Accurate and rapid background estimation in single-molecule localization microscopy using the deep neural network BGnet

Abstract

Background fluorescence, especially when it exhibits undesired spatial features, is a primary factor for reduced image quality in optical microscopy. Structured background is particularly detrimental when analyzing single-molecule images for 3D localization microscopy or single-molecule tracking. Here, we introduce BGnet, a deep neural network with a U-net-type architecture, as a general method to rapidly estimate the background underlying the image of a point source with excellent accuracy, even when point spread function (PSF) engineering is in use to create complex PSF shapes. We trained BGnet to extract the background from images of various PSFs and show that the identification is accurate for a wide range of different interfering background structures constructed from many spatial frequencies. Furthermore, we demonstrate that the obtained background-corrected PSF images, both for simulated and experimental data, lead to a substantial improvement in localization precision. Finally, we verify that structured background estimation with BGnet results in higher quality of super-resolution reconstructions of biological structures. Significance:A main factor that degrades image quality in fluorescence microscopy is unwanted background fluorescence. Background is almost never uniform, especially in complex samples. Rather, background usually exhibits some structure, making it very difficult to distinguish from the signal of interest. Due to this challenging problem, background fluorescence is often assumed to be uniform even though it is not. This assumption leads to deteriorated image quality, e.g. in localization-based superresolution microscopy, and to the introduction of uncharacterized biases. To overcome this challenge, we developed a general framework rooted in deep learning to accurately and rapidly estimate arbitrarily structured background using several distinct image shapes for the single emitter. Proper background estimation allows for critical performance improvement of various optical microscopy methods. Main Text:

Deep learning in single-molecule microscopy: fundamentals, caveats, and recent developments [Invited]

“…8) A second recent project from our lab focused on the identification and quantitative extraction of background fluorescence from PSF images, a key issue in all single-molecule-based fluorescence imaging methods. [103] Background fluorescence is an umbrella term for any contributions to a fluorescence image that does not originate from the species investigated, but from other sources. For example, background fluorescence in biological samples frequently is caused by unspecific binding of labeling agents to other cellular components than the one that should be labeled.…”

“…9): using quantitative measures, the prediction of BGnet matched the ground-truth cases quite well. [103] Fig. 9.…”

Probing Chromatin Compaction and Its Epigenetic States in situ With Single-Molecule Localization-Based Super-Resolution Microscopy

“…Alternative image processing methods can also be used to mitigate the high background. Spatial filters such as wavelet ( Izeddin et al, 2012 ) or background correction methods ( Ma et al, 2019 ; Ma and Liu, 2020 ; Mockl et al, 2020 ) can be used as a pre-processing step prior to single molecule localization to reduce the background noise. On the other hand, emitter sparsity is a requirement for high-quality image reconstruction.…”

AutoSmarTrace: Automated Chain Tracing and Flexibility Analysis of Biological Filaments

“…To date, machine learning has proven to be a powerful tool for microscopy in biological fields, and it can improve significantly the efficiency and performance of large-scale image analysis, such as with medical diagnostics (15) and cellular imaging (16). More recently, deep learning techniques have been applied to imaging at the single-molecule level (17) to extract molecular parameters (18), improve localization (19), and automate analysis (20). A neural network thus presents a good candidate for the automation of the chain tracing process.…”