Photo-activated raster scanning thermal imaging at sub-diffraction resolution

Abstract

Active thermal imaging is a valuable tool for the nondestructive characterization of the morphological properties and the functional state of biological tissues and synthetic materials. However, state-of-the-art techniques do not typically combine the required high spatial resolution over extended fields of view with the quantification of temperature variations. Here, we demonstrate quantitative far-infrared photo-thermal imaging at sub-diffraction resolution over millimeter-sized fields of view. Our approach combines the sample absorption of modulated raster-scanned laser light with the automated localization of the laser-induced temperature variations imaged by a thermal camera. With temperature increments ∼0.5-5 °C, we achieve a six-time gain with respect to our 350-μm diffraction-limited resolution with proof-of-principle experiments on synthetic samples. We finally demonstrate the biological relevance of sub-diffraction thermal imaging by retrieving temperature-based super-resolution maps of the distribution of Prussian blue nanocubes across explanted murine skin biopsies.

Fig. 1

Photo-activated thermal imaging at sub-diffraction resolution. A Gaussian laser beam (red spot in a) is focused on the sample at (x₀, y₀) with a square pulse of duration τ_on. Light absorbing entities (blue) induce a local temperature increase, which is measured by the thermal camera as an evolving 2D Gaussian peak. A Gaussian fit of the frame collected at time τ_on (b, c) provides the temperature variation ΔT_max= ΔT_max − T₀ and the peak coordinates (x_c ± σ_x, y_c ± σ_y) that localize the center of the distribution of the absorbing objects within the laser spot size (magnified region in a). The procedure is repeated over sets of isolated points as exemplified in e, f. The center coordinates provided by the Gaussian fit of temperature peaks identify the position of the absorbing entities on the scan grid (d, g) with uncertainty σ_x,y and a color code assigned by the best-fit ΔT_max. A maximum projection of the signal across the raw thermo-camera image sequence provides a low (∼mm) resolution image of the sample (i), whereas a maximum projection of the stack containing all the localized absorptive centers provides the super-resolution image (j) of the scanned region (h). Scale bars (100 μm in the magnification of a, 1000 μm elsewhere) indicate the typical imaged fields of view; note that the ~50-μm laser spot and the whole magnified area in a lie within a single ~400-μm typical thermo-camera pixel on the sample plane. All panels have been derived from experimental data acquired on an explanted murine skin biopsy treated with 30-nm Prussian blue nano-cubes analogous to the one employed for Fig. 3.

Fig. 2

Proof-of-principle experiments. a, f Transmitted-light images of microfiche samples. b, e Temporal maximum projection of the raw thermo-camera image stacks acquired with modulated illumination on the samples in a and f, respectively; acquisition parameters: N_xxN_y = 120 × 36, δx = 34.4 μm, Δx = 60, Δy = 6, τ_on = 300 ms, P = 4.6 mW, beam 1/e² diameter 56 ± 2 μm in b, N_xxN_y = 160 × 42, δx = 15.3 μm, Δx = 160, Δy = 6, τ_on = 300 ms, P = 15 mW, beam diameter 22 ± 1 μm in e. c, g Super-resolution images of the samples in a and f, obtained from the image stacks employed for b and e, respectively; ΔT_min = 0.3 °C in c, 0.7 °C in g. d Black: intensity profile along the dashed line in a upon a lookup-table inversion on the image; red: ΔT_max profile along the dashed line in c. h Black: average intensity profile along the vertical direction in f upon lookup-table inversion; red: average ΔT_max profile along the vertical direction in g with contrast percentages between adjacent peaks. In d, h, uncertainties on ΔT_max values equal the thermo-camera sensitivity 0.1 °C and x-axis error bars for non-zero ΔT_max values are extrapolated from the σ_{x, y}-versus-ΔT_max plot. Scale bars = 500 μm.

Fig. 3

Sub-diffraction thermal imaging on murine biopsies. a Transmitted-light tile-scan image of an explanted murine skin biopsy treated with 30-nm PB nanocubes, which are highlighted in navy by an upper threshold on intensity counts. b Super-resolution image of ROI 1 in a (N_xxN_y = 100 × 48, δx = 37.6 μm, Δx = 50, Δy = 1, τ_on = 1 s, P = 15 mW, beam 1/e² diameter 22 ± 1 μm, ΔT_min = 0.3 °C) overlaid to the transmitted-light image (nanocubes in navy as in a). c Magnification of the red boxed region in b. d Temporal maximum projection of the thermo-camera stack employed for b. e Same as b with doubled pixel size (N_xxN_y = 50 × 24, δx = 75.3 μm, Δx = 25, Δy = 2), overlaid to the transmitted-light image (nanocubes highlighted in navy as in a); only a 2.7 × 1.6 mm² ROI is shown for the sake of display. f Transmitted-light tile-scan image of a nanoparticle-untreated explanted murine skin biopsy; pixels are highlighted in navy by the intensity upper threshold as in a. g Super-resolution image of ROI 1 in f (same acquisition parameters of b, ΔT_min = 0.1 °C); analogous results have been obtained in ROIs 2 and 3 in f (Supplementary Fig. 11). h Temporal maximum projection of the thermo-camera stack employed for g. i Histograms of the ΔT_max values of the super-resolution images of ROI1 in a and ROIs 1–3 in f. Scale bar = 345 μm in c, 1 mm elsewhere.