Fig 1.
Variation of the PSF in three dimensions.
A vls (green plane, A) is defined as a volume above and below the focal plane of the microscope from which an emitter is imaged as a diffraction-limited spot on the detector (green optical path, A). A fluorophore emitting from outside the vls (red volume above and below the vls, A), is blurred on the image plane of the detector (red optical path, A). The z-stacks of 28 sub-diffraction beads were superposed to image the axial variation of the PSF of the instrument. The contrast-adjusted rendered volume (B) of the z-stack shows the axial variation of the width of the PSF. Three examples of (xy) planes of the z-stack in, above and below the vls are shown in (C-E). For each plane, a contrast-adjusted image (left column) and an intensity surface plot (right column) of the plane underlines the axial variation of the width (orange arrows) and the amplitude (blue arrows) of the PSF.
Fig 2.
Building the virtual-‘light-sheet’.
A calibration z-stack of images of immobile and separated, sub-diffraction fluorophores is imaged and its 2D-PSFs fitted. For each fitted PSF, the width is plotted against the amplitude in a parameter plot (A). PSFs detected in planes from within the vls are plotted in green; PSFs coming from outside of the vls are plotted in red. Two imaging modes are described, aimed at two different SMLM analyses: the structural mode (B) aims to increase the contrast of the image without rejecting a majority of localisations; the confidence mode (C), to reject more localisations and only accept localisations from within the vls with a higher certainty. For each mode, three steps are described: first the confidence (dotted grey line) and the recall (solid grey line) ratios are calculated for different width thresholds (left column). A width threshold (black vertical line) is identified to optimise both ratios in accordance with the aim of the experiment of interest. This step is repeated with the new thresholded list of localisations to identify an amplitude threshold (middle column). Finally, both chosen thresholds are visualised on the parameter plot (PSFs in the grey areas are rejected).
Fig 3.
Fixed S. pombe expressing cytoplasmic Cdc22-mEos proteins were imaged during a PALM experiment. 5,000 frames were analysed with Peak Fit and the resulting list of localisations was used to produce a super-resolved picture directly after fitting (A), or after applying the vlsPALM thresholds defined in Fig 2B (B). The corresponding diffraction-limited image of the two cells is shown as an inset in (A). Close-ups of the white rectangles in (A-B) are shown in (C-D). The contrast of the large intracellular vesicles of the yeast is increased after vlsPALM filtering (white arrows in (C-D)).
Fig 4.
Embryonic stem cells expressing Cenp-A-mEos proteins were fixed and imaged. The corresponding movie (summed in A) was analysed with Peak Fit and the resulting list of localisations was separated between in vls (green) and out of vls (red) localisations using the vlsPALM thresholds defined in Fig 2C. vlsPALM allows the identification of the in-focus localisations (B). All localisations were plotted either as fitted (C) or in a super-resolved picture (D), but coloured according to the vlsPALM filtering. Three categories of Cenp-A clusters were observed: some were almost entirely within the vls (D-F, 1); others were spanning one extremity of the vls, partly in the vls (D-F, 2); the last ones were entirely out of the vls (D-F, 3). (E) shows the diffraction-limited and super-resolved close-ups of the Cenp-A clusters defined in (D). (F) displays the number of localisations in (green) and out of (red) the vls for each cluster. Such classification allows selecting in-focus clusters for further quantification and preventing under-counting due to undetectable out-of-focus emitters.