Volume Tells: Dual Cycle-Consistent Diffusion for 3D Fluorescence Microscopy De-noising and Super-Resolution

TL;DR

The paper tackles the challenge of de-noising and super-resolution in 3D fluorescence microscopy under spatially varying noise and anisotropic axial resolution, where paired ground-truth data are unavailable. It introduces Volume Tells (VTCD), a dual cycle-consistent diffusion framework that learns intra-volume priors via two conditional diffusion processes: a Spatially Iso-Distributed Denoiser to progressively reduce noise along the Z-axis and a Cross-Plane Global-Propagation SR module to transfer high-resolution XY information into XZ and YZ planes. The method operates in an unsupervised, cycle-trained setting and demonstrates substantial improvements over state-of-the-art unsupervised methods on a novel 4D live-cell dataset, including an axial-resolution enhancement from $430~\mathrm{nm}$ to $90~\mathrm{nm}$. This approach enables accurate denoising and 3D SR without paired HR data, offering a practical pathway for high-quality volumetric cell imaging under live-cell constraints.

Abstract

3D fluorescence microscopy is essential for understanding fundamental life processes through long-term live-cell imaging. However, due to inherent issues in imaging principles, it faces significant challenges including spatially varying noise and anisotropic resolution, where the axial resolution lags behind the lateral resolution up to 4.5 times. Meanwhile, laser power is kept low to maintain cell viability, leading to inaccessible low-noise and high-resolution paired ground truth (GT). To tackle these limitations, a dual Cycle-consistent Diffusion is proposed to effectively mine intra-volume imaging priors within 3D cell volumes in an unsupervised manner, i.e., Volume Tells (VTCD), achieving de-noising and super-resolution (SR) simultaneously. Specifically, a spatially iso-distributed denoiser is designed to exploit the noise distribution consistency between adjacent low-noise and high-noise regions within the 3D cell volume, suppressing the spatially varying noise. Then, in light of the structural consistency of the cell volume, a cross-plane global-propagation SR module propagates high-resolution details from the XY plane into adjacent regions in the XZ and YZ planes, progressively enhancing resolution across the entire 3D cell volume. Experimental results on 10 in vivo cellular dataset demonstrate high improvements in both denoising and super-resolution, with axial resolution enhanced from ~ 430 nm to ~ 90 nm.

Figures (4)

• Figure 1: Problem statement and results of our method. Top-Left: The fluorescence microscopy can observe long-time live-cell life processing, but face problems like spatially varying noise and anisotropic resolution. Top-Right: The raw slice (above) and our results (bottom). Bottom: The comparisons among our methods and other SOTA methods.
• Figure 2: The overall framework of the proposed Method. Left: two targeted slicing strategies are modeled as the forward stage of diffusion model via the imaging prior for de-noising and SR 3D cell volume. Middle: the spatially iso-distributed denoiser model is a conditional diffusion model, which progressively reduces the noise of 3D cell volume along the Z-axis, and the cross-plane SRM enables the conditional diffusion model propagates the content distribution of the HR XY plane to the adjacent 3D volume space, eventually the whole 3D space. Right: The comparison of our results (above) and original slices (below). Compared to the original slice, our results show significantly reduced noise, and previously unobservable structural information becomes clear and discernible.
• Figure 3: The qualitative comparison of the cell fluorescence dataset from three extremely different aspects. These results are from different cell and different life time, thus there are different number and shape of cells.
• Figure 4: Comparison of PSNR values across different methods in 3 datasets (under multiple imaging conditions), highlighting the statistical performance variations.