1.

Introduction

Melanoma is acknowledged as the most dangerous form of skin cancer, accounting for 90% of the deaths associated with cutaneous cancers.¹ According to a recent report,¹ the incidence of melanoma is increasing worldwide with the regional incidence rate of up to 10 to 20, 20 to 30, and 50 to 60 per 100,000 populations in Europe, USA, and Australia, respectively. The standard treatment of an operable melanoma involves surgical excision with a safety margin. The safety margin is required in order to completely remove not only the primary cancer but also melanoma cells that might have spread into the surrounding skin.² The determination of a proper safety margin is critical during surgical treatment because a too-large safety margin leaves functionally and cosmetically poor outcome, whereas a too-small safety margin risks incomplete removal of cancer or cancer cells. For the determination of safety margin, Mohs micrographic surgery is widely adopted in dermatology, in which the process of cancer excision and histologic examination is repeated until no cancer cells are found from the excised layer.³ An accurate interpretation of the histologic examination of excised layer during the surgery is critical. One of the shortcomings of the Mohs procedure is the prolonged operation time due to repeated histology test during the surgery, which increases the risk of complications such as wound infection.⁴ Thus an alternative or supporting technique to accurately and rapidly determine the cancer region and appropriate safety margin is highly desired.

Laser-induced breakdown spectroscopy (LIBS) is a technique, in which the concentrations and distributions of constituent elements of a sample are analyzed by measuring the emission from laser produced plasma. LIBS measurement can be completed in a short time (on the order of seconds or shorter) and performed in air with little or no sample preparation. Owing to the speed and high-spatial resolution achievable with LIBS, the application of LIBS for the classification of biological samples, such as liver,⁵ breast and colorectal tissues,⁶ and nerve and fat of pig heads,⁷ carious teeth⁸ has been investigated by many researchers, demonstrating the potential as a fast classification method. Although LIBS is a relatively new technique, other spectroscopic methods have been widely investigated for detection and imaging of skin cancer, such as infrared spectroscopy⁹ or Raman spectroscopy,¹⁰ which provides the molecular information of a sample. On the contrary to these molecular spectra-based techniques, LIBS provides the atomic spectra of the constituent elements. Thus it may be used independently or complementarily with a molecular spectroscopic method for the diagnosis of skin cancer. For instance, a recent study reported the combination of LIBS and Raman spectroscopy for the classification of bacteria.¹¹ For the case of melanoma, we reported in our recent study that melanoma and surrounding dermis could be distinguished by LIBS with high sensitivity using Mg and Ca emission peaks as the biomarkers.¹² Although it was an early stage of research yet, this study confirmed that LIBS can be an effective method to identify melanoma from normal tissue.

Although the spectral information itself in our previous study¹² is useful in determining the pathological state of the tissue, an elemental mapping image based on LIBS spectral intensity can provide a direct information to assess the skin cancer region and required margin during a surgical excision. In the case of Mohs micrographics surgery, LIBS elemental mapping may be applied to a sectioned tissue to rapidly create a map showing the cancer and normal tissue regions. Although elemental mapping of solid samples by LIBS has been investigated by many researchers previously,^13–16 LIBS elemental mapping of biological samples is relatively new. Recently, Motto-Ros et al. published a number of papers for LIBS imaging of biosamples. For example, Sancey et al.¹⁷ reported that the elements in mouse kidney embedded with Gd-based nanoparticles could be imaged with $10\text{-}\mu \mathrm{m}$ resolution using a nanosecond (ns) 266 nm Nd:YAG laser-based LIBS. Also, Moncayo et al.¹⁸ took LIBS image of different types of paraffin embedded human tissue. Bonta et al.¹⁹ used LIBS for elemental mapping of human tumor in conjunction with laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS), where ns-LIBS was used for the mapping of major and minor elements while LA-ICP-MS was for trace elements. As shown in these studies, a nanosecond infrared or ultraviolet laser was commonly adopted in LIBS elemental mapping. However, a femtosecond-laser-based LIBS (fs-LIBS) can provide several advantages over ns-LIBS such as smaller heat-affected zone, lower ablation threshold, potentially finer spatial and depth resolution, and lower background emission.^20–22 On the other hand, changes in sample condition, e.g., water content variation, can also affect the LIBS signal for biological samples.^6,23 Therefore, for LIBS analysis of biological samples, it is desirable to perform the analysis under the conditions as close as possible to the original samples.

In this work, we report the results of fs-LIBS analysis and elemental mapping of cryosectioned melanoma sample as an imaging technique to differentiate the tumor and dermis regions. The sample was prepared by frozen tissue technique, the same procedure as in Mohs’ surgery so that the LIBS elemental mapping image could be directly compared with the haematoxylin and eosin (H&E) staining image and evaluated for accuracy. It is shown that the skin cancer and dermis regions are clearly distinguished on the LIBS elemental mapping image and consistent with the histologically determined ones, demonstrating the feasibility of LIBS as a useful technique for faster determination of skin cancer regions.

2.

Materials and Methods

2.1.

Sample Preparation

For the preparation of melanoma sample for LIBS imaging, $1\times {10}^6$ B16/F10 cells, a well-known murine melanoma cell line, in $100\mu \mathrm{L}$ of 0.9% NaCl solution were implanted in the back skin of 6 to 8 weeks old female SKH-1 mice. After 10 days from the implantation, melanoma lesions were developed on the mice as shown in Fig. 1(a). Then the skin tissues including both melanoma lesion and normal dermis were excised from the mice by a procedure shown in Fig. 1(b) for both LIBS elemental mapping and histologic analysis. For LIBS analysis, the excised tissue was first frozen and sectioned to $50\text{-}\mu \mathrm{m}$ thickness using a cryostat (Cryocut 3000, Leica Biosystems, Nussloch, Germany) and then placed on a piece of silicon wafer for ablation. The frozen sectioning is a technique being used for histologic examination in Mohs surgery.²⁴ Figure 1(c) shows an optical image ($50\times$; SZ1145TRPT, Olympus Korea Co., Republic of Korea) of thereby produced tissue section over which LIBS elemental mapping was performed. The histology sample was also collected from the same excised tissue but was sectioned to $4\text{-}\mu \mathrm{m}$ thick slices after fixing in 4% paraformaldehyde and embedding in paraffin, and stained with H&E.

Fig. 1

(a) Clinical image of a mouse after 10 days of subcutaneous melanoma cell implantation, (b) procedure for LIBS sample preparation, and (c) optical image of the melanoma tissue section for LIBS elemental mapping.

Note that the sectioned LIBS sample and histology sample were not the same. Instead, they were collected from the same excised tissue within a few tens of micrometer thickness range. Therefore, it was assumed that the cancer profiles on the LIBS sample and histology sample were the same.

All procedures for sample preparation were approved by the Animal Care and Use Committee of Gwangju Institute of Science and Technology and performed in accordance with NIH guidelines (GIST-05).

2.2.

LIBS Measurement

Figure 2 shows a schematic of the fs-LIBS set up used in this study. The ablation of sectioned tissue sample was carried out using a diode-pumped ytterbium femtosecond laser ($\lambda =1030\mathrm{nm}$, $\tau =550\mathrm{fs}$; s-pulse HP, amplitude systems). The laser beam was focused to a spot diameter of about $65\mu \mathrm{m}$ on the sample surface using an objective lens ($\text{magnification}=5\times$) and the laser pulse energy was set to $250\mu \mathrm{J}$, in which the corresponding laser fluence became about $7.42{\mathrm{J}/\mathrm{cm}}^2$. On the laser spot, argon gas was injected in the direction of about 30 deg from the sample surface at the flow rate of $15\mathrm{L}/\min$ for signal enhancement. Argon is often used as the buffer gas during LIBS experiments, mainly for signal enhancement. We used argon only during the measurement and the experiment was carried out in a well-ventilated environment. So there was little danger of asphyxiation.

Fig. 2

Schematic diagram of the LIBS system.

An intensified charge coupled device (ICCD; PI-MAX3, Princeton Instruments, $\text{grating}=2400\text{grooves}/\mathrm{mm}$, $\text{spectral width}=\sim 60\mathrm{nm}$, $\text{resolution}=\sim 0.1\mathrm{nm}$, and $\text{gain}=5$) with high sensitivity was employed to detect the plasma emission because the emission intensity of LIBS plasma decreases rapidly at reduced spot diameter. The gate width of the ICCD spectrometer was set to $1\mu \mathrm{s}$ with $0.1\mu \mathrm{s}$ delay. Previously, we reported the LIBS data of melanoma and dermis measured using a nanosecond laser and a CCD spectrometer over a wide spectral range (200 to 900 nm).¹² The results showed that among the various detected elements (C, Mg, Ca, Na, H, O, N, and Cl), Mg and Ca showed significantly higher intensities in melanoma than in dermis, whereas other elements showed little difference. It was also shown that the Mg and Ca signals appeared at Mg (II) 279.553 nm, Mg (II) 280.170 nm, Ca (II) 393.366 nm, and Ca (II) 369.847 nm. The ICCD spectrometer used in this study had a limited spectral width ($\sim 60\mathrm{nm}$), which could not cover the spectral regime including both elements. Thus either of the Mg or Ca peaks had to be measured in experiments, and we selected the Mg peak because it was well isolated and had low background.

3.

Results

3.1.

LIBS Signal Intensity

Figure 3 shows the LIBS spectra measured from the pigmented melanoma and the surrounding tissue of the sectioned ablation sample [see Fig. 1(c) for the LIBS measurement locations on the pigmented melanoma (red dot) and surrounding tissue (blue dot)]. It is seen that the intensity of the carbon peak, C(I) 247.856 nm, is nearly the same for both pigmented melanoma and surrounding tissue. This result is consistent with the earlier report that carbon content is independent of malignancy status of tissue.⁶ In contrast, the intensities of magnesium peaks, Mg(II) $279.553+280.270\mathrm{nm}$ and Mg(I) 285.213 nm, of the pigmented melanoma were found to be significantly higher than those in the surrounding tissue, which reconfirmed our previous results.¹² Note that the silicon peak Si(I) 236.128 nm, obtained by ablating a bear silicon wafer, is also shown in Fig. 3 in order to show that the measured carbon and magnesium signals were not interfered by the substrate.

Fig. 3

Intensities of LIBS spectra measured from the pigmented melanoma, surrounding tissue, and Si wafer substrate.

3.2.

Elemental Mapping

LIBS elemental mapping was performed by ablating the sectioned tissue on silicon wafer in Fig. 4(a) along a $25\times 25$ spot array (total 625 spots). Since the laser spot diameter was about $65\mu \mathrm{m}$, the spot-to-spot distance was kept $70\mu \mathrm{m}$ during mapping so that no overlap between spots may have taken place. It took about 10 min to complete the mapping at the repetition rate of 1 Hz, but the mapping time may be reduced substantially by increasing the laser repetition rate.

Fig. 4

(a) CCD image of the melanoma tissue section on silicon wafer before ablation, and the LIBS intensity maps of (b) C(I) 247.856 nm and (c) Mg(II) $279.553+280.270\mathrm{nm}$ lines, and (d) the map of Mg(II)/C(I) intensity ratio.

The intensity of each spectral line in Fig. 3 was calculated by integrating the area under the peak using MATLAB. Also since the Mg(II) 279.553 nm and Mg(II) 280.270 nm lines are heavily overlapped and these lines have the same upper and lower level energies, the intensities of these lines were added and represented by a single value. Note that the intensities of spectral lines of an element having the same upper and lower level energies change with similar tendency during LIBS measurement.²⁵ Figures 4(b) and 4(c) show the LIBS elemental maps constructed with the C(I) 247.856 nm and Mg(II) $279.553+280.270\mathrm{nm}$ intensities, respectively, using Origin. It is clear from Fig. 4(b) that the intensity of C(I) 247.856 nm line is little or weakly influenced by the existence of melanoma, showing only a slight increase of intensity in the pigmented melanoma region. This result reconfirms that carbon concentration is hardly influenced by the cancer status of tissue. On the contrary, the intensity of Mg(II) $279.553+280.270\mathrm{nm}$ line in Fig. 4(c) shows a strong contrast between the pigmented melanoma region and the surrounding tissue. It is known that melanoma cells tend to infiltrate into the surrounding tissue. Considering the common existence of border region, in which isolated melanoma cells are frequently found, it is uncertain if the observed sharp contrast of Mg signal intensity in Fig. 4(c) reflects the actual variation of elemental concentration across the border of pigmented melanoma and surrounding tissue regions. In addition, the intensity of Mg(II) $279.553+280.270\mathrm{nm}$ line within the pigmented melanoma region appears nonuniform, which could be due to either true Mg concentration change or plasma fluctuation.

To reduce the possible variation of emission intensity due to plasma fluctuation, the intensity map was reconstructed by normalizing the Mg(II) $275.553+280.270\mathrm{nm}$ intensity by that of the C(I) 247.856 nm line. The resulting map in Fig. 4(d) shows a gradual decrease of normalized intensity, $I_{\mathrm{Mg}(\mathrm{II})279.553+280.270\mathrm{nm}}/I_{\mathrm{C}(\mathrm{I})247.856\mathrm{nm}}$, toward the surrounding tissue. The contours of the pigmented melanoma and the surrounding tissue in this elemental map, Fig. 4(d), seem to closely match the morphological features observed by the optical image in Fig. 4(a) [or that in Fig. 1(c)].

4.

Discussion

To interpret the LIBS elemental mapping results, the intensity ratio map ($I_{\mathrm{Mg}}/I_{\mathrm{C}}$) in Fig. 4(d) was compared with the histology image shown in Fig. 5. Again, note that the histology tissue section was not the same as the ablation tissue section, but the histology sample was collected from the same excised tissue within a few tens of micrometer thickness so that the observed features of melanoma may remain as close as possible between the sectioned histology sample and sectioned LIBS sample. The histologic features in Fig. 5 reveal mainly three different regimes, i.e., the pigmented melanoma, surrounding tissue locally containing nonpigmented melanoma cells (or mixed inflammatory cells), and normal dermis. The size and location of these three regimes appear to match the high-intensity region (shown in red color, $I_{\mathrm{Mg}}/I_{\mathrm{C}}=\sim 5$ to 100), medium intensity region (yellow and orange color, $I_{\mathrm{Mg}}/I_{\mathrm{C}}=\sim 3.5$ to 5; greenish color, $I_{\mathrm{Mg}}/I_{\mathrm{C}}=\sim 2$ to 3.5), and the low-intensity region (bluish color, $I_{\mathrm{Mg}}/I_{\mathrm{C}}<2$) represented by the contours in the LIBS elemental map, Fig. 4(d). The $I_{\mathrm{Mg}}/I_{\mathrm{C}}$ value over the elemental map shows a gradual decrease toward the dermis as expected. However, the location of tumor border is presumed to be near the light green and dark blue area on the basis of the histology image in Fig. 5 rather than the intensity ratio values. Note that the laser spot diameter in this study was about $65\mu \mathrm{m}$, which is much greater than the size of melanoma cells ranging from 10 to $30\mu \mathrm{m}$.²⁶ Therefore, the elemental mapping image in Fig. 4 cannot reflect the microscopic or cell-level variations of the cancer (e.g., pagetoid spread or small nests). Nonetheless, since the overall signal intensity within the melanoma cell containing areas is supposedly higher than that of dermis, it is considered that the variation of signal intensity still represents a meaningful information about cancer cell containing area. In principle, a further reduction of laser spot diameter close to cell size can be possible if a higher magnification focusing optics and a higher sensitivity detector are employed, allowing an elemental mapping at an improved spatial resolution.

Fig. 5

Histologic features of the skin section showing pigmented melanoma, surrounding tissue containing nonpigmented melanoma cells, mixed inflammatory cells, and overlying dermis.

In conjunction with the histology results, the variation of Mg(II)/C(I) intensity ratio over the LIBS elemental map in Fig. 4(d) may be explained by the role of Mg in metabolism. In general, melanoma has high cellular density as observed in Fig. 5 and shows active metabolism with robust protein biosynthesis. In addition to the melanoma cells, diverse forms of inflammatory cells observed between the pigmented melanoma and normal dermis also have active metabolism and produce a variety of enzymes.²⁷ On the other hand, Mg is known to be an essential component for protein biosynthesis^28,29 and a cofactor of enzymes for general metabolic pathways and nucleic acid biochemistry.^30,31 Thus compared with dermis which mainly consisted of collagen bundles and fibroblasts, the cell density and metabolic activity within the pigmented melanoma are expected to be significantly higher than those of dermis. Similarly, the cell density and metabolic activity would be high in the immediate vicinity of pigmented melanoma and drop gradually as approached the dermis. The intensity map of Mg(II)/C(I) in Fig. 4(d), which shows the highest Mg concentration in the pigmented melanoma region with a gradual decrease toward the surrounding tissue, is understood to reflect these differences in cell density and metabolic activity depending on the cancer status of tissue over these regions. On the contrary, the difference in pigment density between melanoma and dermis is considered to have little influence in the signal intensity change, because, in our previous study for LIBS analyses of melanoma with pellet and excised tissue samples, Mg intensity was found to be also higher in melanoma even after the absorption property difference between pigmented and nonpigmented regions was compensated by normalization.¹²

5.

Conclusion

It is demonstrated that by fs-LIBS, a detailed elemental map showing the changes of LIBS signal intensity over melanoma and surrounding tissue could be established. The variation of normalized LIBS signal intensity over the melanoma and surrounding tissue closely matched the morphological and histologic features of the lesion. The results in this study showed the potential of LIBS as a method to differentiate melanoma and dermis. Since the spatial resolution can be further improved with appropriate lasers and optics, it is considered that with clinical verification by more histologic comparison and other modalities, fs-LIBS mapping of melanoma can provide a realistic option for fast histologic interpretation of melanoma during micrographic surgery. Although this study was for LIBS analysis of melanoma cryostat sections, the same approach may be applicable for other types of skin cancer.