Physics-informed single-frame end-to-end learning for denoising and background removal in fluorescence imaging.

Physics-informed deep learning achieves single-frame denoising and background removal in fluorescence microscopy, enabling high-quality imaging across diverse biological samples.

• Why it matters: Fluorescence microscopy faces challenges from low signal-to-noise ratios and out-of-focus backgrounds that obscure structural details, limiting imaging quality and speed.
• What they did: The authors developed a physics-informed end-to-end neural network trained entirely on simulated data using a microscope-parameterized forward model, allowing deployment without retraining on new specimens.
• The result: The framework demonstrated robust denoising in wide-field and confocal microscopy, enabled superresolution imaging with tens of frames, and can extend to background removal in thick samples, improving rapid, data-efficient fluorescence imaging.