1.

Introduction

For decades, metal nanoparticles such as colloidal gold and silver have been used as contrast agents for electron microscopy (EM) of biological specimens.^{1, 2} Recently, the enhanced localized surface plasmon resonance (LSPR) of metal nanoparticles has been exploited to develop optical sensors for in vitro biological assays. ^{3, 4, 5, 6, 7, 8} As a result of the LSPR, gold- and silver-based metal nanoparticles exhibit significantly higher extinction coefficients than organic fluorophores.⁹ Metal nanoparticles have unique absorption and scattering properties, which are strongly dependent on particle size, shape, material, and interparticle spacing. Most of the current bioassays that utilize metal nanoparticles rely on substantial changes in the color and strength of nanoparticle absorption upon target-induced aggregation.^{9, 10, 11}

More recently, a number of studies have suggested that light scattering from targeted gold nanoparticles can be used to image changes in biomarker expression in living biological specimens, including cell culture^{12, 13, 14} and fresh human tissue explants.¹⁵ In vivo molecular-specific reflectance imaging using targeted metal nanoparticles offers important advantages for point-of-care detection of disease; metal nanoparticles resonantly scatter light with distinct spectral features^{14, 15, 16} and backscattered light can be collected from optically thick biological specimens. The scattering cross-section of metal nanoparticles is several orders of magnitude larger than that of polymeric particles of similar size.¹⁴ The scattering signal from a single nanoparticle has been shown to be equivalent to $\sim {10}^6$ fluorophores in a homogeneous assay^{6, 9} and is not affected by photobleaching, providing the potential for these nanoparticles to detect biological molecules at very low concentrations over long periods of time. Theoretical calculations suggest that light scattering can be detected from as low as ${10}^{-16}\mathrm{M}$ concentration of nanoparticles in solution.^{17, 18} Furthermore, several studies suggest that metal nanoparticles can be safely delivered in vivo for diagnostic imaging applications. Metallic gold and silver nanoparticles have shown low bioreactivity and toxicity.^{19, 20, 21}

Recent developments in photonic technology provide the ability to noninvasively image cellular morphology in vivo; the use of optically active metal nanoparticles provides an opportunity to simultaneously image cellular morphology and changes in biomarker expression with subcellular spatial resolution. For many diagnostic applications, it is desirable to combine high-resolution imaging with low-resolution widefield imaging. The ability to initially interrogate large areas of suspicious tissue with widefield imaging, followed by high-resolution imaging of suspicious regions, has the potential to improve the sensitivity and specificity of disease detection. A number of simple surgical microscopes and endoscopes that can capture multispectral digital images of large areas of tissue in real time have recently been developed to survey large surface areas of tissue and identify suspicious regions of interest. These areas can then be imaged with higher spatial resolution using high-resolution chromoendoscopy, optical coherence tomography, or in vivo confocal microscopy to achieve subcellular image resolution in near real time to confirm the presence of disease.^{22, 23}

Targeted contrast agents can enhance image contrast in both modalities; however, the production of image contrast in widefield and high-resolution optical imaging differs. In widefield imaging, detected photons can undergo multiple scattering events and image contrast is produced as a result of local fluctuations in both light scattering and absorption. On the other hand, high-resolution reflectance imaging relies predominantly on single backscattering to generate image contrast. An ideal molecular-targeted contrast agent should provide significant contrast in both widefield and high-resolution imaging modes at physiologically relevant concentrations of target biomarkers.

Image contrast depends on a number of factors, including the optical properties of the particles themselves, the properties of the target tissue including the baseline optical properties and the concentration of the targeted biomarker concentration, as well as properties of the system used to acquire images. Maximizing image contrast for a particular application requires an understanding of the interplay of these factors. A number of theoretical approaches have been developed to predict the absorption and scattering properties of single metallic nanoparticles as a function of their size, shape, and material composition.^{24, 25} These studies have provided insights into how changes in particle design affect the optical properties of single particles, but they do not fully address the important question of the expected image contrast when a particular particle is used as a contrast agent in biological tissue.

In this paper, we describe a methodology to evaluate expected image contrast as a function of particle optical properties, tissue optical properties, and imaging modality. We illustrate the application of this methodology to predict image contrast for gold and silver nanospheres targeted against a biomarker expressed in epithelial tissue in both widefield and high-resolution imaging modalities. Predictions based on diffuse reflectance spectroscopy are compared with experimental results of widefield and high-resolution images of labeled tissue phantoms. Predictions of image contrast agree well with those observed experimentally. Results illustrate the potential of gold and silver nanospheres to produce significant contrast in both widefield and high-resolution imaging at subnanomolar concentration. Interestingly, image contrast for a particular particle can differ dramatically for widefield and high-resolution reflectance imaging.

The goal of this study is to develop and evaluate a method to predict the anticipated image contrast properties of targeted gold and silver nanospheres in widefield and high-resolution imaging for biomedical imaging applications. This basis of this method is the use of diffuse reflectance spectroscopy to evaluate the remitted light from nanoparticles as a function of both the optical properties of the surrounding environment and any biomarker induced aggregation of targeted nanoparticles.²⁶ We illustrate that image contrast in both high-resolution imaging and widefield imaging can be predicted by measuring diffuse reflectance spectra of nanoparticles in a variety of phantoms of progressively increasing complexity, including water, turbid gelatin, and cell-culture-based tissue phantoms. Reflectance spectra of nanoparticles in water provide a measure of their scattering in a dilute suspension and can predict image contrast anticipated in high-resolution reflectance imaging. Reflectance spectra of nanoparticles in gelatin and cell-culture-based tissue phantoms provide a measure of their expected image contrast in widefield reflectance imaging. Gelatin phantoms provide a simple model, which mimics the extracellular environment of a tissue without any contribution from cells, while cell-based tissue phantoms model both the cellular and extracellular environment of a tissue. To validate the predictions of this method, we present experiments that compare contrast predicted based on diffuse reflectance spectroscopy measurements directly to that achieved in both high-resolution and widefield reflectance images.

For the results of this methodology to be applicable to biomedical imaging, it is important to choose both the size and concentration of contrast agents to be biologically relevant. For biomedical imaging applications, the size range of nanoparticles is often determined based on the potential for effective in vivo delivery to a target tissue. Using various delivery routes, molecular imaging applications have successfully used nanoparticles in the size range of 20 to $100\mathrm{nm}$ in diameter.^{26, 27} Larger sized nanoparticles are not effective for targeting deep within a tissue, while small-sized particles (1 to $15\mathrm{nm}$ ) do not have a high scattering cross section.²⁸ Thus, in this study we focused on gold and silver nanospheres with a mean diameter of $50\mathrm{nm}$ as a representative size to determine the contrast properties of these particles. In addition, experiments were also carried out with $25\text{-}\mathrm{nm}$ gold nanoparticles, to evaluate the effect of size on contrast properties of nanoparticles.

In many molecular imaging applications, the target biomarker is often present at low concentration, often at subnanomolar levels. Thus, in phantoms lacking cells we used subnanomolar concentrations of particles. In cell-culture-based phantoms, we targeted the epidermal growth factor receptor (EGFR); in these phantoms, the total number of EGFR receptors determines the maximum concentration of nanoparticles. For phantoms containing approximately 30 million cancer cells per ml with an estimated ${10}^5$ EGFR receptor molecules per cell²⁹ and assuming a 100% efficiency of targeting, the effective concentration of nanoparticles in a tissue phantoms is still below the nanomolar level. Thus, the contrast properties of nanoparticles in this study were evaluated at a subnanomolar level of concentration in all model systems.

2.

Material and Methods

3.

Results and Discussion

3.1.

Characterization of Gold and Silver Nanospheres

Figure 1 shows UV-Vis extinction spectra of gold ( $50\mathrm{nm}$ , $25\mathrm{nm}$ ) and silver nanoparticles $\left(50\mathrm{nm}\right)$ in water. In each case, spectra are normalized to a peak extinction value of 1. Solutions of silver nanospheres are light yellowish in color and exhibit peak extinction at approximately $430\mathrm{nm}$ , while solutions of gold nanospheres are deep red in color with peak extinction at approximately $530\mathrm{nm}$ . The extinction peak of gold nanoparticles does show a slight shift with a change in size from 25 to $50\mathrm{nm}$ . The peak extinction value for $25\text{-}\mathrm{nm}$ -diameter particles is $524\mathrm{nm}$ , while that for $50\text{-}\mathrm{nm}$ -diameter particles is observed at $529\mathrm{nm}$ .

Fig. 1

UV-Vis spectra of gold and silver nanospheres. The peak absorption of silver (Ag nanospheres of $50\text{-}\mathrm{nm}$ mean diameter) and gold nanospheres (Au nanospheres of 25- and $50\text{-}\mathrm{nm}$ diameters) is $\sim 430$ and $\sim 530\mathrm{nm}$ , respectively.

3.2.

Diffuse Reflectance Spectroscopy of Gold and Silver Nanospheres in Model Phantoms

Figure 2 shows the diffuse reflectance spectra from nanoparticles in three model systems: water, gelatin, and a tissue phantom. In each case, the spectra shown are divided pointwise by that measured from a blank of identical composition except for the nanoparticles. Thus, relative signal intensity greater than 1 indicates that the particles are expected to provide positive image contrast whereas values lower than 1 predicts absorptive contrast. The spectra measured in aqueous suspension are predominantly governed by single scattering, and can be used to predict contrast expected in high-resolution reflectance imaging, while the data measured in gelatin and tissue phantoms include contributions from multiple scattering, and can be used to predict contrast expected in widefield reflectance imaging.

Fig. 2

Diffuse reflectance spectroscopy of (a) $25\text{-}\mathrm{nm}$ gold (b) $50\text{-}\mathrm{nm}$ gold, and (c) $50\text{-}\mathrm{nm}$ silver nanoparticles in water, gelatin, and molecular targeted tissue phantoms.

Figure 2a shows reflectance spectra measured for phantoms containing $25\text{-}\mathrm{nm}$ -diameter gold nanospheres, Fig. 2b shows that for $50\text{-}\mathrm{nm}$ gold nanospheres, and Fig. 2c shows results for $50\text{-}\mathrm{nm}$ -diameter silver nanospheres. In water, the scattering of the subnanomolar concentration of $25\text{-}\mathrm{nm}$ gold nanoparticles is only slightly higher than the background signal at wavelengths above $580\mathrm{nm}$ ; below this wavelength, contrast is predominantly absorptive even in suspension. In contrast, the diffuse reflectance spectra of the subnanomolar concentration of $50\text{-}\mathrm{nm}$ -diameter nanospheres are significantly higher than that of the background at wavelengths above $500\mathrm{nm}$ . This can be explained based on the lower scattering coefficient of $25\text{-}\mathrm{nm}$ nanoparticles as compared to $50\text{-}\mathrm{nm}$ -diameter gold nanoparticles. In both gelatin and tissue phantoms, the diffuse reflectance of samples containing 25- and $50\text{-}\mathrm{nm}$ nanoparticles is lower than that of the corresponding background with a maximal absorption around $530\mathrm{nm}$ . This suggests that these particles are expected to provide predominantly absorptive contrast in widefield imaging. Similar to the UV-Vis results, we did not observe a significant change in spectral characteristics as the particle size increases from 25 to $50\mathrm{nm}$ in diameter in gelatin and tissue phantoms.

Figure 2c shows the diffuse reflectance spectra of silver nanospheres in water, gelatin, and tissue phantoms, normalized to the appropriate background. Results demonstrate that silver nanoparticles provide positive contrast in all model systems, including tissue phantoms. This is consistent with the higher scattering coefficient expected of silver nanoparticles. The maxima of the scattering contrast from silver nanoparticles in gelatin and tissue phantoms occurs at approximately 500 to $550\mathrm{nm}$ , and is red-shifted relative to the diffuse reflectance peak at approximately $475\mathrm{nm}$ observed in water. Determination of these shifts in peak wavelength is important to optimize contrast properties of nanoparticles for biomedical imaging.

3.3.

Widefield Imaging of Gold and Silver Nanospheres in Model Phantoms

In order to evaluate the predictions based on Fig. 2, we obtained widefield reflectance images of nanospheres in gelatin phantoms at several illumination wavelengths. Results are shown in Fig. 3 . Figure 3a shows widefield reflectance images of $50\text{-}\mathrm{nm}$ -diameter gold and silver nanospheres in gelatin phantoms, as well as control images of gelatin phantoms without nanoparticles. The images are collected using same exposure time at each particular wavelength range. Figure 3b shows the normalized average intensity of measured from the nanoparticle containing phantom relative to that of the gelatin blank at each illumination wavelength. Figure 3b indicates that the contrast produced by the $50\text{-}\mathrm{nm}$ gold nanospheres is absorptive (less than 1) at illumination wavelengths below $600\mathrm{nm}$ ; this is consistent with the predictions of diffuse reflectance measurements in gelatin. Based on multispectral imaging, the peak absorptive contrast from gold nanoparticles is obtained at $530\mathrm{nm}$ . The contrast produced the silver nanospheres is positive at all wavelengths, with peak contrast measured at $530\text{-}\mathrm{nm}$ illumination. These results illustrate that these materials can provide both absorptive and scattering based contrast. The nature and wavelength dependence of contrast in widefield imaging are predicted well by diffuse reflectance measurements in gelatin-based phantoms.

Fig. 3

(a) Widefield multispectral reflectance microscopy of gold and silver nanoparticles ( $50\text{-}\mathrm{nm}$ mean diameter) in gelatin phantoms using white-light, 450-, 530-, 600-, and $650\text{-}\mathrm{nm}$ excitation, respectively. (b) Normalized scattering intensity of gold and silver nanoparticles in gelatin phantoms. The intensity is normalized based on gelatin scattering $(\mathrm{gelatin}=1)$ at each excitation wavelength.

Figure 4a shows widefield images of molecular targeted contrast agents in tissue phantoms using multispectral widefield imaging as well as images from an unlabeled tissue phantom. In these tissue phantoms, gold or silver nanospheres ( $50\text{-}\mathrm{nm}$ mean diameter) conjugated with anti-EGFR antibodies were targeted to $\mathrm{SiHa}$ cells, which after removal of excess contrast agents were added to gelatin solution to form tissue phantoms. Figure 4b shows the normalized mean reflectance intensity of phantoms containing molecular targeted contrast agents relative to the signal from the unlabeled control tissue phantom. The results indicate that tissue phantoms labeled with gold nanospheres produce primarily absorptive contrast at all illumination wavelengths examined in widefield imaging. Again, results agree well with diffuse reflectance spectroscopy measurements in tissue phantoms. Phantoms labeled with molecular targeted silver nanoparticles show positive contrast at all wavelengths, with peak contrast observed at $530\text{-}\mathrm{nm}$ illumination.

Fig. 4

(a) Widefield multispectral reflectance microscopy of gold and silver nanoparticles ( $50\text{-}\mathrm{nm}$ mean diameter) in molecular targeted tissue phantoms using white-light, 450-, 530-, 600-, and $650\text{-}\mathrm{nm}$ excitation, respectively. (b) Normalized scattering intensity of gold and silver nanoparticles in molecular targeted tissue phantoms. The intensity is normalized based on untargeted tissue phantom (untargeted tissue phantom=1).

In many widefield-imaging applications, an intensity ratio of approximately 3 units is considered a significant contrast. Based on the literature, studies with fluorescence molecular imaging using IR and near-IR labels^{34, 35} have reported intensity ratios ranging from 1.2 to 3.0. This ratio is achievable only with selection of an optimized wavelength range. In other imaging modalities such as MR imaging, intensity ratios above 1.5 are considered as a strong signal.³⁶

3.4.

Contrast Properties of Nanospheres Using High-Resolution Imaging

Figure 5a shows high-resolution confocal reflectance images of cells labeled with targeted gold and silver nanospheres obtained at $633\text{-}\mathrm{nm}$ illumination. Confocal reflectance images were obtained from cells targeted with anti-EGFR conjugated nanoparticles; IgG conjugated nanoparticles were used a negative control to assess specificity of targeting. The images show an increase in perimembrane reflectance in EGFR labeled cells, with some intracellular signal; this intracellular signal could result from internalization of targeted receptors. The results demonstrate specific targeting of nanoparticle-based contrast agents. Additionally, the results show positive contrast from both anti-EGFR targeted gold and silver nanoparticles, and are consistent with the results of the diffuse reflectance spectra measured in water.

Fig. 5

(a) High-resolution imaging of gold and silver nanoparticles targeting EGFR receptors in cancer cells using confocal reflectance microscopy at $633\text{-}\mathrm{nm}$ laser excitation: (A) representative IgG conjugated gold or silver nanoparticles as a negative control, (B) anti-EGFR conjugated gold nanoparticles, and (C) anti-EGFR conjugated silver nanoparticles. (b) Multispectral high-resolution confocal reflectance imaging of anti-EGFR targeted gold and silver nanospheres at 633-, 543-, and $488\text{-}\mathrm{nm}$ laser excitation, respectively.

Figure 5b shows high-resolution confocal reflectance images of cells labeled with gold and silver nanoparticles as the illumination wavelength is varied from 633 to 533 to $488\mathrm{nm}$ . Based on the gray-scale intensity of image contrast at each wavelength, we calculated ratiometric contrast of silver to gold as 0.93 at $633\mathrm{nm}$ , 3.0 at $533\mathrm{nm}$ , and 3.98 at $488\mathrm{nm}$ . The results agree well with diffuse reflectance spectra measured in water and show that gold-nanoparticle-based molecular contrast agents provide strongest scattering contrast at $633\mathrm{nm}$ , while their contrast at 488- and $533\text{-}\mathrm{nm}$ excitation is significantly lower. For silver-nanoparticle-based molecular contrast agents, strong positive contrast is observed across a broad range of wavelengths showing high peaks at 488 and $533\mathrm{nm}$ . The results of multiwavelength high-resolution imaging of contrast agents are in agreement with the predicted contrast based on diffuse reflectance spectroscopy of gold and silver nanoparticles in water.

4.

Summary and Conclusion

In summary, we have demonstrated an approach to help predict the effective image contrast from exogenous nanoparticle-based contrast agents for biomedical imaging applications. This approach integrates spectroscopy, widefield imaging, and high-resolution imaging to understand optimal contrast properties of nanoparticles as a function of nanoparticle size and composition, imaging modality, and surrounding environment. Contrast properties predicted from diffuse reflectance spectral measurements can provide an estimate of the effective image contrast as a function of these parameters. Using widefield and high-resolution imaging, the predicted contrast properties can be validated and optimal imaging wavelengths for maximizing the image contrast can be determined.

Based on this integrated approach, we demonstrate that gold and silver nanoparticles can provide a unique source of contrast for widefield and high-resolution imaging at subnanomolar concentration levels. Contrast properties of nanoparticles observed with diffuse reflectance spectroscopy were validated with multispectral widefield and high-resolution microscopy. While the nanoparticles investigated here can produce contrast either in diffuse reflectance mode for widefield imaging, or through single backscattering for high-resolution reflectance imaging, the resulting contrast depends strongly on both the imaging technique and the illumination wavelength, as well as the optical properties of the background medium. The results of diffuse reflectance spectroscopy and widefield imaging highlight that the expected image contrast from nanoparticle-based contrast agents is strongly dependent on the optical properties of the surrounding environment. This illustrates the need to evaluate the contrast properties of various nanomaterials in an appropriate biological environment.

Based on the results presented here, widefield reflectance imaging at $530\text{-}\mathrm{nm}$ illumination provides an optimal operating point to achieve high contrast with both silver and gold nanospheres, although the type of contrast is absorptive for gold nanospheres and scattering for silver nanospheres. For high-resolution imaging, gold nanoparticles provide peak scattering contrast at $633\mathrm{nm}$ while the peak scattering contrast for silver nanoparticles was observed at $543\mathrm{nm}$ .

Evaluating the contrast properties of nanoparticles in biologically relevant models is an important step in their translation from bench to bedside. Additional studies to develop efficient contrast agents will require development of novel coatings to improve stability and bioconjugation, noninvasive delivery and clearance routes for nanoparticles, and validation of their biocompatibility. These developments, in combination with engineering of novel instruments to detect specific contrast from nanoparticles, will enable the clinical translation of these technologies.