SR-Wiki Represents WeisongZhao-lab’s imaging-tool kit implementation. We develop biomedical imaging and analysis solutions across diverse modalities.

This organization hosts code resources developed by members of the Weisong Zhao Lab, covering super-resolution reconstruction, denoising, and bioimage analysis, implemented in MATLAB, Python, and ImageJ plugins.

• 2026.02: COMET released. Cortex-wide Observational Miniature Epifluorescence Technique
• 2026.01: aSN2N released; see also SN2N. Adaptive Self-inspired Noise2Noise learning for denoising
• 2025.11: MGAN-SIM released. Organelle-aware Markovian Generative Adversarial Network for single-frame SIM reconstruction
• 2025.07: FLAMEm released. FLuctuation-based high-order super-resolution Acoustic MicroscopE

• Sparse deconvolution is a universal post-processing framework for microscopy that enhances SNR, SBR, contrast, and spatial resolution. By leveraging PSF-induced sparsity and continuity, it employs a multi-constrained deconvolution algorithm to extend resolution up to twofold beyond the physical limits of existing microscopy systems under low-SNR conditions. This method has been tested on various types of Confocal microscopy & STED, Wide-field microscopy & TIRF microscopy, Light-sheet microscopy, Multi-photon microscopy, OCT, SIM, X-ray, MRI, (micro)CT, ultrasound, SEM/TEM, Cryo-EM/ET, photoacoustic imaging, and etc., feasible for single-slice, time-lapse, and volumetric datasets. MATLAB & Python versions@Sparse deconvolution & sparse-deconv-py. Paper @Nature Biotechnology 2022.

• Sparse deconvolution PRIME an optimized variant designed specifically for ultra-low SNR conditions.

For laser-excited fluorescence

• SACD (super-resolution method based on auto-correlation with two-step deconvolution) is a fast fluorescence fluctuation-based super-resolution method that requires ~50× fewer frames. It achieves twofold lateral and axial resolution enhancement with as few as ~20 frames, without additional optical hardware. SACD can be directly integrated into commercial or custom microscopy systems as a flexible add-on. MATLAB & ImageJ-plugin versions @SACDm & SACDj. Paper @Nature Photonics 2023.

For ultrasound

• FLAME (FLuctuation-based high-order super-resolution Acoustic MicroscopE) a high-efficiency ultrasound super-resolution framework for both contrast-free and contrast-enhanced imaging. It reduces data requirements by ~50× and achieves ~4–10× spatial resolution improvement using only 30–120 frames. MATLAB version @FLAMEm. Preprint @bioRxiv 2025.

For reaction-driven fluorescence

• RIED. Electrochemiluminescence (ECL), chemiluminescence (CL), and bioluminescence (BL) super-resolution imaging. Preprint @bioRxiv 2026.

• MGAN. (Markovian Generative Adversarial Network for single-frame SIM) is a single-frame SIM reconstruction framework based on a conditional GAN with an organelle-aware, multi-scale Markovian discriminator. This is also an cGAN discriminator architecture framework designed particularly for fluorescence image. Python version @MGAN-SIM

• SN2N (Self-inspired Noise2Noise) denoising framework that achieves performance competitive with supervised methods while eliminating the need for paired training data or clean ground truth. It has been validated across diverse imaging modalities, including various types of Confocal microscopy & STED, Wide-field microscopy & TIRF microscopy, Light-sheet microscopy, Multi-photon microscopy, OCT, SIM, X-ray, MRI, (micro)CT, ultrasound, SEM/TEM, Cryo-EM/ET, photoacoustic imaging, and etc., feasible for single-slice, time-lapse, and volumetric datasets. Python & ImageJ-plugin versions @SN2N & SN2Nj. Preprint @Nature Methods 2024.

• aSN2N (adaptive SN2N) is a preprocessing strategy designed to stabilize training and suppress background artifacts in SN2N and other self-supervised denoising frameworks. Python version @aSN2N. Preprint link: PhotoniX Life 2026.

• COMET (Cortex-wide Observational Miniature Epifluorescence Technique) enables single-cell resolution imaging of large-scale cortical neuron populations in freely moving mice. This repository includes 3D-printed components, PCBs, DAQ-QT software, and image analysis pipelines, allowing users to build the COMET system from scratch.

• Bio-Correction is a framework for quantitative correction of super-resolution microscopy measurements, enabling more accurate recovery of biological structure properties.
• Adaptive median filter, A median filter with adaptive threshold to avoid blurring effects for removing hot pixels (abnormal black/white pixels).
• t-varianceJ, T-axial variance (with rolling average) calculation will highlight the regions that have calcium signal transients.
• Palette.ui, Multi-color imaging tool to merge/composite frames.

• img2vid, A light weight framework for making exsiting images to videos.
• ImagePy, An open source image processing framework (A Pythonic ImageJ).
• AdaptiveOptics.simulation, A light weight simulation framework for adaptive optics in microscopy.

This organization is maintained by Weisong Zhao and members of the lab. For inquiries, please contact Weisong Zhao or other team members. If you encounter a bug or have questions about a specific project, please open an issue in the corresponding repository—this helps others as well.

Latest updates from the Lab are available on Weisong Zhao's X account or Bluesky account.