Abstract:Microscopy images are crucial for life science research, allowing detailed inspection and characterization of cellular and tissue-level structures and functions. However, microscopy data are unavoidably affected by image degradations, such as noise, blur, or others. Many such degradations also contribute to a loss of image contrast, which becomes especially pronounced in deeper regions of thick samples. Today, best performing methods to increase the quality of images are based on Deep Learning approaches, which typically require ground truth (GT) data during training. Our inability to counteract blurring and contrast loss when imaging deep into samples prevents the acquisition of such clean GT data. The fact that the forward process of blurring and contrast loss deep into tissue can be modeled, allowed us to propose a new method that can circumvent the problem of unobtainable GT data. To this end, we first synthetically degraded the quality of microscopy images even further by using an approximate forward model for deep tissue image degradations. Then we trained a neural network that learned the inverse of this degradation function from our generated pairs of raw and degraded images. We demonstrated that networks trained in this way can be used out-of-distribution (OOD) to improve the quality of less severely degraded images, e.g. the raw data imaged in a microscope. Since the absolute level of degradation in such microscopy images can be stronger than the additional degradation introduced by our forward model, we also explored the effect of iterative predictions. Here, we observed that in each iteration the measured image contrast kept improving while detailed structures in the images got increasingly removed. Therefore, dependent on the desired downstream analysis, a balance between contrast improvement and retention of image details has to be found.

Image restoration of degraded time-lapse microscopy data mediated by near-infrared imaging

Enhanced Stimulated Raman and Fluorescence Imaging by Single-Frame Trained BDN

Long-term monitoring of intravital biological processes using fluorescent protein-assisted NIR-II imaging

Time-lapse Image Super-resolution Neural Network with Reliable Confidence Quantification for Optical Microscopy

Enhancing Total Optical Throughput of Microscopy with Deep Learning for Intravital Observation

Evaluation and development of deep neural networks for image super-resolution in optical microscopy

Research of NIR-II small animal living fluorescence imaging system

Rapid and Extreme Low-light Superresolution Imaging Via Artificial Intelligence

Time-lapse Image Super-resolution Neural Network with Reliable Confidence Evaluation for Optical Microscopy

Rationalized deep learning super-resolution microscopy for sustained live imaging of rapid subcellular processes

Deep image restoration for infrared photothermal heterodyne imaging

Live-cell imaging in the deep learning era

CIEGAN: A Deep Learning Tool for Cell Image Enhancement

Intravital Confocal Fluorescence Lifetime Imaging Microscopy in the Second Near-Infrared Window

DeepContrast: Deep Tissue Contrast Enhancement using Synthetic Data Degradations and OOD Model Predictions

Wide-Field, High-Resolution Reconstruction in Computational Multi-Aperture Miniscope Using a Fourier Neural Network

Near-infrared-II Deep Tissue Fluorescence Microscopy and Application

Surmounting photon limits and motion artifacts for biological dynamics imaging via dual-perspective self-supervised learning

Advanced Imaging Integration: Multi-Modal Raman Light Sheet Microscopy Combined with Zero-Shot Learning for Denoising and Super-Resolution

De-scattering Deep Neural Network Enables Fast Imaging of Spines Through Scattering Media by Temporal Focusing Microscopy

Fast live-cell conventional fluorophore nanoscopy with ImageJ through super-resolution radial fluctuations