Development of a real-time flexible multiphoton microendoscope for label-free imaging in a live animal

Abstract

We present a two-photon microendoscope capable of in vivo label-free deep-tissue high-resolution fast imaging through a very long optical fiber. First, an advanced light-pulse spectro-temporal shaping device optimally precompensates for linear and nonlinear distortions occurring during propagation within the endoscopic fiber. This enables the delivery of sub-40-fs duration infrared excitation pulses at the output of 5 meters of fiber. Second, the endoscopic fiber is a custom-made double-clad polarization-maintaining photonic crystal fiber specifically designed to optimize the imaging resolution and the intrinsic luminescence backward collection. Third, a miniaturized fiber-scanner of 2.2 mm outer diameter allows simultaneous second harmonic generation (SHG) and two-photon excited autofluorescence (TPEF) imaging at 8 frames per second. This microendoscope's transverse and axial resolutions amount respectively to 0.8 μm and 12 μm, with a field-of-view as large as 450 μm. This microendoscope's unprecedented capabilities are validated during label-free imaging, ex vivo on various fixed human tissue samples, and in vivo on an anesthetized mouse kidney demonstrating an imaging penetration depth greater than 300 μm below the surface of the organ. The results reported in this manuscript confirm that nonlinear microendoscopy can become a valuable clinical tool for real-time in situ assessment of pathological states.

Figure 1. Scheme of the TPME system with linear and nonlinear pulse shaping.

(a) Scheme of the experimetal setup; CM: cut mirror; DM: dichroic mirror; PZT: piezzoelectric tube. The miniature fiber-scanning imaging probe is embeded inside a 2.2 mm outer diameter (OD) stainless steel biocompatible tube (for more details see Supplemenatry Fig. S2a–d). (b) Second order autocorrelation (AC) of the IR excitation pulse at the exit of the 5-m-long endoscopic fiber for a delivered power of 20 mW. The pulse duration has been calculated from the AC duration by using the suitable conversion factor (i.e. 1.54 = (AC duration)/(pulse duration) at FWHM, assuming a sech intensity shape pulse). Accordingly, the pulse duration was equal to 39 fs (FWHM).

Figure 2. Custom-design air-silica DC-PCF used as the endoscopic fiber within the TPME.

(a) Close view of the inner core of the fiber through scanning electron microscopy (SEM). Pure silica is in grey and air in black. (b) SEM image of the fiber cross-section without its outer polymer mechanical cladding. The silica jacket and the second core diameters are respectively equal to 266 μm and 188 μm. The red square denotes the inner core and its microstructured cladding which are depicted in (a). (c) DC-PCF flexibility.

Figure 3. TPME optical resolutions. Intensity spatial distributions obtained during imaging a 0.1 μm-diameter fluorescent bead.

Blue circle: measurements; red line: Gaussian fit. Transverse and axial resolutions were deduced from the FWHM of the Gaussian fits of the intensity distributions. (a) Transverse resolution: Δx = 0.83 μm. (b) Axial resolution: Δz = 12 μm.

Figure 4. Label-free microendoscopy images of fixed tissue samples ex vivo.

Intrinsic TPEF in red and SHG in green. (a–d) raw optical sections. (e–h) perspective view from ImageJ 3D software from a set of sixty optical sections each one corresponding to a given depth below the tissue surface, from depth 0 μm to 300 μm. (a,b) unstained intact and fresh rat tail tendon with 5 mW onto the sample. The red arrows indicate the rectilinear polarization impinging the sample. (c) mouse ear section. D: dermis; E: epidermis; IC: internal cartilage. (d): healthy human distal lung (alveolar area); alveolar wall and alveolar entrances; this optical section has been taken 100 μm below the sample surface. (e) perspective view of a rich collagen mouse aorta sample. (f–h): three perspective views of the extracellular matrix network at 3 different locations within a healthy human distal lung sample. (d,h) correspond to the same location. Scale bars, 50 μm.

Figure 5. Label-free in vivo experiment.

(a) Anesthetized mouse with one kidney being elevated from the body and clamped between two tongue depressors, beneath the 2.2 mm TPME probe (red arrow). A constant power of 30 mW was launched onto the tissues. (b) SHG (in green) and TPEF (in red) raw image of respectively the collagen of the capsule and the intracellular flavins of epithelial cells of the kidney tubules. (c) same as in (b) but with a larger FOV of 450 μm. Scale bars, 50 μm. (d): Successive optical sections of a fibrotic kidney, 6 days after fibrosis induction, taken just after mouse death; the imaging depth below the organ surface is indicated in the bottom in white; in (d), FOVs are 250 μm wide.

Figure 6. Comparison between healthy and fibrotic kidney capsules.

Top line: in vivo label-free raw images delivered by the TPME; SHG in green and TPEF in red. (a,b) correspond to the same image, without TPEF in (b). Bottom line: reference ex vivo SHG images of the same tissues taken post-mortem with a high NA bench-top SHG microscope. (a,b,d) pathologic kidney, 6 days after fibrosis induction; (c,e): healthy kidney. Scale bars, 50 μm.