Abstract:A fundamental challenge in fluorescence microscopy is the defocused background caused by scattering light, optical aberration, or limited axial resolution. Severe defocus backgrounds will submerge the in-focus information and cause artifacts in the following processing. Here, we leverage a priori knowledge about dark channels of biological structures and dual frequency separation to develop a single-frame defocus removal algorithm. It stably improves the signal-to-background ratio and structural similarity index measure of images by approximately 10-fold, and recovers in-focus signal with 85% accuracy, even when the defocus background is 50 times larger than in-focus information. Our Dark-based optical sectioning approach (Dark sectioning) is fully compatible with various microscopy techniques, such as wide-filed microscopy, polarized microscopy, laser-scanning / spinning-disk confocal microscopy, stimulated emission depletion microscopy, lightsheet microscopy, and light-field microscopy. It also complements reconstruction or processing algorithms such as deconvolution, structure illumination microscopy, and super-resolution optical fluctuation imaging.

What problem does this paper attempt to address?

This paper attempts to solve the out - of - focus background problem in fluorescence microscope imaging caused by scattered light, optical aberration or limited axial resolution. Severe out - of - focus background can overwhelm the in - focus information and generate artifacts in subsequent processing. By utilizing the prior knowledge of the dark channel of biological structures and the dual - frequency separation technique, the authors developed a single - frame defocusing algorithm - Dark sectioning - to stably improve the Signal - to - Background Ratio (SBR) and the Structural Similarity Index Measure (SSIM) of the image and restore the in - focus signal, achieving an accuracy of about 85% even when the out - of - focus background is 50 times larger than the in - focus information. Specifically, this method can: 1. **Significantly improve image quality**: By removing the out - of - focus background, the SBR and SSIM of the image are increased by approximately 10 times respectively. 2. **Be widely compatible with multiple microscopic techniques**: Including wide - field microscopes, polarization microscopes, laser - scanning/rotating - disk confocal microscopes, stimulated - emission - depletion microscopes, light - sheet microscopes and light - field microscopes. 3. **Supplement reconstruction or processing algorithms**: Such as deconvolution, structured - illumination microscopy and super - resolution optical fluctuation imaging, etc. ### Main contributions - **Innovatively apply the dark - channel - first principle**: By adjusting the block size and introducing an inhomogeneous background distribution, the limitations of the traditional dark - channel method in fluorescence image applications are solved. - **Efficiently remove the out - of - focus background**: Even when the out - of - focus background is much larger than the in - focus information, it can still effectively retain the weak in - focus signal. - **Wide applicability**: Applicable to multiple microscopic techniques and image - processing algorithms, significantly improving the axial - sectioning ability. ### Method principle 1. **Definition of the dark channel**: \[ I_{\text{dark}}(x, y)=\min (I(x+\Delta x, y+\Delta y)) \] where \(\Delta x\) and \(\Delta y\) represent the block size, and the dark channel represents the minimum value in the RGB channels within a certain block size. 2. **Redefine the block size**: \[ [\Delta x, \Delta y]=\text{rec}(\text{PSF}) \] where \(\text{rec}(\cdot)\) represents the circumscribed rectangle size slightly larger than the point - spread function (PSF), which helps to better distinguish between in - focus and out - of - focus information. 3. **Image decomposition and processing**: - Decompose the image into a low - frequency part \(I_{\text{Lo}}(x, y)\) and a high - frequency part \(I_{\text{Hi}}(x, y)\). - Only perform dark - based optical sectioning on the low - frequency part to better preserve the weak signal. - Divide the low - frequency part into a background part \(A(x, y)\) and a background - free part \(J_{\text{Lo}}(x, y)\) through the transmission ratio \(t(x, y)\): \[ I_{\text{Lo}}(x, y)=t(x, y)\cdot J_{\text{Lo}}(x, y)+(1 - t(x, y))\cdot A(x, y) \] 4. **Final image synthesis**: \[ J(x, y)=J_{\text{Lo}}(x, y)+J_{\text{Hi}}(x, y) \] where \(J_{\text{Hi}}(x, y)=I_{\text{Hi}}(x, y)\), that is, the high - frequency part remains unchanged. ### Experimental verification - **Simulation experiment**: Using hollow spheres and linear structures for simulation, it is verified that Dark sectioning can still effectively restore the in - focus information when the in - focus intensity is only 2% of the out - of - focus information. - **Multi - modal verification**: Verified by multiple microscopic techniques such as HiLo, 2DSIM, 3DSIM, etc.

Dark-based Optical Sectioning assists Background Removal in Fluorescence Microscopy

Sub-diffraction Dark Spot Localization Microscopy

Deep Learning Enhanced Fluorescence Emission Difference Microscopy

Axial Resolution Enhancement for Light Sheet Fluorescence Microscopy Via Using the Subtraction Method

Accurate background reduction in adaptive optical 3D-STED nanoscopy by dynamic phase switching

Resolution Enhancement and Background Suppression in Optical Super‐Resolution Imaging for Biological Applications

Background Suppression with Dual Modulation by Saturated Absorption Competition Microscopy

A Nnovel Wide Field Fluorescence Sectioning Microscope Based on Polarization Filtering Image Enhancement and Dynamic Speckle Illumination

Computational Optical Sectioning with an Incoherent Multiscale Scattering Model for Light-Field Microscopy.

Rapid Image Deconvolution and Multiview Fusion for Optical Microscopy.

Computational Optical Sectioning by Phase-Space Imaging with an Incoherent Multiscale Scattering Model

Efficient high-resolution fluorescence projection imaging over an extended depth of field through optical hardware and deep learning optimizations

Single-scan HiLo with line-illumination strategy for optical section imaging of thick tissues

Deep Learning Optical-Sectioning Method

Noise-robust, physical microscopic deconvolution algorithm enabled by multi-resolution analysis regularization

Combined evanescent-wave excitation and supercritical-angle fluorescence detection improves optical sectioning

Noise-robust, Physical Genuine Deconvolution Algorithm for Fluorescence Microscopy Enabled by Multi-Resolution Analysis

Pixelation with concentration-encoded effective photons for molecular optical sectioning microscopy

Extended Focused Imaging in Microscopy Using Structure Tensor and Guided Filtering

Deep learning-based aberration compensation improves contrast and resolution in fluorescence microscopy

A Deep-Learning-based Workflow to Deal with the Defocusing Problem in High-Throughput Experiments