2.1.

Instrumentation

Figure 1 shows the schematic diagram of the prototype LCE system developed by Optiscan Imaging Ltd. (Victoria, Australia) for in vivo fluorescence imaging of human tissue. A bandpass filter in front of the argon ion $\left({\mathrm{Ar}}^{+}\right)$ laser selectively passes the $488\text{-}\mathrm{nm}$ wavelengths while rejecting all plasma emissions. The selected $488\text{-}\mathrm{nm}$ laser beam then passes through a beamsplitter and is focused into a single-mode fiber via two reflecting mirrors and a focusing lens. The beamsplitter also serves as a reflecting mirror for the return fluorescence signal and directs it to a photomultiplier tube (PMT) via a mirror, as shown in the diagram. A high-rejection $505\text{-}\mathrm{nm}$ long-pass filter further absorbs the backscattered laser beam while transmitting the fluorescence signal to a sensitive PMT for detection.

Fig. 1

Schematic diagram of the laser confocal endomicroscope (LCE) system with a handheld rigid probe that was used for fluorescence imaging of the oral cavity. The inset shows details within the probe tip containing the miniaturized scanning mechanism.

The handheld rigid probe consists of a handle with a focusing knob and an $8\text{-}\mathrm{mm}$ $(Ø)$ metal tube (see inset) that houses a miniaturized tuning fork. The output end of the single-mode fiber is attached to one of the arms of the tuning fork. A driving magnetic coil induces a $700\text{-}\mathrm{Hz}$ resonant oscillation in the tuning fork, providing a fast $x$ scan, while a second coil causes the entire tuning fork to pivot in the $y$ direction at a slower scanning rate of 1 to $2\mathrm{Hz}$ . An objective lens is used both to focus the excitation laser into the tissue as well as to collect the fluorescence signal. These give a maximum imaging area of approximately $390\mathrm{\mu }\mathrm{m}\times 390\mathrm{\mu }\mathrm{m}$ . Due to the small size of the single-mode fiber core, the fiber tip acts as both a point source and a point detector, thereby facilitating confocal detection, i.e., only fluorescence from a small volume at the focal plane is collected. $Z$ sectioning can be achieved by rotating a knob on the handle, which moves the entire scanning mechanism (including both magnetic coils and the tuning fork) axially along the tube. This provides a mechanical range of almost $300\mathrm{\mu }\mathrm{m}$ below the surface. A maximum power of $<0.5\mathrm{mW}$ can be obtained at the surface of the tissue.

Synchronization between the electronic signals from the PMT and the sampling positions is provided by the main control unit (MCU), which communicates with a personal computer (PC) via a frame grabber card and a COM port. The frame grabber in the MCU outputs one frame of $512\times 512$ pixels image per s to the PC. The pixel resolution in a $512\times 512$ pixels image is $0.8\mathrm{\mu }\mathrm{m}$ per pixel.

2.2.

Imaging with ALA in Human Subjects

A 0.4% solution of ALA was freshly prepared by dissolving 5-aminolevulinic acid hydrochloride (Medac, Germany) in phosphate buffered solution (PBS) with the pH adjusted to between 6.5 and 7.4. Three healthy volunteers and three patients with SCC of the tongue were recruited for the pilot study following informed consent. Topical application in the oral cavities of the subjects was achieved by $15\min$ of continuous rinsing under supervision, followed by an incubation period of at least $30\min$ . Imaging using the LCE was carried out between 30 to $120\min$ after administration of ALA. Excised tissue of the SCC tongue was sectioned and processed for hematoxylin and eosin (H&E) staining.

2.3.

Imaging with ALA in Animal Models

Male Sprague-Dawley rats (4 to 6 weeks old) were used as animal models. The animals were anesthetized for survival procedures, and topical application was carried out by the insertion of cotton buds soaked in freshly prepared 0.4% ALA solution (preparation described in Sec. 2.2) into the oral cavity for $15\min$ , followed by an incubation period of $30\min$ . After the incubation period, the rats were sacrificed by an overdose of carbon dioxide, and the tongue and buccal cavity tissue were excised for imaging using the LCE. Images obtained using the LCE were compared to images obtained using a conventional benchtop confocal laser scanning microscope (Carl-Zeiss LSM Meta 510) operating with a $488\text{-}\mathrm{nm}$ excitation source.

2.4.

Imaging with Fluorescein Sodium in Animal Models

Fluorescein sodium was also used as a contrast agent in 4-to-6 week-old male Sprague-Dawley rat models. The animals were anesthetized for survival procedures, and topical application was carried out by the insertion of cotton buds soaked in freshly prepared 0.1% fluorescein sodium solution (Sigma-Aldrich) into the oral cavity for $15\min$ . The rats were then sacrificed by an overdose of carbon dioxide, and the tongue and buccal cavity tissue were excised for imaging.

2.5.

Use of Pharmasolve in Animal Models

The pharmaceutical-grade solvent N-methyl-pyrrolidone, trade name Pharmalsolve (ISP Technologies, Inc., Wayne, New Jersey), was used with 5-ALA and fluorescein sodium to enhance the penetration depths of these drugs in the oral cavity of animal models. A 5% solution of Pharmasolve was prepared in PBS and mixed with a solution of 0.4% ALA (prepared as described in Sec. 2.2). Topical application of the ALA-Pharmasolve solution was carried out in the rat oral cavity as described in Sec. 2.3. Likewise, a 5% Pharmasolve solution was mixed with a 0.1% fluorescein sodium solution (preparation described in Sec. 2.3) and topically applied to the rat oral cavity as described in Sec. 2.4.

2.6.

Carcinogen-Induced Model of Oral Cavity SCC

The water-soluble carcinogen 4-nitroquinoline-N-oxide (4-NQO) was used to induce a rat model of oral cavity SCC. 4-NQO (Sigma-Aldrich) was dissolved in acetone and diluted in water to make up a 0.002% solution. The 4-NQO solution was then fed to the rats as drinking water for 24 weeks. The rats were allowed to drink freely of the 4-NQO solution. After 24 weeks, a solution of 0.4% ALA and 5% Pharmasolve was prepared as described in Sec. 2.5 and topically applied to the oral cavity of the 4-NQO rats. Fluorescence imaging of the oral cavity was carried out after $30\min$ , as described in Sec. 2.3. The excised tissue was sectioned and processed for H&E staining.

3.1.

Imaging the Human Oral Cavity

In a pilot study, ALA-induced PpIX fluorescence imaging of the human tongue was carried out in healthy volunteers and patients with malignant SCC lesions of the tongue. A 0.4% ALA solution was administered topically, and fluorescence imaging was carried out using the LCE after an incubation period of at least $30\min$ . Figure 2 shows ALA-induced PpIX fluorescence LCE images of a normal human tongue in a healthy volunteer. Figures 2a, 2b, 2c show images captured $45\min$ , $90\min$ , and $120\min$ after ALA administration, respectively. Clear images of the tongue could be obtained as early as $45\min$ after ALA administration. Imaging after longer time periods of up to $120\min$ after ALA administration did not offer any advantage over imaging at an earlier time point. The images at later time points $90\min$ and $120\min$ carry more noise and are not as sharp as the image at $45\min$ . This is probably due to clearance of PpIX, leading to reduced fluorescence signals at later time points.

Fig. 2

ALA-induced PpIX fluorescence LCE images of a normal human tongue in a healthy volunteer. (a), (b), and (c) show images captured $45\min$ , $90\min$ , and $120\min$ after ALA administration, respectively. Clear images of the tongue could be obtained as early as $45\min$ after ALA administration. The scale bars in the images represent $50\mathrm{\mu }\mathrm{m}$ in these images, with approximate fields of view of $390\mathrm{\mu }\mathrm{m}\times 390\mathrm{\mu }\mathrm{m}$ .

Figure 3a shows an ALA-induced PpIX fluorescence LCE image obtained from a patient with SCC of the tongue, while Fig. 3b shows a white light microscope image of an H&E–stained section of an SCC tongue. In image (b), the SCC region is characterized by dark nuclei (red arrows) and slightly less dark cytoplasm. The surrounding tissue has many inflammatory cells with smaller nuclei (black arrows). The fluorescence image in (a) shows small structures of similar size that are probably the inflammatory cells (white arrows). Larger structures (red arrows) represent the squamous epithelium of the SCC.. Thus, LCE images of normal (Fig. 2) and SCC [Fig. 3a] tongues appear distinctly different, allowing us to distinguish between normal and lesion tissue using ALA-induced PpIX fluorescence imaging. This shows the potential of the LCE for diagnosis of oral cavity lesions such as SCC of the tongue.

Fig. 3

(a) An ALA-induced PpIX fluorescence LCE image obtained from a patient with SCC of the tongue. (b) A white light microscope image of a hematoxylin and eosin (H&E) stained section of a human SCC tongue. In (b), the SCC region is characterized by dark nuclei ([red] arrows) and slightly less dark cytoplasm. The surrounding tissue has many inflammatory cells with smaller nuclei (black arrows). The fluorescence image in (a) shows small structures of similar size that are probably the inflammatory cells (white arrows). Larger structures ([red] arrows) represent the squamous epithelium of the SCC.

3.2.

Imaging the Rat Oral Cavity with ALA

ALA-induced PpIX fluorescence endomicroscopic imaging of the oral cavity was also carried out in rat models. ALA-induced PpIX fluorescence images of the dorsal surface of the normal rat tongue obtained using the LCE and a conventional CLSM $30\min$ after topical application of 0.4% ALA solution in the rat oral cavity are shown in Figs. 4a and 4b , respectively. Fungiform papillae appear broad and round and are indicated by dotted white arrows in both images. Conical filiform papillae are more numerous and are indicated by solid white arrows in both images. Thus a comparison of (a) and (b) shows that the LCE can capture ALA-induced PpIX fluorescence images of structures on the rat tongue similar to those that can be obtained using a conventional benchtop CLSM.

Fig. 4

ALA-induced PpIX fluorescence images of the dorsal surface of the normal rat tongue obtained $30\min$ after topical application of 0.4% ALA solution in the rat oral cavity obtained using the LCE (a) and a conventional benchtop CLSM (b). Fungiform papillae appear broad and round and are indicated by dotted white arrows in both images. Conical filiform papillae are more numerous and are indicated by solid arrows in both images.

3.3.

Imaging the Rat Oral Cavity with Fluorescein Sodium

As a comparison, ALA-induced PpIX fluorescence images of the rat oral cavity were compared to fluorescence images obtained using fluorescein sodium, a fluorophore that is commonly used for fluorescence imaging. A 0.1% fluorescein solution was topically applied in the rat oral cavity. The tongue and buccal tissue in the rat oral cavity were imaged using the LCE after an incubation period of at least $30\min$ . Figure 5a shows an LCE fluorescence image of the dorsal surface of the rat tongue. Filiform papillae (dotted white arrows) can be easily identified in this image. These structures are similar to those observed in the ALA-induced PpIX fluorescence image of the rat tongue obtained after ALA application, shown in Fig. 4a. Fluorescence images of the surface of the rat buccal tissue obtained after topical application of fluorescein using the LCE (b) and a conventional benchtop CLSM (c) both show squamous epithelial cells of the surface of the mucosa (solid white arrows in both). These are the mature mucosal squamous cells at the surface of the buccal epithelium. They form a flat pavement-like surface with distinct cell boundaries that is typical of the surface of squamous mucosa.

Fig. 5

(a) An LCE fluorescence image of the dorsal surface of the rat tongue obtained $30\min$ after topical application of fluorescein. Filiform papillae (dotted white arrows) can be identified, similar to those observed in Fig. 4a. Fluorescence images of the surface of the rat buccal tissue obtained after topical application of fluorescein using the LCE (b) and a conventional benchtop CLSM (c) both show squamous epithelial cells of the surface of the mucosa (solid white arrows in both). These are the mature mucosal squamous cells at the surface of the buccal epithelium. They form a flat pavement-like surface with distinct cell boundaries that is typical of the surface of squamous mucosa.

3.4.

Use of Pharmasolve in Animal Models

The pharmaceutical-grade solvent Pharmasolve was used with ALA and fluorescein in rat models to enhance the penetration of these drugs in the rat oral cavity. Pharmasolve was mixed with solutions of 0.4% ALA or 0.1% fluorescein sodium, and the solution mixture was topically applied to the rat oral cavity. Fluorescence imaging of the rat oral cavity was carried out using the LCE after an incubation period of at least $30\min$ . Figure 6a shows the fluorescence image of rat buccal mucosa obtained at the surface of the tissue after topical application of a mixture of fluorescein sodium and Pharmasolve. Figure 6b shows an image of the same tissue area obtained at a subsurface depth of about $60\mathrm{\mu }\mathrm{m}$ , indicating that fluorescein, when mixed with Pharmasolve, penetrated about $60\mathrm{\mu }\mathrm{m}$ into the buccal mucosa. By comparison, in rats that received topical application of fluorescein alone, no images of the buccal mucosa can be obtained below the surface. Thus, the use of Pharmasolve enhanced the penetration depth of fluorescein in the rat buccal mucosa.

Fig. 6

LCE fluorescence images of the rat buccal mucosa obtained at the surface of the tissue (a) and at a subsurface depth of about $60\mathrm{\mu }\mathrm{m}$ (b) after topical application of a solution mixture of fluorescein and Pharmasolve.

Preliminary experiments on the rat tongue were also carried out to estimate the maximum subsurface depth from which images can be obtained after topical application of ALA and fluorescein sodium with and without the addition of Pharmasolve. This maximum depth was estimated by turning the focusing knob to focus in increasingly deeper layers below the tissue surface until no discernible features can be obtained in the image. The depth estimates showed that the use of Pharmasolve with ALA increased the maximum image depth on the rat tongue by up to $80\mathrm{\mu }\mathrm{m}$ , from 195 $\mu$ m without to $275\mathrm{\mu }\mathrm{m}$ with Pharmasolve. Thus, our preliminary results show that the use of Pharmasolve with ALA and fluorescein enhanced the penetration depths of both drugs and enabled us to obtain images from deeper layers of the tissue in animal models.

3.5.

Carcinogen-Induced Model of SCC

The water-soluble carcinogen 4-NQO was used to induce a rat model of oral cavity SCC. After 24 weeks, a small white patch was observed in the tongue of the rat, resembling early SCC of the tongue. Prior to imaging the SCC tongue, a solution mixture of 0.1% fluorescein and 5% Pharmasolve was topically applied to the oral cavity of the rat, followed by $30\min$ of incubation. Figure 7a shows an LCE fluorescence image of the SCC rat tongue. This image shows disorganized structures in the SCC rat tongue. In contrast, Fig. 7b shows the fluorescence image taken of a normal (non-SCC) rat tongue also with a topically applied mixture solution of 0.1% fluorescein and 5% Pharmasolve. This image shows the same organized structures of filiform papillae as were observed in fluorescence images of the normal rat tongue obtained using ALA [Fig. 4a] and fluorescein [Fig. 5a]. Thus, we are able to observe morphological differences between the normal and SCC rat tongues from fluorescence images obtained using the LCE.

Fig. 7

(a) An LCE fluorescence image of a 4-NQO carcinogen-induced SCC rat tongue after topical application of a mixture of fluorescein and Pharmasolve. Disorganized structures are observed in the SCC rat tongue. In contrast, (b) shows the fluorescence image taken of a normal rat tongue also with a mixture solution of fluorescein and Pharmasolve topically applied. This image shows the same organized structures of filiform papillae as were observed in fluorescence images of the normal rat tongue obtained using ALA [Fig. 4a] and fluorescein [Fig. 5a]. Thus, LCE fluorescence imaging allows us to distinguish morphological differences between the normal and the SCC rat tongues.

Fig. 8

(a) A dysplastic region within an H&E stained section of a 4-NQO carcinogen-induced SCC rat tongue. (b) An adjacent normal region within the same specimen. The dysplastic region is characterized by a thicker, proliferating epithelium. The proliferating keratinocytes have scant cytoplasm and relatively large nuclei that are hyperchromatic, with prominent nucleoli. Perinuclear vacuolation is noted around a few of the larger dysplastic nuclei (white solid arrow). Some necrotic dysplastic cells are present, with eosinophilic cytoplasm (white dotted arrow). Dysplastic nuclei appear in the most superficial thickened layer (black arrows). All of these features are lacking in the normal region in (b).

Tissue from the SCC rat tongue was also processed for H&E staining to provide histological characterization. Figure 8a shows a dysplastic region within an H&E stained section of the SCC rat tongue. Figure 8b shows an adjacent normal region within the same specimen. The dysplastic region is characterized by a thicker, proliferating epithelium. The proliferating keratinocytes have scant cytoplasm and relatively large nuclei that are hyperchromatic, with prominent nucleoli. Perinuclear vacuolation is noted around a few of the larger dysplastic nuclei (white solid arrow). Some necrotic dysplastic cells are present, with eosinophilic cytoplasm (white dotted arrow). Dysplastic nuclei appear in the most superficial thickened layer (black arrows). All of these features are lacking in the normal region in (b).