Rapid Cellular-Resolution Skin Imaging with Optical Coherence Tomography Using All-Glass Multifocal Metasurfaces

Abstract

Cellular-resolution optical coherence tomography (OCT) is a powerful tool offering noninvasive histology-like imaging. However, like other optical microscopy tools, a high numerical aperture (N.A.) lens is required to generate a tight focus, generating a narrow depth of field, which necessitates dynamic focusing and limiting the imaging speed. To overcome this limitation, we developed a metasurface platform that generates multiple axial foci, which multiplies the volumetric OCT imaging speed by offering several focal planes. This platform offers accurate and flexible control over the number, positions, and intensities of axial foci generated. All-glass metasurface optical elements 8 mm in diameter are fabricated from fused-silica wafers and implemented into our scanning OCT system. With a constant lateral resolution of 1.1 μm over all depths, the multifocal OCT triples the volumetric acquisition speed for dermatological imaging, while still clearly revealing features of stratum corneum, epidermal cells, and dermal-epidermal junctions and offering morphological information as diagnostic criteria for basal cell carcinoma. The imaging speed can be further improved in a sparse sample, e.g., 7-fold with a seven-foci beam. In summary, this work demonstrates the concept of metasurface-based multifocal OCT for rapid virtual biopsy, further providing insights for developing rapid volumetric imaging systems with high resolution and compact volume.

Keywords: all-glass metasurface; cellular resolution; multifocal beam; noninvasive imaging; optical coherence tomography; phase change materials; virtual biopsy.

Figure 1.

Principle of multifocal metasurface. (a) Multifocal metasurface composed of multiple phases of M foci via random spatial multiplexing. (b) Multifocal metasurfaces (MS) combine with a lens (L) to create hybrid lenses capable of generating multifocal beams. Simulations of four distinct multifocal beams are shown in linear scale. (c) With multifocal metasurfaces, OCT can generate two, three, or seven focal planes, as experimentally confirmed by OCT B-scans of 0.8 μm PS beads (displayed in logarithmic scale). Gaussian beam has one focal plane. The four multifocal beams correspond to the ones in (b).

Figure 2.

Metasurface design and fabrication. (a) Phase modulation library of SiO₂ nanopillars with varying diameters ranging from 340 to 850 nm on the silica substrate. (b) Simulated phase map of the various nanopillar diameters for the wavelengths 800–1050 nm. (c) An optical photograph of the fabricated metasurface. (d) SEM images of the nanopillars.

Figure 3.

Characterization of multifocal beams. Beam profiles of Gaussian beam, two-foci beam 1 (2Foci-1), two-foci beam 2 (2Foci-2) and three foci beam (3Foci) in (a) XZ plane and (b) XY planes and their (c) axial intensity profiles and (d) beam diameters along the depth.

Figure 4.

Volumetric imaging of 0.8 μm PS beads by Gaussian, two-foci (2Foci-1 and 2Foci-2), and three-foci (3Foci) beams. (a) Maximum intensity projections, scale bar = 25 μm. (b) XY images at different depths by Gaussian, 2Foci-1, 2Foci-2, and 3Foci beams. (c) The axial intensity profiles, (d) bead diameters at different depths, and (e) the SNR ratios along the depth are measured from the volumetric images.

Figure 5.

OCT imaging of normal human nasal skin by (a) Gaussian, (b) two-foci (2Foci-2), and (c) three-foci (3Foci) beams. The Gaussian beam is focused on the skin’s surface (Z = 0) and one bright layer is observed in B-scan (XZ-plane). The first focus of 2Foci-2 and 3Foci beams are placed on the skin’s surface with two and three focal planes captured in B-scans. At Z = 0, all three beams resolve identical structures in the stratum corneum, e.g., the feature marked by the orange arrows. Both 2Foci-2 and 3Foci reveal epidermal cells at Z = 40 μm. 3Foci beam further resolves cells at Z = 80 μm. Several epidermal cells are indicated by red arrows. Scale bar = 30 μm.

Figure 6.

Volumetric imaging of normal human nasal skin by Gaussian with 42 Z-stacks, two-foci (2Foci-2) with 21 Z-stacks, and three-foci (3Foci) beams with 16 Z-stacks. The sample volume is 250 μm × 250 μm × 250 μm (X × Y × Z). (a–c) In the 3D images captured by the three beams, the stratum corneum, epidermis, and dermis are equally and clearly identified. The extracted B-scan (located at Y = 125 μm) shows the same cells can be identified across all three volumes, with some examples denoted by red arrows. Scale bar = 40 μm. (d–f) XY images at different depths across all three beams demonstrate equivalent image qualities, showing the features in the stratum corneum (marked by orange arrows), epidermal cells (red arrows), and dermal-epidermal junctions (white circles). SC, stratum corneum; ED, epidermis; D, dermis.

Figure 7.

Volumetric imaging of a BCC nasal sample by the three-foci (3Foci) beam (imaging volume, X × Y × Z = 250 μm × 250 μm × 250 μm). (a) 3D imaging, scale bar = 40 μm. (b) The B-scan (XZ plane) located at Y = 120 μm. (c–e) The XY cross sections at three depths. SC, stratum corneum; ED, epidermis; D, dermis.