Handheld multiphoton and pinhole-free reflectance confocal microscopy enables noninvasive, real-time cross-sectional imaging in skin

Abstract

Biopsy-based histology has been the foundation of disease diagnosis and management for over a century. A long-sought goal in dermatology is the development of an imaging modality with sufficient resolution and compositional detail to noninvasively interrogate skin histology in vivo. Here, we describe a system that achieves this goal using cross-sectionally scanned, multimodal microscopy (cross-modal). Cross-modal combines multiphoton and reflectance confocal microscopy into one compact system with coordinated three-axis scanning that preserves optical resolution in cross-section. A custom pinhole-free mechanism employing finite-infinite conjugates further simplifies and stabilizes confocal alignment. Evaluated in participants ages 9-81 and Fitzpatrick skin types (FST) 1-5, cross-modal images revealed histological details analogous to those obtained from traditional biopsied tissue. We observed dermal elastosis in sun-damaged skin, elevated melanin in pigmented skin, basaloid nests in basal cell carcinoma, and elongated rete ridges in seborrheic keratosis, supporting cross-modal's potential to deliver histological insights noninvasively.

Fig. 1

A benchtop, handheld imaging system for multimodal, cross-sectional images. (a) Handheld wand. (b) System hub, including display and wrapped cable. (c) Example H&E section of skin with the same FOV as that of cross-modal images, and (d) example cross-modal composite image of normal skin. FOV setting; standard. Example H&E and cross-modal are from different participants. Scale bars: 100 µm. (e) Schematic of how oblique scanning produces a slanted transection of the point spread function (PSF) for improved cross-sectional resolution. A-PSF, axial PSF; L-PSF, lateral PSF; O-PSF, oblique PSF; and angle, θ (30º). The resulting image is represented as a vertical projection (VP) such that the vertical axis of the images represents a true axial depth. (f) Schematic of the internal layout of the handheld wand and light path (dashed line). (g) Schematic of the compact reflection chamber within the wand used to reject out-of-focus light for pf-RCM.

Fig. 2

Simultaneous multimodal, four-channel imaging with cellular resolution. (a) An in vivo, four-channel capture of normal human skin (Fitzpatrick skin type [FST] 4) with each channel shown individually as a grayscale image: (a) 2PS, (b) 2PL, (c) RCM, (d) SHG, and (e) composite. All four channels are collected simultaneously in vivo for registered features (f–i). Zoom of the dash-bounded box: (f) 2PS, (g) 2PL, (h) RCM, and (i) SHG. (a–i) Scale bars, 100 µm. FOV setting; standard. (j) In vivo, cross section detail of summed 2PS and 2PL signal in the basal layer of the epidermis. Intercellular spaces and filamentous structures (white arrows) within the epidermis are smaller than 2 µm diameter white circles. FOV setting; zoom. (k) Representative image of a 0.5 µm fluorescent bead. Scale bar, 15 µm. FOV setting; zoom. (l) Representative intensity plot through bead axis used to calculate system resolution.

Fig. 3

Tradeoff between resolution and FOV addressed with user-selected FOV in freehand, live skin imaging. In a single session of freehand imaging of living skin, the user captured an image of interest with (a) a FOV optimized for signal-to-noise ratio for improved clarity of small features (“zoom”; 133 µm × 100 µm). Cell nuclei in the epidermis and thin fibrils in the dermis are resolved. Variations in cytoplasmic color from green to yellow indicate variations in melanin concentration. White circles, 1 µm diameter. (b) While freehand imaging the same region of skin (approximate location of dotted rectangle), the user captured a larger FOV (“standard”; 400 µm × 300 µm), which captures a different view and context of histologic features. Scale bars, 100 µm.

Fig. 4

Demonstration of imaging across a variety of subjects with normal skin. Images of skin from sun-exposed, dorsal forearm in female participants with FST 2 or 3 of different ages: (a) age 24, (b) age 51, and (c) age 81. White triangles point to examples of green autofluorescence within the dermis. FOV setting; standard. Scale bar, 100 µm. SAAID values of dorsal forearm and ventral forearm dermis for (d, e) FST 1 and 2 (age < 30, n = 3; age > 49, n = 8).; (f, g) FST 3 (age < 30, n = 3; age > 49, n = 11); and (h, i) FST 4 and 5 (age < 30, n = 4; age > 49, n = 3). **, p < 0.01, effect size (ES) > 1.3; NS, not significant, p ≥ 0.05. (e, g, i) Linear trends of SAAID by age. Data points are skin regions of individual participants; ventral forearm by open circles with solid trendlines (R²: 0.77 [FST 1–2], 0.57 [FST 3], < 0.001 [FST 4–5]) and dorsal forearm by solid triangles with dotted trendlines (R²: 0.53 [FST 1–2], 0.47 [FST 3], 0.04 [FST 4–5]).

Fig. 5

Cross-modal imaging of lesional skin with paired histopathology from the same lesion. Seborrheic keratosis (a) H&E and (b) cross-modal. White arrows, elongated rete ridge; asterisks (*), pseudohorn cyst. Basal cell carcinoma (c) H&E and (d) cross-modal. White arrows, basaloid nodule. H&E histopathology is from the same lesion as cross-modal, but not the same plane or location within the lesion. Images (a–d) at the same scale, scale bar: 100 µm. (b, d) FOV setting; standard.