Abstract:Optic deconvolution in light microscopy (LM) refers to recovering the object details from images, revealing the ground truth of samples. Traditional explicit methods in LM rely on the point spread function (PSF) during image acquisition. Yet, these approaches often fall short due to inaccurate PSF models and noise artifacts, hampering the overall restoration quality. In this paper, we approached the optic deconvolution as an inverse problem. Motivated by the nonstandard-form compression scheme introduced by Beylkin, Coifman, and Rokhlin (BCR), we proposed an innovative physics-informed neural network Multi-Stage Residual-BCR Net (m-rBCR) to approximate the optic deconvolution. We validated the m-rBCR model on four microscopy datasets - two simulated microscopy datasets from ImageNet and BioSR, real dSTORM microscopy images, and real widefield microscopy images. In contrast to the explicit deconvolution methods (e.g. Richardson-Lucy) and other state-of-the-art NN models (U-Net, DDPM, CARE, DnCNN, ESRGAN, RCAN, Noise2Noise, MPRNet, and MIMO-U-Net), the m-rBCR model demonstrates superior performance to other candidates by PSNR and SSIM in two real microscopy datasets and the simulated BioSR dataset. In the simulated ImageNet dataset, m-rBCR ranks the second-best place (right after MIMO-U-Net). With the backbone from the optical physics, m-rBCR exploits the trainable parameters with better performances (from ~30 times fewer than the benchmark MIMO-U-Net to ~210 times than ESRGAN). This enables m-rBCR to achieve a shorter runtime (from ~3 times faster than MIMO-U-Net to ~300 times faster than DDPM). To summarize, by leveraging physics constraints our model reduced potentially redundant parameters significantly in expertise-oriented NN candidates and achieved high efficiency with superior performance.

BGnet: Accurate and rapid background estimation in single-molecule localization microscopy with deep neural nets

Universal and High-Fidelity Resolution Extending for Fluorescence Microscopy Using a Single-Training Physics-Informed Sparse Neural Network

Sparsity-based background removal for STORM super-resolution images

Low-frequency background estimation and noise separation from high-frequency for background and noise subtraction

Fast Background Removal Method for 3D Multi-channel Deep Tissue Fluorescence Imaging.

Photon-Efficient 3D Imaging with A Non-local Neural Network

High-axial-resolution Single-Molecule Localization under Dense Excitation with a Multi-Channel Deep U-Net.

Spatially-Variant CNN-based Point Spread Function Estimation for Blind Deconvolution and Depth Estimation in Optical Microscopy

Fast Confocal Microscopy Imaging Based on Deep Learning

Localization of Ultra-dense Emitters with Neural Networks

Robust single-shot 3D fluorescence imaging in scattering media with a simulator-trained neural network

Local Mean Suppression Filter for Effective Background Identification in Fluorescence Images

Deep Learning architectures for generalized immunofluorescence based nuclear image segmentation

A geometric approach to estimate background in astronomical images

Feature matching driven background generalization neural networks for surface defect segmentation

ABPNet: Adaptive Background Modeling for Generalized Few Shot Segmentation.

A Universal Foreground Segmentation Technique using Deep-Neural Network

Solving the inverse problem of microscopy deconvolution with a residual Beylkin-Coifman-Rokhlin neural network

GAB-Net: A Robust Detector for Remote Sensing Object Detection Under Dramatic Scale Variation and Complex Backgrounds

Non-resonant background removal in broadband CARS microscopy using deep-learning algorithms

Patch-based Non-Local Bayesian Networks for Blind Confocal Microscopy Denoising