Journal: Biomedical Optics Express

Article Title: Cusp-artifacts in high order superresolution optical fluctuation imaging

doi: 10.1364/BOE.382296

Figure Lengend Snippet: Gamma-corrected high-order SOFI-processed experimental images displaying cusp artifacts. Fixed HeLa cells were labeled with QDs (emission wavelength = 800 nm) by immunostaining using primary antibody (eBioscience, Cat#: 14-4502-80) and secondary antibody conjugated to QD800 (ThermoFisher Scientific. Ref#: Q11071MP). Total of 2000 frames (exposure time of 30 ms) were processed to obtain SOFI cumulants with up to 7th order using both auto- and cross-correlations. In order to better illustrate source of cusp artifacts, final SOFI processing steps of deconvolution and Fourier reweighting were skipped. Each SOFI image of particular order is presented in three panels: large field-of-view (left), magnified absolute value SOFI image of box area (middle), and magnified positive/negative values SOFI image of box area (right). Positive/negative domains are color coded separately as shown by color bars for each panel, with color scheme shown at bottom. Cusp artifacts can be seen clearly for cumulants of orders greater than two: spatial distributions of cusps for cumulants of different orders differ and are located at boundaries between positive and negative domains. Scale bars: 3.2 µm (left) and 1.6 µm (middle/right). Image intensities are displayed with gamma correction to highlight the cusps, therefore resolution enhancement is not evident. Gamma values are the multiplicative inverse of the cumulant order. More comprehensive displays are available in Appendix 5 [28].

Article Snippet: On the other hand, when we performed deconvolution with solvers that imposed positivity constraints (for example, MATLAB’s “ deconvlucy ” or “ deconvblind ” functions), deconvolution failed ( ).

Techniques: Labeling, Immunostaining

Journal: Biomedical Optics Express

Article Title: Cusp-artifacts in high order superresolution optical fluctuation imaging

doi: 10.1364/BOE.382296

Figure Lengend Snippet: Post-processing of SOFI reconstructions containing cusp artifacts (simulation). Amplitudes of reconstructions are shown in grayscale (each panel has different dynamic range). Background of each panel is always zero (and therefore should be used as reference). Negative pixel values have darker colors than background; positive pixel values have lighter colors than background. (i) Ground-truth virtual emitters with both positive and negative values. (ii) Corresponding 3rd-order cumulant image (convolved with PSF). (iii) Amplitude (absolute value) of (ii) cusps are clearly visible. (iv) Ideal deconvolution result obtained by dividing Fourier-transformed image by optical transfer function (OTF) and subsequently performing inverse Fourier transformation. (v) Ideal Fourier reweighting, where, in contrast to the case for ideal deconvolution, Fourier spectrum is multiplied by extending the OTF before followed with inverse Fourier transform. (vi) Deconvolution result obtained using “deconvlucy” function, which imposes positivity constraint, that could affect the deconvolution when the corresponding ground-truth contains a mixture of positive and negative virtual brightnesses. PSF is simulated as perfect Gaussian with standard deviation of 4 pixels, as shown by the isolated emitter at the bottom right corner on each panel.

Article Snippet: On the other hand, when we performed deconvolution with solvers that imposed positivity constraints (for example, MATLAB’s “ deconvlucy ” or “ deconvblind ” functions), deconvolution failed ( ).

Techniques: Transformation Assay, Standard Deviation, Isolation