We investigate the application of hyperosmotic optical clearing agents to improve the image contrast and penetration depth in two-photon microscopy of human dermis ex vivo. We show that the agents glycerol, propylene glycol, and glucose all convey significant improvements and we provide results on their dynamic behaviour and the reversibility of the effect. At suitable concentrations, such agents have the potential to be compatible with living tissue and may possibly enhance in-vivo deep-tissue imaging.
© 2005 Optical Society of America
For microscopic imaging techniques that perform optical sectioning, the penetration depth in biological tissue is commonly limited by scattering produced by the high turbidity . Clearing agents have long been used to reduce scattering in fixed animal and plant tissue sections . It is pertinent to consider whether such agents could be employed in imaging tissue in vivo. Toxicity generally renders agents used for fixed sections as unsuitable for in vivo imaging of tissues and cells and has led to the consideration of potentially biocompatible alternative optical clearing agents (OCAs). The study of such OCAs and other exogenous agents capable of reducing scattering, enhancing contrast, and increasing penetration depth has been relatively recent [3–5]. The application of OCAs may prove to be particularly relevant for enhancing two-photon microscopy , since it has been shown that the effect of scattering is to drastically reduce penetration depth to less than that of the equivalent single-photon fluorescence whilst largely leaving resolution unchanged [7,8]. Improvement in the penetration depth of two-photon microscopy has been obtained by optimising the pulse shape and repetition rate for the sample under investigation . Here we report on our investigation of an alternative means of improving imaging performance through application of an exogenous optical clearing agent (OCA). Two-photon microscopy has generated impressive in vivo images of human skin [10–12], but it is potentially applicable to many tissue types. The focus of the present paper is to demonstrate the effect of OCAs on biological tissue, to quantify the magnitude of the effect and its timescale, and to elucidate mechanisms for optical clearing. To do this, we study thick frozen sections of human dermis. We anticipate that our results will not be directly applicable to in vivo imaging, but they provide a useful starting point.
The use of hyperosmotic agents, such as glycerol, propylene glycol, and acetic acid, has been reported in optical coherence tomography [5,13], second harmonic generation microscopy , and confocal reflectance microscopy [15,16], but their usefulness in two-photon microscopy has yet to be examined. An understanding of the clearing process at the microscopic level is still far from complete. A common hypothesis [3,4,13,16] is based on the assumption of diffusion of the OCA into the tissue and the subsequent outflow of water under osmotic pressure. The combined effects of OCA ingress and water egress are generally thought to provide better matching to the refractive index of the remaining ground matter, thus reducing scattering.
In this work, we used the potentially bio-compatible OCAs glycerol, propylene glycol, and glucose. For all agents, we observed an image contrast enhancement and a penetration depth increase due to the clarifying effect of the agent on the tissue. The dependence of contrast on image depth and on agent incubation time was also investigated.
2. Materials and methods
2.1 Experimental set-up
The experimental apparatus was built around an inverted microscope (Nikon Eclipse TE300), modified to perform both two-photon scanning fluorescence microscopy and conventional wide-field microscopy. For the wide-field subsystem, the source of light is a halogen lamp and the sample image is collected in transmission by a CCD camera (Hamamatsu C3077). For the two-photon subsystem, a mode-locked Ti:Sapphire laser (Coherent MIRA900) provides the excitation light, which comprises 100-fs width pulses at an 80-MHz repetition rate, tuneable in wavelength between 700 and 1000 nm. The laser light passes through a collimating telescope and is attenuated with a variable neutral density filter, before passing to the scanning head.
The scanning head comprises two galvanometer mirrors (GSI Lumonics VM500), rotated about orthogonal pivots, and coupled by a relay lens pair. The laser beam is raster scanned such that the beam focus is raster scanned in the focal plane of the objective lens at 1 frame per second. A scan lens (f=50 mm) and microscope tube lens (f=200 mm) enlarge the beam to a dimension of around 1 cm, before it is focussed onto the specimen by an objective (Nikon Plan-Apo 60X, N.A. 1.4, W.D. 170 µm, oil immersion). The working distance of the objective sets a limit on the depth of penetration independent of any optical clearing effect. The optical power at the sample was 3 mW (corresponding to an estimated intensity at the focal point of 1.6 MW/cm2) and the center wavelength was 750 nm.
The fluorescence is collected by the objective and retraces the same optical path as the laser excitation, encountering the two galvo-mirrors (descanning mode), before being separated out by a dichroic mirror and detected by an avalanche photodiode (Perkin Elmer SPCM-AQR). The wavelength range of the detection has an upper limit of 670 nm set by a cut-off optical filter, and a lower limit of 370 nm set by the detector response. The photocurrent is integrated using custom electronics to provide a proportional voltage. The acquisition and control is provided by a PC and I/O board (National Instruments PCI-6070E). A LabView program that also allows the visualization of the acquired images controls the input settings. For a more detailed description of the apparatus, see Ref. .
2.1 Sample preparation
Samples were taken from normal human skin excised during plastic surgery procedures performed on patients who agreed to participate in the study. After excision, the fresh skin samples were oriented the same as for routine histology and immediately frozen at -60°C in a compact routine cryostat (Leica Microsystems, CM 1850). Sections 150 µm thick (1cm×0.5 cm) were sliced and imaged within the same day. Each section was mounted on a glass coverslip that constituted the bottom of a plastic chamber of approximately 1 ml volume.
2.2 Image acquisition and analysis
Images of the connective tissue of the dermis were acquired at depths measured from the sectioned surface of the sample. Images were collected in stacks, each comprising four images of dimensions (500×500 pixels, 100 µm square) taken at depths of 20, 40, 60, and 80 µm from the sectioned surface of the skin tissue. Before acquisition, the sample was immersed in 0.1 ml of phosphate buffered saline (PBS) in order to prevent drying and shrinkage. Then, the tissue was immersed in 0.5 ml of an OCA and one image stack was acquired every 30 seconds for 6–7 minutes. We note that the sample orientation effectively bypassed the barrier effect provided by the stratum corneum. Finally, the OCA was removed and the sample was immersed again in 0.1 ml of PBS, in order to observe the reversibility of the clearing process. The OCAs investigated were: glycerol, propylene glycol, both in anhydrous form, and glucose (aqueous solution, 5M). Aqueous dilutions of these agents were also investigated. After application of the OCA, we repeatedly checked for axial shrinkage of the samples by checking the measured distance from the sample surface and comparing the image with previous images. We estimate an upper limit of 2% shrinkage in the 6–7 minutes duration of our experiments. We confirmed that the detected signal was predominantly two-photon autofluorescence by observing the complete loss of signal after insertion of a band pass filter centered at half the pump wavelength.
To estimate the average contrast in each image, we define a contrast function as follows:
where 〈Iij〉 is the mean intensity of the nearest eight pixels and Nlines=N-2, with N=500. Other measures of contrast exist, but would be expected to give similar results. Contrast, as defined by Eq. (1), is linearly dependent on the fluorescence intensity and varies according to structures in the image. Hence, its usefulness is primarily to enable comparison between images of the same sample at the same depth maintaining the same field of view. Normalization to the total intensity would be required in order to compare different images.
3. Results and discussion
Figure 1(A) shows two typical image stacks; the first acquired with the sample immersed in PBS and the second acquired 7 minutes after the application of glycerol. Figure 1(B) shows the corresponding total intensity (i.e., the intensity summed over all pixels). The images show connective tissue in human dermis, which is primarily composed of collagen and elastin fibers with very few cells in comparison with the epidermis. The contrast enhancement and the increased penetration depth (from 40 µm to 80 µm), as well as a higher intensity signal, are clearly distinguishable from the images. The corresponding contrast levels, calculated using Eq. (1), are plotted in Fig. 1(C). For the purposes of comparison, it is helpful to define the relative contrast (RC) by
where Contrast[OCA] and Contrast[PBS] are calculated using Eq. (1), for OCA and PBS immersion, respectively. RC is plotted in Fig. 1(D). It has a value of 200 at 40 µm and dramatically increases with increasing depth. The effect on deeper sections is greater because of the cumulative effect of the reduction in scattering in the sections closer to the tissue surface. In addition, this effect is further enhanced by the dependence of the contrast on the intensity of the fluorescence signal, which is in turn dependent on the square of the excitation intensity.
The time dependence of the optical clearing is shown in Fig. 2. The images in Fig. 2(A) show the evolution in time of a section at 60 µm depth, after immersion in glycerol. The corresponding contrast levels are plotted in Fig. 2(B).
The contrast increases with time at a decreasing rate, suggesting the existence of a saturation that was not reached in this case. The same broad trends are supported by the results on all OCAs shown in Fig. 3, in which contrast versus time and depth is presented. In each case, the sequence PBS-OCA-PBS was followed, as indicated on the figures. Figure 3 shows that each optical clearing agent resulted in a contrast enhancement with varying degrees of saturation. The dynamics and the final contrast level attained depend on the agent and on the depth. Saturation of contrast occurs most rapidly in sections that are closest to the sample surface. This is broadly consistent with a simple diffusion model for the penetration of the agent from the surface into the tissue, i.e., if the contrast is proportional to OCA concentration, then the saturation time at a given depth will be proportional to the depth.
Of the three agents, glycerol is the most efficient with respect to saturation level (RC=16.3 at 20 µm depth), but also the slowest. Propylene glycol is similarly efficient (RC=12.6 at 20 µm depth), whereas, glucose is the worst (RC=5.1 at 20 µm depth) but diffuses three times faster than glycerol and five times faster than propylene glycol. We also conducted experiments on aqueous solutions of the agents. We observed similar effects in terms of contrast and depth enhancement, but reduced with respect to the results presented in Fig. 3. In general, the higher the OCA concentration, the higher the contrast and depth enhancement we observed.
To date the favoured mechanism proposed for the reduction in scattering caused by OCAs has been refractive index matching. Recently, evidence has been presented for the disassociation of collagen fibre bundles in the presence of glycerol , suggesting that reduction in the size of collagen structures may also play a role. The effects of glycerol have also been shown to be somewhat reversible [5,14], which is confirmed by our results (Fig. 3(A)), and not to differ significantly between in vivo and ex vivo applications [5,14], suggesting that homeostasis does not play a large role. The group refractive index of synthetic type I collagen has been shown to vary according to its level of hydration from 1.43 to 1.53 (dry) ; the (phase) refractive index of glycerol is 1.47 ; that of propylene glycol is 1.43 ; and that of glucose is 1.46 . Since the level of hydration is unknown and also depends on the action of the agent itself, it is not obvious, based on the refractive indices alone, which agent should provide the best match to that of collagen. The precise modes of operation of propylene glycol and glucose have not yet been clarified. Our results (Fig. 3(B),(C)) show a slowing in the rate of contrast increase following addition of PBS rather than a decrease. This suggests a mode of operation distinct from that of glycerol. For live cellular tissue such as the epidermis, we expect significantly different behaviour than that reported here, resulting from the effects of homeostasis and the expected variations in osmotic pressure arising from live cells .
In conclusion, the use of a bio-compatible optical clearing agent in in vivo applications of two-photon microscopy or other non-linear optical biopsies could provide an option for deep tissue imaging. However, much work remains to be done to establish a detailed understanding of the action of the agent as well as what are suitable bio-compatible concentrations  and the means of delivery. In this paper, we have shown that at high concentration, the hyperosmotic agents glycerol, propylene glycol, and glucose, are effective in improving the image contrast and penetration depth (by up to a factor of two) in two-photon microscopy of ex vivo human dermis. Such improvements were obtained within a few minutes of application; only for glycerol was the effect shown to be partially reversible through reapplication of PBS. This adds some support to the hypothesis of alteration to the collagen structure as being a significant contributor to the clearing effect.
We thank LENS (European Laboratory for Non-linear Spectroscopy), Contract Number RII3-CT2003-506350 and Ente Cassa di Risparmio di Firenze for funding this project. David Sampson thanks LENS and Department of Physics, University of Florence for his sabbatical
References and links
1. W. F. Cheong, S. A. Prahl, and A. J. Welch, “A review of the optical properties of biological tissues,” IEEE J. Quantum Electron. 26, 2166–2185 (1990) [CrossRef]
2. J. A. Kiernan, Histological and Histochemical Methods, 3rd Edition, (Oxford University Press, New York), 1999
3. H. Liu, B. Beauvoit, M. Kimura, and B. Chance, “Dependence of tissue optical properties on solute-induced changes in refractive index and osmolarity,” J. Biomed. Opt. 1, 200–211 (1996) [CrossRef]
4. V. V. Tuchin, I. L. Maksimova, D. A. Zimnyakov, I. L. Kon, A. H. Mavlutov, and A. A. Mishin, “Light propagation in tissue with controlled optical properties,” J. Biomed. Opt. 2, 401–417 (1997) [CrossRef]
5. G. Vargas, E. K. Chan, J. K. Barton, H. G. Rylander, and A. J. Welch, “Use of an agent to reduce scattering in skin,” Lasers in Surg. and Med. 24, 133–141 (1999) [CrossRef]
7. A. K. Dunn, V. P. Wallace, M. Coleno, M. W. Berns, and B. J. Tromberg, “Influence of optical properties on two-photon fluorescence imaging in turbid samples,” Appl. Opt. 39, 1194–1201 (2000) [CrossRef]
8. M. Gu, X. Gan, A. Kisteman, and M. G. Xu, “Comparison of penetration depth between two-photon excitation and single-photon excitation in imaging through turbid tissue media,” Appl. Phys. Lett. 77, 1551–1553 (2000) [CrossRef]
9. E. Beaurepaire, M. Oheim, and J. Mertz, “Ultra-deep two-photon fluorescence excitation in turbid media,” Opt. Commun. 188, 25–29 (2001) [CrossRef]
12. K. König and I Riemann, “High-resolution multiphoton tomography of human skin with subcellular spatial resolution and picosecond time resolution,” J. Biomed. Opt. 8, 432–439 (2003) [CrossRef] [PubMed]
13. Y. He and R. K. Wang, “Dynamic optical clearing effect of tissue impregnated with hyperosmotic agents and studied with optical coherence tomography,” J. Biomed. Opt. 9, 200–206 (2004) [CrossRef] [PubMed]
15. A. F. Zuluaga, R. Drezek, T. Collier, R. Lotan, M. Follen, and R. Richards-Kortum, “Contrast agents for confocal microscopy: how simple chemicals affect confocal images of normal and cancer cells in suspension,” J. Biomed. Opt. 7, 398–403 (2002) [CrossRef] [PubMed]
16. I. V. Meglinski, A. N. Bashkatov, E. A. Genina, D. Y. Churmakov, and V. V. Tuchin, “The enhancement of confocal images of tissue at bulk optical immersion,” Laser Physics 13, 65–69 (2003)
17. L. Sacconi, I. M. Tolic-Nørrelykke, R. Antolini, and F. S. Pavone, “Combined intracellular three-dimensional imaging and selective nanosurgery by a nonlinear microscope,” J. Biomed. Opt.10 (1), (2005) [CrossRef] [PubMed]
18. X. Wang, T. E. Milner, M. C. Change, and J. S. Nelson, “Group refractive index measurement of dry and hydrated type I collagen films using optical low-coherence reflectometry,” J. Biomed. Opt. 1, 212–216 (1996) [CrossRef]
19. R. K. Wang, X. Xu, V. V. Tuchin, and J. B. Elder, “Concurrent enhancement of imaging depth and contrast for optical coherence tomography by hyperosmotic agents,” J. Opt. Soc. Am. B 18, 948–953 (2001) [CrossRef]
21. G. Vargas, K. F. Chan, S. L. Thomsen, and A. J. Welch, “Use of osmotically active agents to alter optical properties of tissue: Effects on the detected fluorescence signal measured through skin,” Lasers in Surg. and Med. 29, 213–220 (2001) [CrossRef]