Determination of Melanoma Lateral and Depth Margins: Potential for Treatment Planning and Five-Year Survival Rate

Cutaneous malignant melanoma (CMM) is a serious type of cancer accounting for 75% of all deaths associated with skin cancer (Jerant et al., 2000). CMM incidence has dramatically increased in the past few decades and, recently, approximately 160,000 new cases of CMMs are diagnosed worldwide each year (Ries et al., 2003). In 2010, the American Cancer Society estimated that 68,130 cases of melanoma (38,870 males; 29,260 females) and 8,700 melanoma deaths (5,670 males; 3,030 females) were expected in the United States (American Cancer Society (ACS), 2010). In the United States, the lifetime risk for developing CMM has increased from 1 in 1500 in 1930 to 1 in 50 in 2010 (ACS, 2010; King, 2004). Proper staging of CMM is crucial for defining prognosis and for determining the optimal treatment approach. Several cancer staging systems are being used worldwide. One of the most common staging systems is the tumor-node-metastasis (TNM) classification established by the American Joint Committee on Cancer (AJCC) (Balch et al., 2001). The TNM system classifies CMM in three categories: (1) the size and extent of the primary tumor (T), (2) the involvement of regional lymph nodes (N) and, (3) the presence or absence of distant metastasis (M), determining CMM clinical Stage I, II, III, or IV. To remain current and relevant to clinical practice, the TNM classification is updated periodically based on advances in understanding of cancer prognosis. The latest revision of TNM (presented in the 7th edition of the AJCC Cancer Staging Manual) is applied for cases diagnosed on or after January 1, 2010 (Edge et al., 2010). The CMM invasion depth known as the Breslow thickness (Breslow, 1970) in T category is the single most important factor for CMM staging and closely related to survival rate (Mihm et al., 1988). The five-year survival rate is 95%100% if CMM thickness is less than 1 mm, while the survival rate is reduced to 50% if the tumor thickness is greater than 4 mm (Figure 1). Current surgical treatment for primary CMM has often been an excision with a margin determined by CMM thickness (Table 1). Since the risk of local recurrence is dependent on CMM thickness, a narrow margin of 5 mm is recommended for in situ CMMs, 1 cm for tumors thinner than 1 mm, 1-2 cm for tumors between 1.01 and 2 mm, and 2 cm for tumors thicker than 2.01 mm (Sladden et al., 2009). Because sentinel lymph node highly correlates with the metastatic status of CMM, a sentinel lymph node dissection (SLND) procedure is also performed on patients with intermediate thickness (1-4 mm) lesions (Balch & Ross,

1999; Kanzler & Mraz-Gernhard, 2001). Therefore, detection of CMM thickness is clinically significant and essential for five-year survival rate and surgical margin determination. Fig. 1. CMM thickness and survival rate. Image adopted from Rubin., 2010. The current gold standard for CMM diagnosis and tumor thickness measurement is achieved by taking a small biopsy for standard histology (Gambichler et al., 2006). Biopsy and histology allows the visualization of structures in a vertical section of the skin (i.e., from www.intechopen.com the epidermis through to the reticular dermis or even subcutaneous tissue). However, this procedure is invasive, time consuming and the sensitivity/specificity of CMM detection is highly dependent on location of biopsy. A real clinical need is recognized for non-invasive imaging techniques for in vivo evaluation of CMM. Currently, magnetic resonance imaging (MRI), computed tomography (CT), positron emission tomography (PET), high-frequency ultrasound (HFUS), optical coherence tomography (OCT), photoacoustic imaging (PAI), pulsed photothermal radiometry (PPTR), scanning confocal microscopy (SCM), multiphoton luminescence microscopy (MPLM), second harmonic generation (SHG), dermoscopy, multispectral imaging (MSI), diffuse reflectance spectroscopy (DRS) and Raman spectroscopy (RS) or some combination of these are being investigated for noninvasive diagnosis of CMM. Although MRI, CT and PET have the ability to identify nodal and distant metastasis, their routine use for localized CMM investigation is not indicated due to insufficient spatial resolution (King, 2004). Dermoscopy, MSI, DRS and RS are able to identify intrinsic differences between CMM, dysplastic nevi and normal skin but without providing the depth profile of CMM. SCM, MPLM and SHG provide the highest spatial resolution among all these imaging techniques and can also identify the morphological differences in CMM compared with dysplastic nevi. Moreover, SCM, MPLM and SHG have been successfully used to preoperatively delineate CMM lateral margins. However, the penetration depth of SCM, MPLM and SHG limits their use to detect depth margins of intermediate (1-4 mm) and more advanced (>4 mm) CMMs. HFUS, OCT, PAI and PPTR have spatial resolutions and penetration depths between MRI/CT/PET and SCM/MPLM/SHG, which underlines their potential applicability to not only diagnose CMM but also to detect lateral and depth margins. This chapter examines the evidence for use of non-invasive imaging techniques, in particular, MRI, CT, PET, HFUS, OCT, PAI , P P T R , S C M , M P L M a n d S H G , i n C M M diagnosis as well as tumor lateral and depth margin detection for preoperative CMM staging and surgical margin definition. Comparison between these imaging techniques in terms of spatial resolution, penetration depth, sensitivity/specificity, correlation with histology and temporal monitoring (possibility to monitor CMM changes at multiple time points) are described and, recommendations for future studies are indicated.

CMM imaging techniques 2.1 Magnetic Resonance Imaging (MRI)
Atomic nuclei in a magnetic field oscillate in the direction of the field at a specific frequency directly related to the field strength and the magnetic properties of the nuclei. If a pulse of current of the same frequency is applied to the coil surrounding the nuclei, an oscillating magnetic field is produced that creates a radio frequency within the coil. Magnetic field energy is absorbed by the nuclei and re-emitted as a radio-frequency signal immediately after the applied pulse. The re-emitted radio frequency energy is measured by the surface coil and reconstructed to form an MRI image (Baddeley, 1984). MRI has been widely employed in clinical oncology and introduced to examine cutaneous melanocytic and other types of skin lesions since 1989 (Zemtsov et al., 1989) due to good contrast between tumor regions and soft tissues (Totty et al., 1986;Weeks et al., 1985). The application of MRI to dermatology has become practical with the development and application of specialized surface coils that allow higher resolution imaging than standard MRI coils (Bittoun et al., 1990;Hyde et al., 1987;Marghoob et al., 2003;Querleux et al., 1988Querleux et al., , 1995Rajeswari et al., www.intechopen.com 2003; Richard et al., 1991;Zemtsov et al., 1989). Although imaging thin cutaneous tumors (i.e., <1 mm) is not possible due to insufficient resolution, MRI was utilized to evaluate advanced skin tumors, particularly to determine the depth of malignant tumors and the degree of invasion. el Gammal et al combined a strong homogeneous magnetic field of 9.4 T with gradient fields of 11.7 G/cm and an imaging unit to obtain a voxel resolution of 40×40×300 μm 3 , allowing differentiation between normal skin and skin tumors including CMMs ( Figure 2) (el Gammal et al., 1996). Using this approach, a primary nodular melanoma with Breslow thickness of 1.65 mm was imaged. The tumor is visible in the image (Figure 2d) in the upper left part of the histological section. Skin layer, sub-layers and tumor were visualized by MRI (Figure 2c) with different signal contrast. An excellent correlation between MRI image and corresponding histological features was achieved. Ono et al applied MRI to the diagnosis of malignant skin tumors and reported selected cases (Ono & Kaneko, 1995). A more advanced CMM with lateral dimensions of 36×26 mm 2 was imaged ( Figure 3). The MRI image (Figure 3a) was found to reflect precisely the actual morphology and tumor depth ( Figure 3b). The reconstructed 3-D MRI image ( Figure 3c) yielded accurate information regarding the relationship between tumor and its surrounding www.intechopen.com tissue and also provided a 3-D view of the state of infiltration, useful for deciding on a resection area prior to surgery. MRI has also been used to image nodal and distant metastasis in a murine model (Foster et al., 2008) as well as different organs in humans (King, 2004). However, the availability and cost of MRI is a limiting factor. Therefore, although MRI is useful for advanced CMM thickness measurement and individual metastasis characterization, at present this approach is not a first-line investigation tool.

Computed Tomography (CT) and Positron Emission Tomography (PET)
CT produces images with contrast given by the attenuation of X-ray photons by differing body tissues, in a manner similar to conventional radiography (Baddeley, 1984). Because CT has comparable or lower resolution than MRI (Link et al., 2003), it is not indicated to diagnose patients with primary CMMs but rather metastasis. The identification of CMM metastasis is dependent on the morphologic alteration of tissue malignancy and, therefore, CT is insensitive at detecting small tumor masses. Chest CT is sensitive to detect nodules (<1 cm), calcification within a nodule and to distinguish solitary from multiple pulmonary lesions (Armstrong et al., 2001;Halton, 1992). Although these findings are often non-specific for malignancy, they can provide a baseline scan allowing serial assessment of changes.
Heaston et al showed that chest CT detected CMM metastasis in 19% of patients with normal chest X-ray (Heaston et al., 1983). PET, a whole-body imaging technique, is widely used in the diagnosis of metastatic cancer. In a PET imaging procedure, an isotopic tracer is injected and used to label cancer cells.

www.intechopen.com
Fluoro-deoxy-glucose (FDG) is one of the most widely applied PET tracers used to survey cell metabolism (Gritters et al., 1993;Strobel et al., 2007;Wagner et al., 1999). Because metabolic characterization of tumor cells usually exceeds physiological metabolic activity, excessive FDG uptake has consequently been demonstrated in most cancers in vivo, making whole-body FDG-PET a sensitive indicator of metastatic CMM compared with conventional diagnostic imaging modalities (Blessing et al., 1995;Boni et al., 1995;Damian et al., 1996;Gritters et al., 1993;Macfarlane et al., 1998;Rinne et al., 1998;Steinert et al., 1995;Wagner et al., 1997). The improved sensitivity and potential cost-effectiveness of FDG-PET are rational arguments for PET staging of patients with recurrent CMM (Yao et al., 1994). The sensitivity of PET depends on the location and size of the tumor. A resolution of 4-6 mm is usual, which suggests that PET may lack the sensitivity to detect small nodular CMM metastasis that are usually 1-2 mm in size . As PET alone does not provide sufficient resolution to detect small CMM metastasis, it is usually used in combination with CT (Akcali et al., 2007;Essner et al., 2006), allowing mapping of PET images onto CT images acquired simultaneously ( Figure 4). Reinhardt et al studied 251 patients with PET/CT and showed a sensitivity of 98.7% compared to 88.8% for PET alone and 69.7% for CT alone (Reinhardt et al., 2006). Moreover, Iagaru et al recently reported a study involving 163 patients and showed a sensitivity of 89%, and recommended the use of PET/CT in the evaluation of high-risk CMM metastasis (Iagaru et al., 2006). Fig. 4. 18 F-fluorodeoxyglucose PET/CT for restaging of a 56-year-old patient with a superficial spreading melanoma of 4.3 mm depth at the right shoulder after resection of a lymph-node metastasis at the right neck. (a) CT, (b) PET and (c) fused PET/CT images showed two lymph-node metastasis infraclavicular, and PET showed an additional metastasis that was found to be localized in the right humerus head after image fusion. Image adopted from Reinhardt et al., 2006.

High-frequency Ultrasound (HFUS)
Ultrasound (US) uses a transducer to transmit sound pulses and to receive backscattered echo signals. Because the interfaces between tissues have different acoustic impedances and, therefore, different reflectivities, the received US signal contains boundary information of tissue with different elastic properties (T. Wang et al., 2011). US frequencies of 7.5-15 MHz are routinely used to visualize subcutaneous structures deeper than 1.5 cm including muscles, tendons, vessels and internal organs to identify pathologies or lesions (Aaslid et al., 2010;Shia et al., 2007;Tuzcu et al., 2010). The traditional US equipment used in the clinical environment does not allow precise measurement of tissues due to insufficient spatial resolution (Ulrich et al., 1999). High-frequency ultrasound (HFUS), however, is capable of visualization of dermis (20 MHz) and even epidermis (50-100 MHz). The resolution of the image and depth of tissue penetration is largely dependent on the frequency of the US transducer. High-frequency scanners offer finer resolution (e.g., axial resolution of 50 μm and lateral resolution of 300 μm for a 20 MHz transducer) but poorer tissue penetration. HFUS at 20 MHz have been used with success to distinguish between benign skin lesions and CMM. Harland et al showed 100% sensitivity in distinguishing between 29 basal cell carcinomas and 25 CMMs (Harland et al., 2000). Bessoud et al in a study involving 111 patients, identified 65 of 70 CMMs (81%). There was a 100% sensitivity and specificity for melanoma and 32% specificity for non-melanoma lesions (Bessoud et al., 2003). HFUS at 20 MHz has also been shown to allow preoperative assessment of CMM thickness that correlated well with histological measurement Hoffmann et al., 1992;Lassau et al., 1999;Serrone et al., 2002;Tacke et al., 1995). A prospective study and systematic review of literature from 1987 to 2007 on CMM thickness measurement using 20 MHz HFUS reported that measurement of CMM thickness was possible except for thin CMMs (<0.4 mm) in areas with marked photoaging, and in the case of very thick CMMs exceeding the explored depth (7.6 mm) ( Figure 5a) (Machet et al., 2009). This study also demonstrated a linear correlation between ultrasound and histology in CMM thickness measurements ( Figure 5b). Another study using 75 MHz HFUS, involving 112 patients with suspicious CMMs, showed that 45 of 52 CMMs had clear hypoechogenic boundaries with tumor thicknesses ranging from 0 to 2.8 mm and correlation between HFUS measurement and histology was high (r=0.908) (Guitera et al., 2008).

Optical Coherence Tomography (OCT)
OCT is an emerging diagnostic optical imaging technique that provides in vivo structure and function of tissues by measuring backscattered or backreflected light. OCT is based on the principle of Michelson interferometry. The light sources used for OCT imaging of skin are broad-band superluminescent diodes or tunable laser sources operating at a wavelength of about 1300 nm (Huang et al., 1991). The broad-band source leads to a small coherence length and achieves an axial and lateral resolution of approximately 15 μm and a penetration depth of www.intechopen.com 500 to 1000 μm (Olmedo et al., 2006;Welzel et al., 2003). OCT is a well established tool in ophthalmology (Welzel, 2003) and currently being advanced in dermatology (Fujimoto et al., 1995;Gambichler et al., 2005;Welzel, 2001), particularly for the diagnosis of CMM (Marghoob et al., 2003). A study of CMM characterization by OCT examined a panel of CMMs and benign nevi and demonstrated that CMMs showed increased architectural disarray, less defined dermal-epidermal borders, and vertically oriented icicle-shaped structures not seen in nevi (Figure 6b,c) (Gambichler et al., 2007). A recent review of OCT investigation in dermatology also showed that the intact border between epidermis and dermis disappears in infiltrative growing CMM compared with healthy skin (Figure 6a) (Smith & MacNeil, 2011). The utility of OCT for early-stage CMM thickness measurement (Figure 7) has not been fully established because correlation studies for CMM thickness determination by OCT and histology have not been reported. Furthermore, an important limitation of OCT is penetration depth. The maximum penetration depth of OCT is currently between 1-2 mm, dependent on the tissue type. Although the use of longer light wavelengths (e.g., 1750 nm) may improve penetration depth, no data is available to suggest that penetration depth greater than 4 mm is possible in the near future (Brezinski & Fujimoto, 1999). Therefore, OCT is currently not an established candidate for CMM thickness measurement.

Photoacoustic Imaging (PAI)
PAI is an emerging hybrid technique that detects absorbed photons ultrasonically through the photoacoustic effect (Sun & Diebold, 1992). When a short-pulsed radiant source (e.g., laser) irradiates biological tissues, wideband ultrasonic waves (referred to as photoacoustic waves) are induced as a result of transient thermoelastic expansion. Magnitude of the photoacoustic waves is proportional to the local optical energy deposition and, hence, the waves divulge physiologically specific optical absorption contrast. As energy deposition is related to optical absorption coefficient of chromophores, concentration of multiple chromophores can be quantified by varying the laser excitation wavelength. Tissues such as blood vessels and CMM, can be imaged by PAI with the spatial resolution of ultrasound, which is not limited by the strong light scattering in biological tissues (X. Wang et al., 2003).

www.intechopen.com
Images of microvasculature as deep as 3 mm were demonstrated with high-resolution PAI (Maslov et al., 2005). Recently, Oh et al used high-resolution and high-contrast PAI in vivo with a near-infrared (NIR) (764 nm) and a visible (584 nm) pulsed laser source, respectively, to image the 3-D CMM distribution inside nude mouse skin and the vascular system surrounding the CMM including tumor-feeding vessels (Oh et al., 2006). Maximum CMM thickness (0.5 mm) was measured with a lateral resolution of 45 μm and an axial resolution of 15 μm (Figure 8). Detection of melanoma cells in circulation was also reported (Holan & Viator, 2008;Weight et al., 2006;Zharov et al., 2006). More recently, in a pilot study Song et al proposed that non-invasive in vivo spectroscopic PAI can map sentinel lymph node using gold nanorods as lymph node tracers in a rodent model (Song et al., 2009). images projected along the z axis of a CMM region, where an MAP image is formed by projecting the maximum photoacoustic amplitudes along a direction to its orthogonal plane. Here, blood vessels are pseudocolored red in the 584 nm image and the CMM is pseudo-colored brown in the 764 nm image. As many as six orders of vessel branching can be observed in the image as indicated by numbers 1-6. (i) 3-D rendering of the CMM from the data acquired at 764 nm. Two MAP images at this wavelength projected along the x and y axes are shown on the two side walls, respectively. Image adopted from Zhang et al., 2006.

Pulsed Photothermal Radiometry (PPTR)
PPTR is a non-invasive technique that utilizes an infrared detector to measure radiometric temperature changes induced in a test material exposed to pulsed radiation. Heat generated as a result of light absorption by subsurface chromophores in the test material diffuses to the surface and results in increased infrared emission levels. By collecting emitted radiation onto an infrared detector, a PPTR signal is obtained that represents the time evolution of temperature near the test-material surface. Useful information regarding the test material may be deduced from analysis of the PPTR signal. PPTR has been applied to depth profiling of strongly absorbing tissues and tissue phantoms (Milanič et al., 2007;Milner et al., 1996), including blood vessels in port wine stain (PWS) birthmarks in human skin (Li et al., 2004). Because different chromophore thicknesses can provide different laser induced initial temperature profiles and eventually produce different radiometric temperatures (T. , the authors proposed that relationship between CMM thickness and detected radiometric temperature increase can be determined using PPTR in tissue phantoms mimicking CMM thicknesses from 120 µm to 2.8 mm (Figure 9) with a penetration depth of 1.7 mm and axial resolution of 75 µm (T. Wang et al., 2011). However, further studies are needed to investigate the capability of PPTR in CMM thickness measurement in skin in vivo.

Scanning Confocal Microscopy (SCM)
SCM is a non-invasive imaging technique that permits in vivo examination of the epidermis and papillary dermis. The basic premise of SCM is the selective collection of light from a specific plane in tissue through a pinhole-sized aperture which allows for light collection from the single in-focus plane and the rejection of light from all out-of-focus planes (Nehal et al., 2008). SCM has been recently employed in CMM diagnosis (Gerger et al., 2005;Marghoob & Halpern, 2005), preoperative and intraoperative margin assessment (Busam et al., 2001), and followup for response to medical treatment (Ahmed & Berth-Jones, 2000;Cornejo et al., 2000;Langley et al., 2006;Tannous et al., 2000Tannous et al., , 2002. Commercial SCM instruments have been developed that image with lateral resolution of 0.5 to 1.0 μm and an optical sectioning thickness of 1.0 to 5.0 μm, to a depth of 200 to 300 μm in human skin (depth of papillary dermis). The spatial resolution of SCM is determined by the pinhole size while imaging depth is limited by the laser wavelength (with a 488 nm laser imaging 50-100 μm into skin (Gareau et al., 2007) and longer wavelengths lasers able to image at depths of up to 300 μm (Gonzalez & Gilaberte-Calzada, 2008;Marghoob et al., 2003;Nehal et al., 2008), providing images of the basement membrane down into the papillary dermis. SCM with an 830 nm light source is ideal for detecting CMM because melanin serves as an endogenous contrast agent. Melanin presence in melanocytic nevi and CMM provides strong contrast, thereby permitting the clear visualization of the architecture and outlines of cells Rajadhyaksha et al., 1995). SCM can be used in either fluorescence or reflectance modes (Meyer et al., 2006). While dermatological research tends to use fluorescence, clinical practice uses reflectance microscopy as this does not require fluorescent labeling of cells and tissues and, therefore, is more suitable for in vivo imaging. SCM criteria have been established to distinguish between CMMs and benign nevi (Gerger et al., 2005;Pellacani et al., 2007). Pellacani et al proposed a diagnostic algorithm that uses 2 major (i.e., nonedged dermal papillae and cytologic atypia at the basal layer) and 4 minor criteria (i.e., roundish pagetoid cells, widespread pagetoid infiltration in the epidermis, nucleated cells within dermal papillae, and cerebriform cell clusters in the dermis) ( Figure  10) (Pellacani et al., 2005). The presence of at least 2 features, 1 major and 1 minor criterion, are required for a positive CMM diagnosis.  SCM is capable to identify the lateral margins of CMM when determining the precise margins by clinical Wood's lamp or dermoscopic examination is virtually impossible. Chen et al reported a case of a patient with a recurrent CMM on the scalp that developed in a background of photodamage with diffuse melanocytic atypia and lentigines (Chen et al., 2005). SCM was able to distinguish the adjacent normal skin from CMM and the lateral tumor margin was preoperatively determined by SCM and the tumor was excised accordingly (Figure 11). Another study by Curiel-Lewandrowski also demonstrated the feasibility of using SCM in preoperative and intraoperative surgical margin assessment of indistinct CMM lesions (Curiel-Lewandrowski et al., 2004). However, SCM detection of tumor depth margins is still difficult due to insufficient penetration depth. The high cost and sophisticated design of current SCM devices are considered to be major barriers. Efforts, such as miniaturizing the device and lowering the cost of production can facilitate wider adoption.

Multi-photon Luminescence Microscopy (MPLM) and Second Harmonic Generation (SHG)
MPLM is a rapidly developing imaging technique in the field of optical sectioning, which has been applied to tissue imaging with intrinsic fluorescence (Zipfel et al., 2003a(Zipfel et al., , 2003b. Endogenous fluorophores such as melanin, elastin and collagen are sources of tissue fluorescence. MPLM of animal and human skin has been reported by So et al, Masters et al, Hendriks et al, König et al and Peuckert et al (Hendriks & Lucassen, 1999;König, 2000aKönig, , 2000bKönig, , 2002Masters et al., 1998;Peuckert et al., 2000;So & Kim, 1998). Teuchner et al reported MPLM detection of melanin fluorescence (Teuchner et al., 1999). MPLM excites fluorophores by a nonlinear multiphoton (e.g., two-photon) process, as opposed to the single photon excitation used in conventional microscopy. Two-photon excitation occurs when two photons of approximately half the one photon energy are absorbed nearly simultaneously by the fluorescent molecule. MPLM allows non-invasive tissue screening with subcellular spatial resolution. Wang et al demonstrated that lateral and axial resolution of MPLM can reach about 0.3 and 1 μm, respectively (H. Wang et al., 2005). Masters et al detected the autofluorescence of human skin in depths down to 200 μm (Masters et al., 1997). Figure 12 illustrates the effectiveness of MPLM in obtaining subcellular resolution images of CMM and normal skin (Dimitrow et al., 2009). Similar to SCM, MPLM has the potential to precisely measure the lateral margins of CMM, however, limited penetration depth prevents use for tumor depth margin delineation. SHG (also called frequency doubling) is also a nonlinear optical process, in which photons interacting with a nonlinear material are effectively combined to form new photons with twice the energy and, therefore, twice the frequency and half the wavelength of the incident photons (Fine & Hansen, 1971;Roth & Freund, 1979;Theodossiou et al., 2006). Because SHG excitation wavelength is off the resonance wavelength of chromophores in tissue, less energy is absorbed, and hence, negligible thermal or photodamage is observed (Lohela & Werb, 2010). Recently, SHG imaging has been employed to detect noncentrosymmetric crystalline structures (e.g., collagen) in tissues (Campagnola et al., 2001;Cox et al., 2003;Lim et al., 2010;). Thrasivoulou et al demonstrated that SHG imaging showed detailed collagen distribution in healthy skin, with total absence of SHG signal (fibrillar collagen) within the melanoma-invaded tissue ( Figure 13) (Thrasivoulou et al., 2011). The presence or absence of SHG signal changed dramatically at the borders of CMM, allowing accurate demarcation of CMM margins that strongly correlated with H&E and Melan-A defined margins (p<0.002). www.intechopen.com

Dermoscopy
Dermoscopy (also called dermatoscopy, skin surface microscopy, epiluminescence microscopy) is a non-invasive technique for the early recognition of CMM. Dermoscopy was first introduced to evaluate pigmented lesions in 1971 (MacKie, 1971), studied rigorously in Europe and Australia during the 1980s and 1990s (Argenziano et al., 1998), and adopted slowly in the United States in recent years (Tripp et al., 2002). To date, dermoscopic and histologic correlations and algorithms for dermoscopic diagnosis of CMM have been established (Argenziano et al., 2003;Henning et al., 2007). Two types of dermoscopes are currently available: (1) nonpolarized dermoscope (NPD) and (2) polarized dermoscope (PD). NPD uses liquid (e.g., oil, water, or alcohol gel) to cover the lesion, which decreases light reflection, refraction, and diffraction, makes the epidermis essentially translucent and allows for visualization of subsurface anatomic structures of the epidermis and papillary dermis ( Figure 14). PD uses a polarizer that preferentially captures backscattered light from below the surface of the skin. The liquid or direct contact with the skin is not needed. However, some subset of CMMs (e.g., amelanotic/hypomelanotic CMMs that lack pigmentations and dermoscopic structures) are difficult to diagnose with dermoscopy and, thus, impossible to determine lateral margins by dermoscopy. Although a dermoscopic image can show subsurface structures of CMM, this approach is a 2-D imaging technique that cannot obtain a depth profile of the lesion. Therefore, depth margins of melanoma are not provided by dermoscopy.

Multispectral Imaging (MSI), Diffuse Reflectance Spectroscopy (DRS) and Raman Spectroscopy (RS)
MSI has been widely used in the fields of astronomy and remote sensing (Colarusso et al., 1998;Curran, 1994) and, recently, is being applied to the field of biology and medicine (Johnson et al., 2007;Mansfield et al., 2005;Weber et al., 2011). An essential part of an MSI microscope is a spectral dispersion element that separates incident light into its spectral components (i.e., 400-1000 nm) (Bearman & Levenson, 2003). MSI of skin acquires spectrally resolved information at each pixel of a multispectral image, providing information on the distributions of collagen, melanin content and blood vessels within skin lesions (Marchesini, 1991(Marchesini, , 1992. Carrara et al developed an algorithm for the automatic segmentation of multispectral images of 1856 cutaneous pigmented lesions including 264 CMMs, which successfully detected lateral margins of the lesions with a contour accuracy of 97.1% (Carrara et al., 2005). Marchesini et al reported that MSI and an artificial neural network could be used in the preoperative evaluation of CMM thickness with sensitivity (i.e., CMM≥0.75 mm thick correctly classified) and specificity (i.e., CMM<0.75 mm thick correctly classified) ranging from 76 to 90% and from 91 to 74%, respectively (Marchesini et al., 2007). DRS (also called elastic scattering spectroscopy) measures the spectral modification of remitted light (i.e., light that has propagated some distance into the skin, been scattered, and recollected at skin surface). DRS was first introduced as a single-point measurement technique by Marchesini et al in 1992 to study CMM and nevus with a 5 mm probe (Marchesini et al., 1992). Wallace et al later reported that DRS in the wavelength range 320-1100 nm had the potential for improving the differential diagnosis of CMM from benign pigmented skin lesions (Wallace, 2000a(Wallace, , 2000b. However, further prospective study of DRS is needed to investigate its capability to detect tumor margins. RS is a non-invasive imaging technique that has been widely used for the past 70 years for nondestructive chemical analysis. RS is based on the principle of Raman scattering, the inelastic scattering of electromagnetic radiation. The Raman scattering effect is caused by molecular vibrations in the irradiated sample and thus gives information about the structure of the molecules. Earlier studies have shown that samples of various benign skin lesions and nonmelanoma skin cancer have characteristic Raman spectra (Gniadecka et al., 1997a(Gniadecka et al., , 1997b(Gniadecka et al., , 1998. Recent studies on basal cell carcinoma by RS demonstrated its feasibility to distinguish malignant tissue from healthy surrounding tissue (Nijssen et al., 2002). More recently, Gniadecka et al developed a neural network system for the automated classification of Raman spectra, allowing CMM differentiation from other clinically similar skin tumors (Gniadecka et al., 2004). Therefore, RS is potentially capable of demarcating lateral margins of CMM. In contrast to other imaging techniques (e.g., dermoscopy), MSI, DRS and RS are more objective and not observer dependent. However, similar to demoscopy, MSI, DRS and RS are also 2-D imaging techniques that cannot image the depth profile of CMM directly. Therefore, detection of depth margins of CMM by MSI, DRS or RS is hardly possible.

Conclusion
Early detection and surgical resection can reduce CMM mortality. To this end, several noninvasive imaging techniques have been developed to realize CMM screening, early diagnosis, preoperative tumor staging, surgical margin definition and in vivo tumor monitoring over multiple time points. Currently available instruments include MRI, CT, PET, HFUS, OCT, PAI, PPTR, SCM, MPLM, SHG, dermoscopy, MSI, DRS and RS. Dermoscopy, MSI, DRS and RS are useful in the evaluation of superficial CMMs, but they are not depth resolved. PET/CT do not have the resolution to detect early stage CMMs (e.g., <1 mm). They can, however, be used to detect tumor metastasis. MRI has comparable or higher resolution than CT and, hence, is suited to imaging advanced tumor invasion (e.g., >1 mm). SCM, MPLM and SHG provide physicians with an unprecedented capability to visualize a CMM lesion at a detail comparable to histology, but their imaging depths are limited to several hundred micrometers from skin surface, preventing them from detection of CMM depth margins beyond penetration depth. HFUS, OCT, PAI and PPTR have spatial resolutions and penetration depths between MRI/CT/PET and SCM/MPLM/SHG, therefore, they are potentially capable of detecting both lateral and depth margins of CMMs. A comparison of spatial resolution, penetration depth, sensitivity/specificity, correlation with histology and temporal monitoring (possibility to monitor CMM changes at multiple time points) of the imaging techniques discussed are summarized in Table 2 below. From a clinical perspective, sensitivity and specificity studies of CMM diagnosis by MRI/CT/PET, HFUS, SCM/MPLM and dermoscopy/MSI/DRS/RS having been performed. Although sensitivity and specificity of OCT, PAI, PPTR and SHG on human CMM diagnosis are not yet established, they are promising imaging techniques that need further investigation on not only CMM diagnosis but also tumor lateral and depth margin delineation. To date, surgical resection is the major treatment for primary CMMs, which significantly increases the five-year survival rate. Surgical margins are currently determined by CMM thickness measured from biopsy with histology -the gold standard. The imaging techniques discussed, however, have the potential to non-invasively measure CMM thickness and detect tumor margins, and even guide CMM surgical resection in real time, which will reduce unnecessary biopsies and, eventually replace this invasive approach. Because each of these imaging techniques has advantages and shortcomings, the best performing imaging tool for CMM evaluation is a combination of different imaging techniques -although at increased cost. A multimodal imaging tool will significantly improve the accuracy of CMM identification as well as the precision of surgical margin definition, which would achieve the five-year survival rate currently determined by biopsybased surgical resection.  (Gammal et al., 1996;Schick, 2005). $ Lateral resolution varies from 0.1 (micro-CT) to 0.29 mm (conventional CT), axial resolution varies from 0.1 (micro-CT) to 1 mm (conventional CT) (Badea et al., 2004;Schroeder et al., 2001). * Both lateral and axial resolution vary from nearly 1 mm FWHM (2 mm FWTM) for a 10-20 cm diameter system typical for animal studies with 18 F to roughly 4 mm FWHM (7 mm FWTM) for an 80 cm diameter system typical for human imaging using 15 O (Levin & Hoffman, 1999). § Lateral resolution varies from 20 (100 MHz) to 300 μm (20 MHz), axial resolution varies from 20 (100 MHz) to 50 μm (20 MHz), penetration depth varies from 4 (100 MHz) to 7.6 mm (20 MHz) (Harland et al., 1993;Passmann & Ermert, 1999;Pavlin et al., 1990). ¶ Lateral resolution can be 10 μm or better, axial resolution varies from 10 to 20 μm (Ding et al., 2002;Schenk & Brezinski, 2002). δ Axial resolution varies from 1 to 1.9 μm (Campagnola et al., 2002;Moreaux et al., 2000). Lateral resolution is 20 μm with no axial resolution (Kopf et al., 1997). Lateral resolution is 30 μm with no axial resolution (Marchesini et al., 2007). κ Lateral resolution is limited by the fiber size, no axial resolution is available (Häggblad et al., 2010). μ Lateral resolution is limited by the laser spot size, no axial resolution is available (Gniadecka et al., 2004). 420 stage III/IV melanoma lesions from 64 patients were examined by MRI, CT and PET respectively (Pfannenberg et al., 2007).
Correlation of collagen fiber density was examined from mid-point of CMM lesion to deep area of non-lesion skin (Thrasivoulou et al., 2011). 128 pigmented skin lesions including 33 CMMs were examined by dermoscopy and ABCD diagnosis rule (Argenziano et al., 2003).
1939 pigmented skin lesions including 250 CMMs were examined by MSI, the correlation coefficient value between tumor thickness and area is not so great to fully assess that lesion dimension increases with thickness (Marchesini et al., 2007).
15 CMMs and 32 compound naevi were examined by DRS (Wallace, 2000a). 134 pigmented skin lesions including 22 CMMs were examined by RS and neural network analysis (Gniadecka et al., 2004). Table 2. Comparison of non-invasive imaging techniques for CMM diagnosis and tumor margin detection in terms of spatial resolution, penetration depth, sensitivity/specificity, correlation with histology and temporal monitoring (possibility to monitor CMM changes at multiple time points)

Acknowledgement
We thank Jingjing Sun for her technical assistance with the preparation of references for this book chapter. The book Skin Cancer Overview is divided into three sections to cover the most essential topics in skin cancer research: Etiology, Diagnosis and Treatment, and Prevention. Due to the complexity of skin cancer, this book attempts to not only provide the basic knowledge, but also present the novel trends of skin cancer research. All chapters were written by experts from around the world. It will be a good handbook for researchers with interests in skin cancer.