - Ze-Hao Wang
AU - Qi Wang
AU - Xi-Hua Wang
AU - Xiang-Dong Chen
AU - Fang-Wen Sun
PY - 2026
DA - 2026/09/15
PB - arXiv
DO - 10.48550/arxiv.2609.16538
UR - https://arxiv.org/abs/2609.16538
AB - Structured illumination microscopy (SIM) extends the optical passband, and reconstruction of detail beyond it depends on prior knowledge. Hand-designed regularizers depend on how well their structural assumptions match the specimen; learned priors can be sensitive to changes in imaging conditions and specimen structure. We introduce SIMAdapter, which pretrains a network that predicts the emitter and the point-spread function (PSF) by self-supervision on 23,237 raw SIM stacks from BioSR, BioTISR, and simulations spanning different PSFs and specimen structures, then adapts it to a single unlabeled target stack. Adaptation refines the network against a differentiable image-formation model, with the light pattern calibrated from that stack. Both stages take their supervision from the raw measurements and need no paired high-resolution reference. On two held-out synthetic domains, SIMAdapter reaches a mean emitter normalized root-mean-square error of 0.156, compared with 0.403 for Sparse-SIM. The same adaptation started from a network pretrained on BioSR alone is less accurate in both domains. In three experimental case studies, adaptation reduces flanking artifacts and yields more distinct profiles across filament pairs, mitochondrial boundaries, and calibration lines. A single pretrained network can thus be reused across SIM measurements, with each reconstruction refined against its own raw data.
ER -