Raman spectroscopy is an optical technique based on inelastic scattering of monochromatic light. It can provide molecular information of specimens via probing vibrational energy transitions in molecules. Researchers have illustrated that Raman spectroscopy could be potentially employed for in vitro and in vivo diagnosis of diseases and malignancies in various tissues (e.g., lung, breast, prostate, ovarian, brain, bone, and etc.), originating from biochemical differences between normal and diseased/malignant tissues [1–5]. Since it can provide information about molecular structure and composition of specimens with no contrast agents, Raman spectroscopy offers valuable complementary information to current anatomical or functional imaging techniques such as electrochemical, electrical, thermal, ultrasound, x ray, and nuclear magnetic resonance.

However, traditional Raman techniques have been restricted by their shallow penetration depth, which are typically less than 1 mm [1–5]. This limitation renders many tissue components, such as deep cancerous tissues and bones, inaccessible by traditional Raman techniques. Recently, the evolvement of spatially offset Raman spectroscopy (SORS) has made a breakthrough in the penetration depth using diffuse optical techniques [6–13]. SORS can detect Raman signals at a penetration depth up to several millimeters and in some cases several centimeters. Moreover, SORS can achieve differentiation of Raman signals in terms of depths inside the sample. It has been investigated in detection of bone diseases, cancers, glucose levels, and pharmaceutical tablets/capsules [6–13].

Current designs of SORS fiber optic probes generally have the fibers fixed in place, which offer constant offsets ($\mathrm{\Delta }s$) between the excitation and collection units and limited amount of fibers in the SORS probe [6–13]. When a large range of $\mathrm{\Delta }s$ is necessary (e.g., for large or deep samples), more collection fibers or multiple channels of illumination/collection units will be required to construct a fiber probe, which might not be cost efficient and has less flexibility when interrogating samples with different depth. In addition, when multiple collection units are utilized in SORS, each fiber requires individual calibration for system and background correction. More fibers thus could lead to more complicated calibration and operation procedures.

In this manuscript, we present a new probe design for SORS, which is based on a mechanical iris concept and offers adjustable spatial offsets. This fiber probe design could potentially provide a design-efficient and cost-effective solution for various Raman applications.

The schematic and photos of the mechanical iris-based fiber optic probe prototype, which was fabricated by using the 3D-printing technique (Laser Imaging Inc., Houston, Texas), are shown in Fig. 1. The fiber probe consisted of an iris, a turnkey, a set of turning rods, a set of frame support rods, one excitation fiber tube, and six collection fiber tubes. The excitation fiber tube (orange) was fixed at the iris center, and collection fiber tubes (blue) were mounted on movable blades (red) of the iris. The set of frame support rods constructed a frame to support the turnkey and the iris. The turning rods between them were used to rotate the iris. The operating principle of this iris-based fiber probe is as follows: rotating the turnkey clockwise and counterclockwise concurrently brings the outer six collection fiber tubes (blue) together toward (close) and away (open) from the central excitation tube (orange), respectively. Consequently, $\mathrm{\Delta }s$ between the collection and excitation fibers can be altered. The movements of six collection fiber tubes are mechanically synchronized to maintain identical $\mathrm{\Delta }s$ at all the times, which means the six collection fiber tubes are located on a circle with the same diameter at all the times.

 

Fig. 1. A–D, Schematic and E–F, photos of the mechanical iris-based fiber optic probe prototype.

Download Full Size | PDF

SORS allows differentiation of Raman signals originating from layers at different depths within a diffusely scattering medium. To validate the performance of this iris-based fiber probe for SORS applications, the fiber probe was first used to characterize a two-layer sample consisting of 1 mm thick Teflon on top of 30 mm thick Delrin by varying $\mathrm{\Delta }s$ from 3 to 12 mm. The schematic of our SORS experimental setup is shown in Fig. 2, the excitation laser was a 785 nm continuous-wave (CW) laser (I0785MM0500MS, Innovative Photonic Solutions). The spectrometer consisted of a spectrograph (HoloSpec f/1.8i, Kaiser Optical Systems, Inc.) and a CCD camera (PIXIS: 400BR, Princeton Instruments Inc.). Excitation and collection fibers were Thorlabs FG200LCC (core $\text{diameter}=200\mathrm{\mu m}$) and Nufern MM-S480-22FA (core $\text{diameter}=400\mathrm{\mu m}$), respectively. Fibers passed through seven openings on the turnkey plate to link to the laser and spectrometer. The sample was placed directly under the fiber probe, so the position order from top to bottom was fibers, Teflon and Delrin in this case. The excitation power was set at 150 mW for all experiments. Six collection fibers were arranged in a vertical array through an adaptor to mount at the input port to the spectrograph (Fig. 2). Light from each collection fiber formed a horizontal spectrum on the CCD camera, and six spectra were located at six different vertical levels of CCD (i.e., separate regions of interest). In our experiments, the averaged spectrum of the six spectra from six collection fibers was taken as the final spectra.

 

Fig. 2. Schematic of our SORS experimental setup.

Download Full Size | PDF

Experimental results are shown in Fig. 3, which displays the ratio of Delrin peak ($918{\mathrm{cm}}^{-1}$) to Teflon peak ($733{\mathrm{cm}}^{-1}$) as a function of $\mathrm{\Delta }s$ after normalization and averaging, and the inset shows a few stacked spectra at different offsets along with those from pure Teflon and Delrin. The inset indicates that the Raman spectrum at $\mathrm{\Delta }s=3\mathrm{mm}$ contained substantial contribution from the top Teflon layer. As $\mathrm{\Delta }s$ increased, Delrin’s contribution (e.g., peak $918{\mathrm{cm}}^{-1}$) gradually increased with respect to Teflon (e.g., peak $733{\mathrm{cm}}^{-1}$). Hence, the ratio of Delrin to Teflon increased as $\mathrm{\Delta }s$ increased. It suggests this iris-based fiber probe could be used for deep-layer Raman detection while suppressing Raman perturbation from the surface layers.

 

Fig. 3. Ratio of Delrin peak ($918{\mathrm{cm}}^{-1}$) to Teflon peak ($733{\mathrm{cm}}^{-1}$) as a function of $\mathrm{\Delta }s$. Inset shows SORS spectra taken by the iris-based fiber probe at $\mathrm{\Delta }s=3$, 6, 9, 12 mm (upward), pure Teflon (bottom) and pure Delrin (top).

Download Full Size | PDF

Because the collection fibers of our fiber probe were movable during the iris operation, and additionally all fibers will be isolated by an optical window from the ambient in the final packaged device, there will exist a gap between the fibers (i.e., excitation and collection fibers) and the sample. As such, we investigated the gap effect on the SORS signal detection by using the same setup and the same sample (i.e., 1 mm thick Teflon on top of 30 mm thick Delrin). First we set $\mathrm{\Delta }s$ at a specific value and then increased the gap ($\mathrm{\Delta }g$) between fibers and the sample from 0 to 12.7 mm (0.5 in.) while recording the SORS spectra at different $\mathrm{\Delta }g$. In this experiment, we set $\mathrm{\Delta }s$ at 3, 5, and 7 mm, respectively, to examine the SORS spectral variations as $\mathrm{\Delta }g$ altered.

The experimental results are illustrated in Fig. 4. Figure 4A shows a few SORS spectra taken by the iris-based fiber probe as the gap between fibers and the sample varied, $\mathrm{\Delta }g=0$, 3.175, 6.35, 9.525, 12.7 mm for $\mathrm{\Delta }s=3$, 5, 7 mm, pure Teflon and pure Delrin. Figure 4B shows the ratio of Delrin peak ($918{\mathrm{cm}}^{-1}$) to Teflon peak ($733{\mathrm{cm}}^{-1}$) as a function of $\mathrm{\Delta }g$ at $\mathrm{\Delta }s=3$, 5, 7 mm, respectively. We note that (1) SORS spectra showed that Raman intensities of Delrin and Teflon as well as their signal-to-noise ratios all decreased when $\mathrm{\Delta }g$ increased; (2) when $\mathrm{\Delta }s$ was constant at a specific value, the ratio of Delrin (at bottom) to Teflon (on top) decreased monotonically as $\mathrm{\Delta }g$ increased, which indicated the Raman signals of Delrin diminished faster than Teflon did as $\mathrm{\Delta }g$ increased. For $\mathrm{\Delta }s=3\mathrm{mm}$, the ratio approached zero when $\mathrm{\Delta }g$ approached 10 mm, which resulted from the fact that the Raman signals of Delrin nearly disappeared or were unable to be discerned due to its burial under background noises; (3) the ratio curves of $\mathrm{\Delta }s=3$, 5, 7 mm were nearly parallel to each other and owned the similar upwardly convex shape; larger $\mathrm{\Delta }s$, higher curve. It suggested that it remained able to detect Raman signals of Delrin for larger $\mathrm{\Delta }s$ setups, while smaller $\mathrm{\Delta }s$ setups might already be unable to detect Delrin signals at the same amount of $\mathrm{\Delta }g$.

 

Fig. 4. A, SORS spectra taken by iris-based fiber probe as the gap between fibers and the sample, $\mathrm{\Delta }g=0$, 3.175, 6.35, 9.525, 12.7 mm (upward) for $\mathrm{\Delta }s=3$ (black), 5 (red), 7 mm (blue), pure Teflon (top orange), and pure Delrin (bottom green). B, Ratio of Delrin peak ($918{\mathrm{cm}}^{-1}$) to Teflon peak ($733{\mathrm{cm}}^{-1}$) as a function of $\mathrm{\Delta }g$ at $\mathrm{\Delta }s=3$, 5, 7 mm, respectively.

Download Full Size | PDF

The cartoons in Fig. 5 can be used to interpret the results in Fig. 4 as follows: according to the diffuse optics principle, for a given excitation-collection fiber pair, most of the SORS signals originate from a “most-sensitive region” in the sample between them within the media below. The most-sensitive region resembles a “banana” shape, as shown in Fig. 5 (blue “banana” region), whose depth is approximately 1/2 of the excitation-collection fiber separation [14,15]. In our case, when $\mathrm{\Delta }g=0$, most parts of the “banana” region were situated in the Delrin area (Fig. 5A). Hence, most Raman contributions were attributed to Delrin while less came from Teflon, resulting in the ratio of Delrin to Teflon being greater than one. As $\mathrm{\Delta }g$ increased, the “banana” region shifted upward (Fig. 5B). As such, compared with Delrin, more and more contribution was attributed to Teflon, decreasing the ratio of Delrin to Teflon. As $\mathrm{\Delta }g$ increased to a value (e.g., g2 in Fig. 5C), the “banana” region nearly fully moved away from the Delrin area, the contribution of Delrin approaching zero. Hence the ratio of Delrin to Teflon approached zero. The results in Fig. 4 suggest the SORS fiber probe could still be used to obtain high-quality Raman signals in practical applications, even with a gap between fibers and the sample.

 

Fig. 5. Most-sensitive SORS “banana” region shifts as $\mathrm{\Delta }g$ alters.

Download Full Size | PDF

Finally, since surface-enhanced Raman scattering (SERS) has exhibited great potential in biomedical imaging due to its merits such as photostable, high sensitivity, spatial resolutions, and multiplexing capability [12,16,17], we demonstrated that this fiber optic probe can be used as a platform for surface-enhanced SORS (SESORS) applications. First we mixed $3\times {10}^{10}$ SERS active gold nanoparticles (SaNPs) (in vitro Raman probes, NanoPartz Inc.) into a 10% gelatin gel (Fisher Scientific Inc.) cylinder of 5 mm diameter by 5 mm height, which was embedded into the surface of a 47 mm thick basal lean pork tissue. Then we covered the SaAuNP-containing basal pork tissue with another piece of lean pork tissue (named as tissue top) of different thicknesses. Afterward, the fiber probe was placed on top of the tissue to conduct the measurements. The center of the fiber probe vertically pointed to the center of the gel cylinder. The experimental results are displayed in Fig. 6. Figure 6A shows normalized SESORS spectra as SaNPs were covered by the tissue top with a thickness of 6, 12, 18, and 24 mm (downward) at $\mathrm{\Delta }s=20\mathrm{mm}$, SaNPs alone (top black) and tissue alone (bottom green, lean pork tissue). The characteristic Raman peaks of the SaNPs and pork tissue (i.e., ${\mathrm{CH}}_2$ wagging in the protein) are 593 and $1445{\mathrm{cm}}^{-1}$, respectively. When $\mathrm{\Delta }s=20\mathrm{mm}$, the SaNP peak ($593{\mathrm{cm}}^{-1}$) remained discernible until the tissue top was 18 mm thick. It was barely visible when the thickness of tissue top was 24 mm, which might be because it required larger offsets to “see” deeper objects, and it was beyond the capability of our current iris-based fiber probe. Figure 6B shows the ratio (solid curves) of SaNPs peak ($593{\mathrm{cm}}^{-1}$) to pork tissue protein peak ($1445{\mathrm{cm}}^{-1}$) as a function of $\mathrm{\Delta }s$ for the tissue top with 6 mm (blue), 12 mm (green), 18 mm (red) thickness and normalized SaNPs signals (dotted curves) by exposure times as a function of $\mathrm{\Delta }s$ for the tissue $\mathrm{top}=6\mathrm{mm}$ (blue), 12 mm (green), 18 mm (red). Apparently, as $\mathrm{\Delta }s$ increased or the thickness of the tissue top increased, the SERS signals (dotted curves) gradually decreased due to increased absorbing and scattering loss of both excitation and emission signals in longer traveling paths. In addition, generally, the ratio of SaNPs peak ($593{\mathrm{cm}}^{-1}$) to pork tissue protein peak ($1445{\mathrm{cm}}^{-1}$) increased first and then decreased at a turning point as $\mathrm{\Delta }s$ increased. It is possibly because that, at the beginning, the detectable depth increased, and more SERS probes were covered (i.e., excited by the laser) by the “banana” region (Fig. 5), as $\mathrm{\Delta }s$ increased, resulting in increased SERS signals as well as the increased ratio. However, since the SERS sample was a small cylinder not an infinite layer, and the “banana” region was moving as $\mathrm{\Delta }s$ varied, as $\mathrm{\Delta }s$ kept increasing, the excitation photons progressively bypassed the SERS probes. Hence the SERS signals and the ratio decreased correspondingly, resulting in a turning point in the ratio curves between the increased (near offset) and decreased (far offset) ratio parts.

 

Fig. 6. SESORS experimental results of SaNPs (SERS probe) embedded in pork tissue at different depths. A, Normalized SESORS spectra taken by iris-based fiber probe as SaNPs were covered by the tissue $\mathrm{top}=6$, 12, 18, 24 mm thick (downward) at $\mathrm{\Delta }s=20\mathrm{mm}$, SaNPs alone (top black) and tissue alone (bottom green). B, Ratio (solid curves) of SaNPs peak ($593{\mathrm{cm}}^{-1}$) to pork tissue protein peak ($1445{\mathrm{cm}}^{-1}$) as a function of $\mathrm{\Delta }s$ for the tissue $\mathrm{top}=6\mathrm{mm}$ (blue), 12 mm (green), 18 mm (red) thick and normalized SaNPs signals (dotted curves) by exposure time as a function of $\mathrm{\Delta }s$ for the tissue $\mathrm{top}=6\mathrm{mm}$ (blue), 12 mm (green), 18 mm (red). Integration time of CCD camera for each data point was shown next to the corresponding point in the same color.

Download Full Size | PDF

In summary, we demonstrated a mechanical iris-based fiber optic probe with adjustable spatial offsets, which can be used for SORS and SESORS applications. In addition, we studied the gap effect between the probe and sample, which illustrates the SORS fiber probe could capture high-quality Raman signals in practical applications even with a gap between fibers and the sample. Compared with conventional SORS fiber probes, this iris-based design offers continuous and customizable spatial offsets between the illumination and the collection fibers and, thus, can be exploited for samples with various shapes and depth. It requires fewer collection units while detecting large/deep samples. Therefore the current probe design could potentially provide simpler and operating procedures and a design/cost-efficient solution for both traditional SORS and advanced SESORS system. Furthermore, this type of engineering invention might potentially have an impact on the scientific community in the area of diffusion-based spectroscopy (not limited to Raman spectroscopy).