Obtaining super-resolved images at the mesoscale through Super-Resolution Radial Fluctuations

Session

Poster Session

Authors

Mollie Brown (1), Liam Rooney (2), Gwyn Gould (2), Gail McConnell (2)

Affiliations

1. Department of Physics, University of Strathclyde
2. Strathclyde Institute of Pharmacy and Biomedical Sciences, University of Strathclyde

Keywords

Mesolens, Super-resolution microscopy, Image Processing

Abstract text

Widefield fluorescence microscopy is commonly used for diffraction limited imaging of fluorescently labelled cellular structures. Super-resolution methods have been able to overcome this diffraction limit to achieve significantly higher spatial resolutions, but current methods often have a restricted field of view, minimising the available information and limiting the understanding of behaviours and interactions on a larger scale. It is possible to circumvent these issues by stitching and tiling many images of the same sample, however, this can introduce artefacts where the edges of the tiles are poorly matched, or where there is inconsistent fluorescence across separate tiles. Here, we have applied Super-Resolution Radial Fluctuations (SRRF)^1,2 in conjunction with the Mesolens³ to obtain super-resolved images over a field of view of 4.4 mm x 3.0 mm. This presents a new means of understanding both cellular structure and the collective behaviour, without the typical artefacts and problems encountered when replicating this large field of view with other conventional super-resolution techniques.

HeLa cells were fixed with 4% paraformaldehyde and the tubulin was fluorescently labelled. Specimens were imaged with the Mesolens using diffraction-limited widefield fluorescence imaging (4X, 0.47 numerical aperture). Forty images of n=3 specimens were taken at 2 second intervals, and images were then analysed with the SRRF plugin². Following processing, the accuracy of the output image was assessed using super-resolution quantitative image rating and reporting of error locations (SQUIRREL)⁴, which analyses the differences between the reference and super-resolution images to create error maps displaying areas of high to low agreement, and two additional quantitative quality measurements to determine the accuracy of the super-resolution reconstruction – resolution-scaled Pearson (RSP) coefficient, and resolution-scaled error (RSE). To further quantity the impact of SRRF processing on Mesolens images the resolution of the original image was calculated using Fourier Ring Correlation⁵ (FRC) and image decorrelation analysis⁶, and this was repeated for the super-resolution image. From this the improvement in resolution could be quantified, with a two-fold improvement when calculated using FRC, and up to three-fold when measured with image decorrelation analysis.

Figure 1 shows that SRRF processed images display a notable improvement in resolution compared with diffraction limited widefield fluorescence illumination with the Mesolens. We also observe increased contrast in the SRRF data at the mesoscale. SQUIRREL error analysis shows consistent structural agreement between the original raw image data and the SRRF processed images. This is also confirmed by high RSP values demonstrating high structural similarities, and low RSE values, showing low error between the original and super-resolution image. For Fig. 1 these were 0.99 and 23.7 respectively, which demonstrates accurate image reconstructions. Additionally, there was a consistent improvement in measured resolution in the SRRF images compared to that measured in the original diffraction-limited widefield images when using both FRC and image decorrelation analysis, with a 1.96x average improvement when using FRC on Fig. 1 and an improvement of 1.96x and 3.09x when using image decorrelation analysis for the selected ROIs.

Figure 1: A raw diffraction limited widefield epifluorescence Mesolens image (A), and SRRF processed image (B). For each ROI, a digital zoom of the raw widefield epifluorescence diffraction limited Mesolens image (C), the same ROI following SRRF processing (D), with the SQUIRREL generated error maps for each (E). The normalised intensity measurements for each of the highlighted line profiles are shown. Error bars are 50𝜇𝑚 for all ROIs.

By using SRRF in conjunction with diffraction-limited widefield Mesolens images, it has been possible to achieve super-resolved images at the mesoscale, over a field of view of 4.4 mm by 3.0 mm. The work here demonstrates a notable improvement in resolution and contrast over the original images, quantifiable using multiple methods, whilst also demonstrating through error mapping and analysis highly accurate super-resolution reconstructions and consistent structural agreement between these and the original images. Cumulatively, this provides a comparatively simple method for obtaining accurate super-resolution images over a large field of view, allowing for a simultaneous understanding of the cell cytoskeleton in the context of large-scale interactions.

References

1.            Gustafsson, N. et al. Fast live-cell conventional fluorophore nanoscopy with ImageJ through super-resolution radial fluctuations. Nat. Commun. 7, 12471 (2016).

2.            Culley, S., Tosheva, K. L., Matos Pereira, P. & Henriques, R. SRRF: Universal live-cell super-resolution microscopy. Int. J. Biochem. Cell Biol. 101, 74–79 (2018).

3.            McConnell, G. et al. A novel optical microscope for imaging large embryos and tissue volumes with sub-cellular resolution throughout. eLife 5, e18659 (2016).

4.            Culley, S. et al. Quantitative mapping and minimization of super-resolution optical imaging artifacts. Nat. Methods 15, 263–266 (2018).

5.            Koho, S. et al. Fourier ring correlation simplifies image restoration in fluorescence microscopy. Nat. Commun. 10, 3103 (2019).

6.            Descloux, A., Grußmayer, K. S. & Radenovic, A. Parameter-free image resolution estimation based on decorrelation analysis. Nat. Methods 16, 918–924 (2019).