2.1.

Chemicals

The chemicals used for synthesis of GNRs, including chloro- auric acid (HAuCl4·3H₂O), CTAB, sodium borohydride (NaBH₄), silver nitrate (AgNO₃) and ascorbic acid, and the chemicals for making phantom, including gelatin and TiO₂ were purchased from Sinopharm Chemical Reagent Co., Ltd., China. 3,3^′-diethylthiatricarbocyanine iodide (DTTC), 3,3^′-diethylthiadicarbocyanine iodide (DTDC) and thiol-polyethylene glycol (SH-PEG) were obtained from Sigma Aldrich. Deionized (DI) water was used in all the experimental steps.

2.2.

Synthesis of Dye-Coated GNRs

GNRs used in our experiment were synthesized through a two-step seed-mediated growth method.¹⁴ To coat Raman reporter molecules (see Fig. 1), GNR solutions were brought to 10,000 times molar amount of both DTDC (DTTC) and 5-kDa thiol-PEG, and dialyzed for 16 h in a 5-kDa-cutoff cellulose membrane.¹⁵ After dialysis, samples were purified with centrifugation to eliminate excess DTDC/DTTC and PEG molecules, and stored at 4°C for later use. In this paper, we call the dye-coated GNRs PEG-dye-GNRs (dye stands for DTTC or DTDC).

Fig. 1

Representative scheme showing the synthesis process of dye-coated GNRs.

2.3.

Experimental Characterization

Transmission electron microscope (TEM) analyses were derived from JEOL JEM-1200EX equipment, which is suitable for the investigation of PEG-dye-GNRs. The UV and visible extinction spectra were recorded from the scale of 250 to 900 nm with a Shimadzu 2550 UV-vis scanning spectrophotometer at room temperature.

2.4.

Phantom Preparation

A tissue phantom of gelatin/TiO₂/India ink was made and detail of the synthesis method is given in Swartling ¹⁶ The gelatin phantom used in our experiment had reduced scattering coefficient μ^′ _sca of about 2.0 cm⁻¹ and absorption coefficient μ_abs of about 0.06 cm⁻¹ under 1030 nm superluminescent diode (SLD) excitation.

2.5.

Raman/Fluorescence Signal Detection

In Fig. 2a, two tiny tubes filled respectively with the solution of PEG-DTTC-GNRs and PEG-DTDC-GNRs, were submerged into the phantom at a depth of 3.0 mm. The diameters of both tubes were 2.0 mm, and the distance between them was 1.0 cm. A Raman probe, which consisted of a 300-mW cw laser at 785 nm and a Raman spectrometer, was used to scan the phantom along the y direction (as shown in Fig. 2). The focal length of the convex lens of the Raman probe was 7.5 mm, and a backscattering geometry (a conventional method in Raman spectroscopy¹¹) was adopted to detect the optical signals. During the scanning, Raman signals, fluorescence signals, and elastic scattering signals were all collected by the probe. A bandpass filter was used to ensure the monochromaticity of the excitation laser and a long-pass filter was used to block the elastic scattering and extract the Raman signals. To introduce more light into the phantom, minimize the random noise, and enhance the Raman intensity, the Raman probe was placed near (about 2 mm away from) the surface of the phantom and the integration time of each detection process was set at 40 s. Furthermore, the probe was slightly tilted from the normal direction of the phantom surface to improve the collection efficiency of the back scattering signals.

Fig. 2

Setup for (a) the Raman DOI system and (b) the Raman probe. (BP, 785 nm bandpass; LP, long-pass filter at 800 nm; DM, dichroic mirror.) This probe is scanning along the y axis during [as shown in (a)] the experiment to acquire the original measured Raman spectra.

3.1.

Raman Spectra of PEG-dye-GNRs

Figure 3 shows the TEM images of GNRs (left) and PEG-DTTC-GNRs (right). As expected, these nanoparticles had a rod shape, and their rough average length and diameter were about 54 and 19 nm, respectively. The TEM image of the PEG-DTTC-GNRs is similar to that of the pure GNRs.

Fig. 3

TEM images for GNRs with longitudinal extinction peak of (a) 632 nm and (b) PEG-DTTC-GNRs.

In our experiment, GNRs with a longitudinal extinction peak of 632 nm [Fig. 4a] were chosen. As mentioned before, a dialysis procedure was used for the synthesis of dye-coated GNRs. The absorption spectra of the synthesized GNRs became smaller since some of the GNRs were lost during the dialysis. Only tiny wavelength shifts can be observed among the absorption spectra of GNRs, the dye-coated GNRs, and PEG-dye-coated GNRs, as shown in Fig. 4a, though the amount of the dye was 5000 times of that of GNRs. Since GNRs dominated the absorption of the dye-coated GNRs or PEG-dye-GNRS, PEG-dye-GNRs could still provide a high absorption contrast, as did the pure GNRs. As shown in Figs. 4c and 4d, the original measured Raman spectra of free DTTC/DTDC molecules (in solution) could not be observed [the curves with circle marked in Figs. 4c and 4d], although the excitation cw laser power (300 mW at 785 nm) was high and the integrated time of the detector was long enough (40 s). However, due to SERS effect, the intensity of the original measured Raman spectra of DTTC/DTDC molecules (in solution) was greatly enhanced by GNRs, even if the integrated time was just 100 ms. From the solid curves in Figs. 4c and 4d, we can find that DTTC molecule has its characteristic Raman peaks at 502, 632, 778, 841, 1129, and 1237 cm⁻¹, and the DTDC molecule has its Raman peaks at 502, 578, 632, 778, 841, 1129, and 1237 cm⁻¹. Here, 578 cm⁻¹ is the unique DTDC Raman peak, which could distinguish it from DTTC. To show that Raman spectra were not from the GNRs, the original measured Raman spectra of (only) GNRs (CTAB coated) was also measured as a control experiment [the curves with square marked in Figs. 4c and 4d], and almost no Raman signal could be observed (as expected).

Fig. 4

(a) Extinction spectra of GNR, GNR + dye and GNR + dye + PEG; (b) extinction and fluorescent spectral of DTTC and DTDC; (c) and (d) Raman spectra of nanoparticles used in our experiment. Compared with pure dye (DTTC or DTDC), the PEG-dye-GNRs provided great Raman enhancement.

3.2.

Scanning Raman/Fluorescence Signal in Phantom

Figures 5a and 5b show the detection results when the Raman probe was placed to aim at the red tube (doped with PEG DTTC-GNRs) and blue tube (doped with PEG-DTDC-GNRs) [shown in Fig. 2a], respectively. The baseline curves [discounted from the original measured Raman spectra by the “Peak analyzer” tool of the Origin program, shown in the inserts of Figs. 5a and 5b] composed of the fluorescence spectra of DTTC/DTDC and the autofluorescence spectra of the phantom. When the baseline was subtracted, the net Raman spectra of DTTC and DTDC were obtained [shown in Figs. 5c and 5d, which were then smoothed by the smoothing tool of Origin program]. Just as shown in Figs. 4c and 4d, it is easy to see that the Raman spectrum of PEG-DTTC-GNRs contains characteristic peaks at 502, 632, 778, 841, 1129, and 1237 cm⁻¹, while the Raman spectrum of PEG-DTDC-GNRs contains characteristic peaks at 502, 578, 632, 778, 841, 1129, and 1237 cm⁻¹. As reported previously, the low signal-to-noise ratio (SNR) of the backscattering Raman signal detected from tissue could be improved by spatially offset Raman spectroscopy¹⁷ (SORS). By using SERS in our experiment, the back scattering Raman signal, as well as its SNR, could be enhanced enormously.

With the advantages of SERS signals, we can determine the locations of the two tubes filled with PEG-dye-GNRs by performing 1-D scanning along the y axis to acquire the original measured Raman spectra, as shown in Fig. 2a. First, the red dotted curve in Fig. 6b, standing for the fluorescence signal of DTTC/DTDC and the autofluorescence of the phantom (which was mainly included in the original measured Raman spectrum) detected along the scanning direction, provides the location information of the two tubes. Second, the imaging could also be acquired from the Raman intensity. The green solid curve in Fig. 6b was the distribution of Raman signal along the scanning direction, which corresponds to the intensity of Raman signal at 502 cm⁻¹ [both the DTTC and DTDC have a Raman peak at 502 cm⁻¹, as shown in Figs. 5c and 5d]. Here, the fluorescence and Raman signals were normalized and magnified in Figs. 6c and 6d to represent these two curves more clearly. From the Raman intensity curves, one could also tell directly the positions of two tiny tubes. Furthermore, comparing these two kinds of signals, we could see that the gradient of the Raman intensity was more abrupt than that of the fluorescence intensity [see Figs. 6c and 6d]. This was so because the fluorescence signal was affected by autofluorescence. Suppressing the autofluorescence is challenging work. As we know, UCNPs (Ref. 3) and spectral component analysis¹⁸ are two useful methods to reduce the autofluorescence during imaging. However, in our study, it was difficult to separate the fluorescence signal from the autofluoresence by utilizing a single spectrometer and two kinds of traditional fluorescence dyes. However, the Raman signal was obtained by subtracted the baseline (including fluorescence signals and autofluoresence background) from the original measured Raman spectrum, and thus the autofluoresence can be easily removed from the Raman signal. Meanwhile, the Raman spectra have also been greatly enhanced by the GNRs due to the SERS effect. Therefore, the Raman signal of the dye-coated GNRs could provide better signal-to-background contrasts than a fluorescence signal. In addition, the Raman spectrum could also distinguish the chemical composition in these tubes, because the Raman spectrum is due to some specific molecular vibration of the dye. From the scanning results of the Raman spectra [shown in Figs. 5c and 5d], we can distinguish DTTC and DTDC from each other very easily. Therefore, we can determine both the position and type of the nanoparticles doped inside the phantom through Raman imaging.

3.3.

Results of Transmission Scanning

In addition, PEG-dye-GNRs could give a high absorption contrast as the third kind of information. As shown in Fig. 2a, the GNRs used in our experiment have a longitudinal extinction peak around 632 nm. After dyes and PEG coating, their extinction peak did not shift very much. Although the absorption peak was reduced by 50% since part of the GNRs was lost during the dialysis procedure, the PEG-dye-GNRs still provided great absorption around 632 nm. A schematic illustrating the transmission scanning of a phantom with two tubes of PEG-dye-GNRs is shown in Fig. 7a, and the input laser beam (from a 632-nm cw laser) and detector probe were collinearly scanning synchronously. The tubes were submerged into the phantom at a depth of 3.0 mm, and the distance between them was about 1 cm. This setup of the phantom was similar to that shown in Fig. 2a. The blue tube (right) was filled with a solution of PEG-DTDC-GNRs, and the red tube (left) was filled with a solution of PEG-DTTC-GNRs. Due to the high absorption contrast of PEG-dye-GNRs, the positions of both tubes could also be obtained from the transmission result. The notches of the curve represented the location of tubes with GNRs [Fig. 7b]. As a result, PEG-dye-GNRs, which were doped in the phantom, could provide the high absorption contrast as the third kind of information in tissue imaging.

Fig. 7

(a) Schematic diagram of the transmission detection and (b) the 1-D transmission collinearly scanning along the yaxis. Because of the absorption of PEG-dye-GNRs, there are two notches in the transmission curve due to the two tubes inside the phantom, giving the location information of these two tubes.