Fig. 1. (a) Principle of blind-${\mathrm{S}}^3$. Schematic of the setup to achieve super-resolution using the SRS process SRG based on a single-pixel SIM scheme. (b1) Transverse and (b2) longitudinal planes of the scanning Stokes beam trajectory (red dashed line) over the stationary Raman-active specimen (blue) and structured pump (green), in this case, a speckle pattern. (b3) For every speckle realization, an SRS image is acquired forming an image stack that is passed to a SIM algorithm to reconstruct a super-resolved image. (c) Conventional SRS, consisting of raster scanning co-propagating pump and Stokes beams, is used as a control to demonstrate the increase in resolution when compared to standard imaging. (c1) Transverse and (c2) longitudinal planes of the Stokes and focused pump beam (green and red dashed lines) scanning trajectories over the stationary Raman-active specimen (blue) (c3) leading to a low-resolution image.

Fig. 2. Proof-of-concept of blind-${\mathrm{S}}^3$ capabilities to image beyond the diffraction limit. (a) Conventional and (b) blind-${\mathrm{S}}^3$ images of 239-nm-diameter polystyrene beads. (c) Line profiles showing the increase in the transverse resolution of blind-${\mathrm{S}}^3$ (solid line) compared to conventional methods (dashed line). (d) Conventional SRS (dashed line) and blind-${\mathrm{S}}^3$ (solid line) sectioning capability characterizations. All scale bars: 500 nm. Pixel dwell time are 73 and 300 µs for conventional and blind-${\mathrm{S}}^3$ methods, respectively.

Fig. 3. Transverse resolution analysis for blind-${\mathrm{S}}^3$. Outcome analysis of the images of various close-contact bead pairs: We use close-contact distances as a proxy for the bead diameter. The inset shows representative images used for analysis, with the dashed lines representing some of the beads chosen for evaluation.

Fig. 4. Bio-compatibility capabilities of blind-${\mathrm{S}}^3$ at reduced excitation energy densities. (a) Large FOV imaging of lipid droplets within HeLa cells (conventional SRS). (b) Two zoomed-in ROIs are depicted by dashed boxes with conventional SRS (left panels) and blind-${\mathrm{S}}^3$ (right panels) methods, with various line profiles shown in (ci), (cii), and (ciii) for conventional SRS (dashed line) and blind-${\mathrm{S}}^3$ (solid line), respectively. (d) Large FOV image of opaque 100-μm-thick mouse cerebellum (conventional SRS). (e) A zoomed-in ROI is depicted by dashed boxes with conventional SRS (left panel) and blind-${\mathrm{S}}^3$ (right panel) methods, (f) with line profiles chosen for conventional (dashed line) and blind-${\mathrm{S}}^3$ (solid line) methods. All scale bars: 500 nm. Pixel dwell time are for the top row (bottom row), 90 µs and 180 µs (100 µs and 270 µs) for conventional and blind-${\mathrm{S}}^3$ methods, respectively, in panel (b), and 100 µs and 300 µs for conventional and blind-${\mathrm{S}}^3$ methods, respectively, in panel (e).

Fig. 5. Maximum temperature rise of the speckle envelope.