Post-processing strategies in image scanning microscopy

Abstract

Image scanning microscopy (ISM) coupled with pixel reassignment offers a resolution improvement of 1.41 over standard widefield imaging. By scanning point--wise across the specimen and capturing an image of the fluorescent signal generated at each scan position, additional information about specimen structure is recorded and the highest accessible spatial frequency is doubled. Pixel reassignment can be achieved optically in real time or computationally a posteriori and is frequently combined with the use of a physical or digital pinhole to reject out of focus light. Here, we simulate an ISM dataset using a test image and apply standard and non--standard processing methods to address problems typically encountered in computational pixel reassignment and pinholing. We demonstrate that the predicted improvement in resolution is achieved by applying standard pixel reassignment to a simulated dataset and explore the effect of realistic displacements between the reference and true excitation positions. By identifying the position of the detected fluorescence maximum using localisation software and centring the digital pinhole on this co--ordinate before scaling around translated excitation positions, we can recover signal that would otherwise be degraded by the use of a pinhole aligned to an inaccurate excitation reference. This strategy is demonstrated using experimental data from a multiphoton ISM instrument. Finally we investigate the effect that imaging through tissue has on the positions of excitation foci at depth and observe a global scaling with respect to the applied reference grid. Using simulated and experimental data we explore the impact of a globally scaled reference on the ISM image and by pinholing around the detected maxima, recover the signal across the whole field of view.

UCsim2: 2D Structured Illumination Microscopy using UC2

“…field curvature) originating from the freeform optics or improper alignment ( Figure 4c). A snippet of 11 × 11 ' around each detected peak position is cut-out, multiplied with a a 2D Gaussian to simulate a "macro-pinhole" [20], [27] for better axial sectioning, summed and reassigned on a two times super sampled canvas (Figure 4c). Reassigning the information for all illumination spots in each frame produces a super-resolved image (Figure 4 d).…”

“…Here, we rely on two consumer-grade laser video projectors (Figure 2 b) to realize a complexity reduced and low-cost ISM setup. Our used image-processing is based on [20].…”

Image scanning microscopy: an overview

“…This has proven problematic in practical applications of ISM, where the pattern must either be determined by regularly calibrating the system using a test slide (York et al 2012;Schulz et al 2013) or by computationally identifying the excitation pattern post-acquisition. (Ströhl & Kaminski 2015;McGregor et al 2015) Both of these techniques have their limitations: calibrating the system is a time-consuming step, and is ineffective if the sample has greatly different optical properties to the test slide; determining the pattern post-acquisition adds another step to an already computationally intensive technique; and, depending on the algorithm used, reconstruction can break down in patterned or sparsely fluorescing samples. affords.…”

Super‐Resolution Confocal Microscopy Through Pixel Reassignment

“…Instead of summing the signals directly as the conventional imaging technologies, each signal is reassigned to a particular location where the signal most probably comes. Figure 1(a) gives the principle of the pixel reassignment in terms of excitation and detection point-spread function (PSF) [7]. The excitation PSF (PSF ex , labeled by blue line) represents the distribution of the corresponding excitation focus.…”