Skin cancer detection using non-invasive techniques

Skin cancer is the most common form of cancer and is globally rising. Historically, the diagnosis of skin cancers has depended on various conventional techniques which are of an invasive manner. A variety of commercial diagnostic tools and auxiliary techniques are available to detect skin cancer. This article explains in detail the principles and approaches involved for non-invasive skin cancer diagnostic methods such as photography, dermoscopy, sonography, confocal microscopy, Raman spectroscopy, fluorescence spectroscopy, terahertz spectroscopy, optical coherence tomography, the multispectral imaging technique, thermography, electrical bio-impedance, tape stripping and computer-aided analysis. The characteristics of an ideal screening test are outlined, and the authors pose several points for clinicians and scientists to consider in the evaluation of current and future studies of skin cancer detection and diagnosis. This comprehensive review critically analyses the literature associated with the field and summarises the recent updates along with their merits and demerits.


Introduction
Skin cancer is a common type of cancer that emerges from the skin. It is extended due to the development of abnormal growth of cells. These cells can invade other parts of the body. More than 90% of cases occur due to the exposure of UV radiation. The UV radiation wavelength ranges from 100 nm to 400 nm. Cancer occurs due to unrepaired DNA damage to skin cells which is caused mostly due to UV radiation. In melanoma, cancer cells arise from moles on the skin which causes inammation around the epidermal layer which in turn increases the temperature around it. 1 A survey in 2015 found about 17.5 million cancer cases worldwide and 8.7 million deaths occurred due to cancer. As part of the epidemiological transition, cancer incidence is expected to increase in the future, considering limited health care resources. Past statistics and trends are shown in Fig. 1. In 2017, out of 5.4 million cases of non-melanoma skin cancer, only 3.3 million people were treated in the US. Over 87 110 new cases of melanoma will be diagnosed in the US, but estimation found that 9730 people will die of melanoma due to lack of healthcare resources. About 2 to 4% cases of people are suffering from melanoma skin cancer in India. 1 In a report published ten years ago, skin cancer ranked 10th as the most common cancer in Malaysia and accounted for 2.6% of all cancer cases in the country. Melanoma is not very prevalent in Malaysia and occurred in only 5.4% of the patients who attended the dermatology clinic in Hospital Kuala Lumpur, but awareness among them is very less. 2 At present, between two and three million non-melanoma skin cancers and 132 000 melanoma skin cancers happen globally every year. According to the recent research on skin cancer trends in Asia, it was found that skin cancer rates among the fairer-skinned were approximately three times higher than those who generally have darker complexions. Prevalence rates of skin cancer in Singapore increased from 1968 to 2006, especially among older Chinese. 3

Classifications of skin cancer
Skin cancer is categorized into three types which are basal cell skin cancer, squamous skin cancer and malignant skin cancer. The rst two skin cancers are grouped together as nonmelanoma skin cancers. The stages of skin cancer is shown in Fig. 1. 4 The features and the treatment involved in each type of cancers are discussed below.

Basal cell skin cancer
Basal cancer cell (BCC) is the least aggressive of non-melanoma skin cancer derived from keratinocytes that resemble epidermal basal cells. 5 BCC appears like a esh coloured pearl likes a bump or a pinkish patch of skin. BCC is due to skin exposure especially found in the sun exposed areas such as the face, head, neck, arms, should be treated at the earlier stage else it will grow in the dense amount in the body. SCC is characterized by proliferation of invasive squamous cells which are able to invade other tissues. It can be treated by photodynamic therapy, tropical chemotherapy, and curettage and electrodessication. SCC is the second common cancer in the world. It is not dangerous than that of melanoma. 4,8 Malignant melanoma Melanoma arises from melanocytes, specialized pigmented cells that are found abundantly in basal layer of epidermis. 9 Malignant melanoma consist of four major histological subtypes frequently reported namely supercial spreading melanoma (70%), nodular melanoma (15%), lentigo maligna melanoma (13%), and acral lentiginous melanoma (2-3%). 10 Melanoma is usually observed as a dark spot on the epidermal (outer) layer of the skin. It is the fast-growing cancer cells that have the ability to spread over the surrounding tissue of the body. The malignant melanoma is caused due to UV radiation and damages the DNA indirectly. The free radical and the reactive oxygen species are the two leading causes of DNA damage. Research indicates that intentness of sunscreen components into the skin, conjugate with 60 min exposure to UV leads to the increase of free radical in the skin. 8 Globally in 2012 malignant melanoma occurred among 232 000 people, and 55 000 people have died. 11 Malignant melanoma is the deadliest cancer, but it can be cured at early stages of cancer. It is normally treated with chemotherapy, targeted therapy, and radiation therapy.
Detection of skin cancer in the early stage is useful in treating the condition very quickly. Detection systems pave the way for the diagnosis, but the result was not that precise comparing to biopsy test which is an invasive method. Along with the cost involved, the aer-effects of biopsy include severe infectious problem around the punctured area. If the condition found to be more severe in the biopsy test, then surgery is performed, this adds to the cost along with the complexities. Other methods like radiotherapy, chemotherapy and drug therapy can be given, but the loss of hair and sittings oen makes the patient to the worst condition. By taking into account all these points, many sensors are fabricated and designed to detect the skin cancer in a non-invasive manner, unlike biopsy. In this comprehensive review article, the recent non-invasive approaches involved in skin cancer detections are summarised.

Literature search performed
The literature search was performed using Google Scholar. The search was carried out for the keywords "Skin cancer, detection techniques, diagnosis, non-invasive" for the time span of 1990-2017. From the search, relevant articles were presented, and some articles were chosen based on the pertinence. Publication trends over the years are shown in Fig. 2. This review article will be promising for researchers who plan to explore further this eld.  Fig. 4. Along with that presented the currently practiced skin cancer diagnosis methods and techniques. 12 Principles and mechanisms involved Optical. When the light is passed into the tissues, the light gets scattered within the tissues of the skin. The change in the property of the reected light is used for diagnosis. 13 Advanced biochemical markers are also developed with respect to cancerous tissue and specic wavelengths are used to enhance the contrast of the tissue by absorbing light. 14,15 Photodynamic based. In order to detect the presence of tumor cells, a photosensitive marker is introduced into the particular area. Protoporphyrin IX is formed as a result of chemical reactions inside the body, which is absorbed by the healthy cells but gets accumulated in the tumour cells. When the resectoscope is introduced into the area, under blue illumination the accumulated pigment in the tumour cells emit red uorescence and normal cells will be in blue. To provide contrast for photodynamic diagnosis (PDD) yellow-light lter is specially designed which enhances the red uorescence. Skin lesions using photodynamic diagnosis are based on the uorescent properties of an exogenous and endogenous compound  Review in response to illumination. Stimulation of photosensitizers is made to produce the photodynamic effect when a certain wavelength of light is exposed to it. 12,16 Sonography. Sound waves are used in this approach. Aer transmitting sound pulses into the body, some of the sound waves get reected back to the probe, while some travel on farther until they arrive another boundary and get reected. The reected waves are sensed by the probe and relayed to the machine. The changes in the property of sound waves are used in this technique. Sonography is used to assess skin lesion depth and margins before doing the biopsy and also used to classify adjacent lymph nodes whether it has become cancerous or not. This is added to a patient's protocol to clarify lesion depth, identify margins for excision, and determine lymph node. It has the ability to differentiate benign from malignant skin cancer. It is used to identify metastatic lymph nodes visually, which avoids unnecessary removal of lymph nodes. The main advantage is that it gives the accurate results of measuring skin cancer lesion thickness. 17 Electrical bio-impedance. The electrical impedance of skin, when measured by topical application of electrodes, is dominated by the high resistance of the stratum corneum. Changes in stratum corneum hydration or structure can be followed by electrical measurements at low frequency. Stratum corneum must be bypassed to measure the living epidermis and dermis. Direct contacts of microneedles on stratum corneum penetrate and measure the electrical activity underlying the skin. 18 Thermal imaging. Thermography is based on the phenomenon that any object of a temperature above absolute zero emits electromagnetic radiation. There is a denite relationship between the surface of a body and the intensity and spectral composition of its emitted radiation. By determining its radiation intensity, the temperature of an object can thereby be determined in a non-contact way. The human skin is a thermal interface with the human environment. Temperature distribution on the skin will be emitted in the infrared band that is useful in imaging the thermal distribution. Infrared cameras are used in clinical medicine to map the thermal distribution in the human body. Stress is applied to the imaging region so that the liquid crystal plate directly over the body make useful in capturing the image. Thermal imaging can also be performed in a non-contact manner through the infrared camera of high sensitivity. Comparing to other techniques which have false negative reports at a high probability, thermography provides high accuracy. 19 Techniques and methods

Photography
Total Body Photography (TBP) is also known as mole mapping which is used to identify patients with numerous nevi and changes in the lesions during regular skin examinations. 20 TBP uses the digital camera for recording and storing the images as shown in Fig. 5. Using a high-resolution, professional medium- format camera, images are taken from the subjects by implementing the camera at different poses and position of the subject to get accurate results. 21 Once the image is taken, it is digitally processed using mirror body mapping soware. Sum of all the images is analyzed using this soware. The soware later generates an interactive, stand-alone program (Derma-Graphix) 22 containing the patient's body map. This stand-alone DermaGraphix body map, which is given to the patient using USB 2.0, tier 1 thumb drive. 23 This system can also be performed at home by oneself, but the cost of equipment is high. 24 Lynn T. Dengel, Gina R. Petroni and Craig L. Slingluff et al. had a signicant nding of photography method in screening of skin cancer, but it is not used widely used due to time and cost. 20 In this technique, captured the surface images of the skin to identify suspicious and pigmented lesions in high-risk patients primarily. Image processing algorithms are applied to the captured images for detecting any abnormalities in the skin. The segmentation technique which is used to detect the skin lesion is classied as region-based segmentation and neural edge detection. In region-based segmentation method, the optimal threshold is determined iteratively by an isodata algorithm. In neural network edge detection, an approximate closed elastic curve between the recognized neural network edge patterns is tted. 25 Computer-aided decision tools are essential in medical imaging for diagnosis and evaluation of various diseases. 26 The acquired image from the subject using high-resolution camera is pre-processed to reduce the artefacts. To overcome the artefacts like the presence of hair in the image, mean, median, Gaussian and anisotropic lters are employed. It is followed by the detection of the lesion by image segmentation technique that partition the image into disjoint areas that are homogeneous concerning a chosen property such as luminance, colour, and texture. Next process is feature extraction. 27 The extraction process takes by ABCD rule that checks the symmetrical, border, colour, and the diameter of the acquired image. Once the lesion is localized, different chromatic and morphological features can be quantied and used for classication. 28 In the classication part, bringing out the algorithm in the examination process combines visual processing with deep learning. 29 In deep learning, the computer is trained to gure out a problem rather than having the answers programmed into it.
The computer-aided technique is used to segment the melanocytes in the images taken from the skin. This segmentation is made mostly to reduce the local intensity variant. The mean-shi algorithm is adapted. 30 The digital images of melanoma skin lesions can provide information about the skin cancer at an early stage without performing any skin biopsies. 31 The mathematical morphology has been used for segmentation for proper identication of skin lesion area. Using TBP, different shape, texture and colour features are extracted and malignant melanomas have been classied sensitively. 32 Delaunay triangulation is a technique used to extract a binary mask of the lesion region, which doesn't require any training. This is highly accurate when dealing with benign lesions, while the segmentation accuracy signicantly decreases when melanoma images are processed. Geometrical and colour features were extracted from the binary masks generated by the algorithm for classication, which gives accurate results for melanoma detection. 33 Computer-aided diagnosis extended its framework for the detection of skin cancer in recent years. The detection of melanoma skin cancer in the early stage will be very useful to cure it and safeguard the life of the affected individuals. Authors Abdul Jaleel, Sibi Salim, R. B. Aswin et al. found that based on imaging techniques and articial intelligence the result of computer-aided detection of skin cancer is based. 34 Computer vision plays a vital role in medical image diagnosis. Reported the computer-aided diagnosis for the pigmentations of skin lesions to determine the presence of skin cancer by performing the boundary detection. 35 Skin lesions can also be analysed by qualication of degree of symmetry. The other paper explains that melanoma can be diagnosed by wavelet and texture analysis. 36 The texture features were derived from the wavelet decomposition and the border features were collected from the lesions borders and the selections of such features were collected by using the gain-ratio method. That was computationally efficient for the melanoma diagnosis. In recent years, computer-aided diagnosis extended its framework in mobile technologies and in cloud platforms. By this method, the system can able to classify the lesions by identifying the moles in skin images and classify them accordingly as melanoma, benign and nevus lesions. 37 Computer diagnosis can also be used as a preventive tool in the detection of skin cancer via mobile phone applications. By using this technique, the melanocytes can be separated from the histopathological images. 38 Smartphone plays an active role in medical diagnostic of skin cancer by considering the sensitivity, specicity etc. The non-clinician users can easily differentiate the malignant and benign tumours. 39 Through the computer-aided method, the patients can identify the skin cancer without going to the hospital. 40 Computer-aided diagnosis can be carried out in the eld of image processing, pre-processing, feature extraction, segmentation and classication of demographics images. 34 Computer-aided melanoma diagnosis is used to provide the support to the physicians in enhancing the sensitivity and specicity in the diagnosis of melanoma condition. To capture the images of skin lesions, optoelectronic probe were used. 41 One of the articles reveals that computer-aided diagnosis can be used to detect the early stage of melanoma by applying the image processing technique. Images can be analysed based on ABCD technique which involves asymmetry, border, colour, diameter etc. From the skin lesion images, the above parameters were collected by considering the texture, size and shape of the image. 26 These features were extracted from the images and it was further processed for image segmentation. The extracted features were used to distinguish the normal skin and the melanoma cancer lesions as shown in Fig. 5b. The other methods deal with the image processing in which the images of the affected areas are captured by using the camera of the mobile phone by moving the diffraction grating. Thus the skin cancer can be diagnosed. 42 Recent updates on photography techniques for skin cancer diagnosis are summarised in Table 1.

Dermoscopy
Dermoscopy is also known as dermatoscopy or epiluminescence microscopy that is useful in examining the skin lesions. 50 Conventional instrumental set up for dermatoscopy includes a magnier (Â10), a non-polarised light source, a transparent plate and a liquid medium between the device and the skin so that skin reections can be avoided. Modern dermoscope as shown in Fig. 6, do not use this liquid medium; instead, it uses polarised light that eliminates the skin reection. 51 From this method, the non-cancerous and cancerous condition can be known. Basically, this dermoscope operates under three modes, rst is, non-polarised in a contact manner, second is polarised light in a contact manner then the third mode is polarised light in a non-contact manner. The main advantage of using polarized light is, it provides better visualization in deeper skin structures whereas for non-polarized the information will be only on the supercial layer of skin. 52 In this system, the accuracy level is increased based on the experience of the clinician. 53 Dermoscopy shows subsurface structures of the skin with the use of handheld devices that extend optical light ray penetration beyond the skin surface and minimize surface reection. In contrast, naked eye examination limits visualization of certain skin structures because the stratum corneum has reective properties. Dermoscopy plays an important role in the diagnosis of skin cancer. It has higher discriminating power than a clinical analysis. The report made by Marie-Lise Bafounta, Alain Beauchet, Philippe Aegerter, on the diagnosis of melanoma skin cancer shown that sensitivity and the speci-city 54 obtained by dermoscopy techniques were comparatively higher than clinical analysis. Skin lesion on the face was usually diagnosed by comparing the dermoscopic features of lentigo senilis and lentigo malinga 55 based on logistic regression method. Dermoscopy plays a crucial role in distinguishing the melanoma cells from benign cells with the help of various diagnostic tools 56 such as pattern analysis, ABCD rule, Menzies method, and 7-point checklist. The dermoscopic studies were made on 20 pigmented skin lesions 57 based on the pattern analysis and the results were found to be more accurate. Dermoscopy techniques are used to analyze the vascular structure 58 of melanocytic and non-melanocytic tumors based on their morphological behaviour of the skin. Dermoscopy extends its application in early diagnosis of malignant melanoma and differential 59 diagnosis of pigmented lesions of the skin by in vivo methods. The study was made on 2522 skin cancer subjects 60 based upon the accuracy rate, dermoscopy techniques showed better results than conventional technique, but the main drawback of this technique is that it requires a highresolution camera to capture the images of skin lesions.
Boundary detection technique which uses gradient vector ow for detecting the boundary is explained. The initialization method is automatic which makes the skin lesion border determination fully automated. 61 The unsupervised approach for skin lesion images based on modied version of JSEG algorithm for detecting the border is used. The border detection error is quantied by a metric which uses manually determined borders by the dermatologist. 62 An approach which is based on detecting the border using statistical region merging algorithm is reported. The border detection error is quantied by a metric which has three sets of dermatologist determined border as the base. This method is compared with four states of automated methods. 63 A new method to detect and visualize the pigment network structures in the images already extracted has been reported. This detection is based on the automatic detection of the edges of pigment network structure which forms cyclic graphs. The extracted image is rst pre-processed which gives the binary image; the binary image is then converted into graph and cyclic subgraphs which corresponds to the skin texture. These cyclic subgraphs are ltered to remove round structures. 64 The mobile application uses the image processing technique for the prevention of melanoma. This system comprises of a microscope and a smartphone. The microscope is to image the area of skin, whereas smartphone is to get the output. 65 The artefact is removed, and the borders of the lesion are detected by using border detection technique in the captured images. 39 Dermoscopy also employed in the diagnosis of melanoma condition in pregnant women. The study was made on 35 pregnant women 66 and the results showed that the pregnancy leads to signicant modications in pigmented skin lesions. Border detection 67 method was used in dermoscopy techniques to enhance the accuracy level of the diagnosis of skin cancer. Mostly for non-melanoma 68 skin cancer such as SCC and BCC are diagnosed by dermoscopic methods due to its high resolution. Using dermoscopy, the irregular streaks in melanoma can be detected by performing a three-way classication of streaks which involves regular, absent and irregular in a pigmented skin lesions. 69 Esteva et al., employed the deep convolutional neural network to classify melanoma and non-melanoma skin cancer. 70 Latest reports on dermoscopy techniques for skin cancer diagnosis are summarised in Table 2.

Ultrasound
The ultrasound signal originates from the mechanical oscillations of many crystals in a transducer, which is excited by electrical pulses, this effect is called as the piezoelectric effect. Piezoelectric materials are ideal for this. 72 The typical transducer consists of a disk-shaped piezoelectric element that is made vibrating by applying an electrical impulse via an electrode on each side of the disc. Likewise, the echo returning to the disk makes it vibrate, creating a small electrical potential across the same two electrodes that can be amplied and recorded. 73 In modern clinical scanners, the transducer consists of hundreds of small piezoelectric elements arranged in a 1D array packed into a small enclosure. 74 Thus the generated acoustic waves propagate in the uid or surrounding tissues, from where these structures reect the echoes which are collected back in the transducer which then transforms into electrical energy. 75 This signal is then processed to get the image of the tumours in a required dimensional form. Photoacoustic imaging (PAI), or optoacoustic imaging, is insensitive to photon scattering in the biological tissue and, unlike conventional optical imaging techniques, this technique makes highresolution optical visualization. [76][77][78][79] Multi-spectral optoacoustic tomography (MSOT) is an imaging method that produces high-resolution optical images in scattering media, including biological tissues. This technique is also known as functional photoacoustic tomography. MSOT illuminates tissue with light of transient energy, typically light pulses lasting 1-100 nanoseconds. The tissue absorbs the light pulses, and as a result undergoes thermo-elastic expansion, a phenomenon known as the optoacoustic or photoacoustic effect. This expansion gives rise to ultrasound waves (photo echoes) that are detected and formed into an image as shown in Fig. 7. [80][81][82] The process is carried out with the help of computer where the image displayed on the screen is useful in diagnosing the lesions. 83 The images are obtained in vertical sections are both the resolution and penetration vary according to the frequency. A retrospective study examined the usefulness of high-frequency ultrasound in the diagnosis of BCC through the identication of hyper-sonographic spots. This study concludes the conicts between hyper-sonographic spots versus hypersonic spots in differentiating BCC and melanoma. 12 Sonography is the pre-operative studies which measures the thickness of the tumor and internal echo density. 17,84 Ultrasound of B-scan technique is interfaced with the computer which uses 20 MHz frequency transducer in acquiring the image. 74 40 MHz ultrasound gives higher resolution than 20 MHz. 85 El Gammal et al. demonstrated the detection of melanoma using 100 MHz frequency which provided higher resolution and detailed visualization of skin layers. 86 High- Review frequency sonography which is coupled with color Doppler gives 100% specicity and 100% sensitivity in detection of melanoma. 87,88 Ultrasonography is used to produce 3-dimensional imaging and detects the border of the lesions. 89 White light digital photography in combination with reex transmission imaging, which is a form of high-resolution ultrasound used to classify the pigmented lesions. 90 Contrast-enhanced ultrasound is used to study the primary lesion and gives  information about the non-surgical treatment. Sonoelastography is used to study the mechanical behaviour of the skin lesions. 91,92 Melanoma appears as hypoechoic or as heterogeneous oval structure under ultrasound detection. 93 Detailed anatomic data about the primary tumor is provided by sonography. 94 Retrospective analysis is done by ultrasound for presurgical examinations. 95 Fine-needle aspiration cytology is also used as a minimally invasive technique of ultrasound in detecting melanoma. [96][97][98] Recent summarization of ultrasound techniques for skin cancer diagnosis is presented in Table 3.
The laser transmits nanosecond pulses of NIR light through the transducer into the tissue. This light is then absorbed inside the tissue producing a localized temporary thermoelastic expansion. This expansion leads to the emission of ultrasound, which is detected by the transducer. Finally, a photoacoustic image is generated and displayed in real-time.

Confocal microscopy
Confocal microscopy is a non-invasive technique that allows examination of the skin with cellular resolution. It uses a focused laser beam to illuminate a specic point within the skin and measures the reection of light from that point as shown in Fig. 8. Multiple points are scanned across an area parallel to the skin surface to construct a grayscale image. 100 Different depths can also be imaged to form optical sections. A confocal microscope includes a light source, a condenser, an objective lens and a detector. A confocal microscope uses point illumination and a pinhole in an optically conjugate plane in front of the detector to avoid out-of-focus signalthe name "confocal". 101 This conguration permits the collection of light from the single in-focus plane and dismissal of light from all out-of-focus planes. The light source illuminates a small 3-D spot (voxel) within a sample, such as skin, from which mirrored light is collected to produce a pixel. 102 The illuminated spot is then scanned horizontally over a 2-D grid to obtain a horizontal microscopic section. This process is known as "optical sectioning" of the image in a series of horizontal planes stacked vertically to produce an image pixel by pixel, with an axial thickness of 2-5 mm. 103 This property enables a confocal microscope to look at a slice in the body of a thick semi-transparent sample, whereas, conventional microscopes visualize all the planes contemporarily.
Confocal microscopy provides the capacity for direct, noninvasive, serial optical screening for thick, living specimens with least of sample preparation as well as a minimal improvement in lateral resolution, commonly called reectance confocal microscopy (RCM). In vivo RCM is a non-invasive technique that allows examination of the skin with cellular resolution. Resolution is almost comparable to the conventional histology. It has the advantage of allowing the clinician to do a "virtual biopsy" of the skin and obtain diagnostic clues while minimizing unnecessary skin biopsies. 104 Various skin disorders of which both hyper-pigmentary and hypo-pigmentary lesions can be detected using confocal microscopy. 105 Also in combination with Raman spectroscopy at various wavelength, confocal microscopy provide better cellular details. 106 The recent study involves the usage of vivascope 107 with the extended version which offers better imaging. From 488 to 700 nm, the laser source can be used to illuminate the tissue non-invasively to know the depth nature of tissues and its complications. 108 Multispectral polarized light imaging (MSPLI) and confocal microscopy combination provide better accuracy details in results rather an individual system. 109 Recent progress in confocal microscopy techniques for skin cancer diagnosis is presented in Table 4.

Raman spectroscopy
Raman spectroscopy is a technique which is used to discover various modes in a system which involves vibrational, rotational and other low-frequency modes. The detailed setup of Raman spectroscopy is as shown in Fig. 9. It depends on Raman scattering of monochromatic radiations, usually from a laser in the visible, near infrared and near UV rays. 116 In Raman scattering, inelastic collisions take place between the photons of an irradiating laser beam and the sample (or tissue) molecules. 117 The obtained spectra can be processed and analyzed in order to provide automated feedback at the time of measurement. 118 This system provides better sensitivity in differentiating the tissues. 119 One of the article deals with the classication of skin lesion using Raman spectra by a non-linear neural classier. 120 The classication model is highly probabilistic and automated. This model involves the feature extraction for Raman spectra and a fully adaptive robust feed-forward neural classier. The component analysis of Fourier transform Raman spectra of ex vivo basal carcinoma can be diagnosed by using the Nd:Yag laser at the wavelength of 1064 nm acting as an excited source in FT Raman RFS100/s spectrometer. 121 Thirty-nine set of human skin samples were collected. Among the 39 samples, 18 samples were found to be histopathologically diagnosed as non-diseased and 21 samples were found to be BCC. Raman microspectroscopy is mostly done on in vivo method. It is a noninvasive, automated and real-time application to diagnosis the non-melanoma skin cancer which includes both BCC and SCC which can calibrate the sample within less than 1 second. 122 By using the confocal Raman system 1 with a handheld probe, the Raman spectra of 21 suspected non-melanoma skin cancers in 19 subjects with matched normal skin spectra were collected to analyse the pathological condition. Small distinctive bands corresponding to specied Raman spectra of lipids and proteins are used to distinguish the skin lesions. Raman Spectra of various compounds can be analysed by using the partial least regression and discriminate analysis. 123 Raman spectroscopy plays a major role in the study of static and dynamic properties of the biologically signicant molecule in solution, living cells and cell culture etc. The contribution of various bio-compounds such as lipids and proteins in the tissue can be estimated by providing a linear least square tting model, to analysis the skin cancer in the tissue regions. 119 Raman spectroscopy can also be used as a developmental tool in medical diagnosis. By shiing the excitation energy of Raman spectroscopy from the visible to NIR regions, the uorescence components present within the normal cells can be minimized. The soware tools have been developed to interpret the spectra of the tissue to diagnose the affected areas of the tissues regions. 124 Cluster analysis plays a signicant role in the classication of Raman spectra of malignant tumours from normal cells. 125 Latest updates on Raman spectroscopy techniques for skin cancer diagnosis are summed in Table 5.

Fluorescence spectroscopy
Fluorescence spectroscopy is electromagnetic spectroscopy also known as uorometry or spectrouorometry that analyses the uorescence from the sample. 128 The light source used in this technique excites the electrons in the molecules, and the uorescence is analyzed from the samples as shown in Fig. 10. 129 This method has been employed in many biological, biochemical and environmental applications. 130,131 Laser-induced uorescence spectroscopy is used to detect the skin cancer by in vivo methods. A study was conducted by E. Borisova and his team on uorescence spectroscopy; they noticed that excitation spectrum are usually identical to the absorption spectrum, as the uorescence intensity is directly proportional to the absorption. 132 The non-melanoma skin cancer can be diagnosed by the uorescence polarization imaging technique, in which the tunable monochromatic light source along with CCD camera is employed to detect the tumour cells. 133 As a result, two uorophores were retained in tumours. Hyperspectral uorescence is the one of the technique used to diagnosis the skin cancer in which the two ultrafast lasers were used operating at the wavelength of 355 nm that excites the auto-uorescence between the bio-molecules that resides in the skin tissue. 134 Auto-uorescence technique is used to capture the images of the pigmented skin lesions, in which both the uorescence and reectance were studied. 135 The excitation source used in this method is bre coupled laser operated at the wavelength of 785 nm and the illumination of a light source is ltered by bandpass lter in order to acquire the uorescence imaging and the reectance imaging were obtained by the selection of components of the long pass lter. Auto-uorescence technique is used to select the sampling points for the Raman spectroscopy, based on that the classication of various skin lesions was performed. 136 Fluorescence spectroscopy is used to determine the distribution of biological substances by employing the uorescent agent. 137 The excitation source that causes uorescence has been integrated with ber optic probes to examine the tissue and it was recorded by using spectrograph. Fluorescence spectroscopy technique can also be used to classify the benign from malignant tumors with the help of uorescence spectra by in vivo method. As a result, the normal tissues were found to be the wavelength of 440 nm and the non-melanoma tissues were found to be at the wavelength of 436 nm. 138 Fluorescence spectroscopy has the ability to distinguish the cancerous cells from the non-cancerous cells. 139 The monochromatic radiation is used as an excitation source at the wavelength of 300 nm. From the tissue two sets of uorescence were emitted at the wavelength of 340 nm and 440 nm. The difference in intensity between the two uorescence will decide whether the tissue is a malignant or a benign tumour. 140 PDD of doubtful skin lesions relies on the uorescent properties of an exogenous drug or endogenous compound in response to a light source. Topical agents are utilized to stimulate the reproduction of endogenous photosensitizers that produce a photodynamic effect when exposed to the light of specic wavelengths and energy. Photodynamic therapy (PDT) is the combination of absorption light and the photosensitizer. 141 This technique also helps prior to surgery to detect the cancerous cells by using the photosensitizer. Fluorescence spectroscopy is performed on the tissues by using the photo sensitizer. 142 Non-melanoma skin cancer can be diagnosed by uorescence spectroscopy by both in vivo and ex vivo method. Auto-uorescence is used to detect the presence of an abnormality in malignancy. The uorescence intensity of  dermal collagen is generally lower in tumour than in the normal tissues. 143 Synchronous uorescence method with time-resolved uorescence technique can also be used for this applications. 144 Modern reports on uorescence spectroscopy methods for skin cancer diagnosis are summarised in Table 6.

Tetrahertz spectroscopy
Tetrahertz (THz) spectroscopy as shown in Fig. 11 comes under the spectroscopy technique which is used to detect and controls the properties of matter with the electromagnetic domain that are in the frequency range between a few hundred GHz to several THz. It is an imaging technique, which is used to detect the epithelial cancers. Time domain analysis is used to differentiate between diseased and healthy tissues, which are signicant statistically. The location of a tumour is well correlated and increased in terahertz absorption. Diagnosis of skin cancer using frequency domain analysis in terahertz spectroscopy is by using in vitro techniques and by using tetrahertz pulsed imaging (TPI). Then data manipulation is done to determine the spectral information which is related to depth. 146 The pulsed THz spectrometry is used to analyse the dielectric properties of the human skin. 147 Using this technique, the contrast between the BCC and normal healthy skin was found. The THz material parameter can encounter the dysplastic and non-dysplastic nevi pigmentation and it was compared with the normal healthy skin. A skin tissue model has been developed for the THz wavelength and has been estimated for a Monte Carlo simulation of the polarized light and scattering. The studies were made based upon the Mueller matrices in order to diagnosis the skin cancer. 148 In accordance with the novel optic system, it is used to capture the images of the neoplastic and the non-neoplastic colon tissue with close to the wavelengthlimited spatial resolution at the spatial resolution of 584 GHz frequency were employed to study the carcinoma tissues. 149 THz technology plays an important role in medical diagnosis. T waves also extends its application in both electrical and optical method to diagnose the skin cancer by means of plamonic photoconductive antenna of high sensitivity. 150,151 THz spectroscopy also used to analyse the biological tissues by integrating the IR camera with IR detector. 152 Using the THz imaging at the range of 1.39-1.63 THz the non-cancerous and cancerous cells can be differentiated. 153 It also extended its application in the domain of biomedical which ranges from the bio-molecules includes lipids, proteins, amino acids to cells and tissues. 154 It is used in sensing applications, involves detection of concealed objects, explosive detection of identications, space exploration and THz cancer detections. 155 Contemporary updates on THz spectroscopy approaches for skin cancer diagnosis are compiled in Table 7.

Optical coherence tomography
The backscattering of near-infrared light by tissue is measured by the technique called optical coherence tomography (OCT) shown in Fig. 12. OCT is a non-invasive method that allows 2 or 3-dimensional cross-sectional visualization of the microstructural morphology of tissues. 158 It provides better depth resolved image than ultrasound imaging. OCT is an in vivo technique which works on the principle of interference (Michelson interferometry) of infrared radiation and living tissues which is the major reason for high resolution. The interferometer in an OCT scanner splits a broadband source eld into a reference eld and sample eld. The sample eld focuses through the scanning optics and the objective lens to some point below the surface of the tissue. The back scattered beam from the tissue combines with the reected beam from the reference mirror, forming an interference pattern which is detected by the photodetector. 159 Structures of normal skin, including the epidermis, dermo-epidermal junction, dermis, hair follicles, blood vessels and sweat ducts can be clearly visualized by this method. This technique allows the differentiation between the benign and malignant epithelial tissue by the qualitative and quantitative assessment of OCT images. It enables the optical biopsy of epithelial lesions. 160 A report was made by Michael R. Hee, Joseph A. Izatt, Joseph M. Jacobson, James G. Fujimoto, and Eric A. Swanson, who made a signicant nding on OCT by introducing a transilluminance property in femtosecond laser. 161 Famous authors Alexander M. Sergeev and Valentin M. Gelikonov found that longitudinal scanning of human tissue has good accuracy comparing to transverse scanning. 162 Authors also used CCD camera to detect the photons but failed in showing the wider eld of view. 163 Cross-sectional image of human skin is obtained with higher accuracy. 164 It has been proved that longer wavelength can minimize the inuence of multiple scattering on image contrast and resolution; thus increase the effective penetration depth of OCT. 165 Combination of both OCT and optical Doppler tomography (ODT) shows better results in scanning speed of acquiring the image. 166 Even for detecting the inammation on the skin, lesions on the skin and other     infection over skin apart from tumour detection can also be performed by OCT. 167 Heterogeneous property of tissue shows that collagen level in tissue can also rate the tumor growth. 168 Later researchers like Aneesh Alex, Jessika Weingast and Martin Weinigel, reported that the combination of multi photon tomography (MPT) and OCT showed better resolution and high penetration depth. 169 To overcome the conicts in differentiating the non-tumour and tumour, deep margins are detected in the region of interest using the OCT. 170 Increased thermal property of tissue shows various dysfunction of metabolic activity. The epidermal region shows higher thermal coefficient in the cancer patient. 171 Also, the density of normal skin and tumour occurred in the skin varies. 172 OCT reduces the Mohs micrographic surgery. 173 Measurement of epithelial density can also be performed by OCT directly. 174 Currently; precise tumour depth is performed by three methods combination namely, OCT, optoacoustic and Raman spectroscopy which shows the better results. 175 Recent updates on OCT approaches for skin cancer diagnosis are compiled in Table 8.

Multispectral imaging technique
In multi-modal spectroscopy (MMS) or multispectral imaging, various linear polarizers of different wavelengths are used. The spectral and spatial information of the samples are simultaneously recorded by the multispectral image spectrometer, in which the acquired images from a monochrome camera is processed using spectral and polarization ltering that provides high contrasting images which is useful in identifying the pathological and morphological features of the suspicious skin lesions. [177][178][179] Refer to Fig. 13 shows the multimodal spectroscopy setup for skin cancer diagnosis. The report was made by Hagen Nathan, Kudenov Michael on multispectral imaging based on spectral bands. In this, the image analysis is done automatically and pattern recognition is used to identify lesions which in turns help in further biopsy. It plays an important role in the diagnosis of skin cancer by considering the parameters such as texture, asymmetry, border irregularities etc. 180 The images are obtained from the affected regions of the skin using the charge coupled camera along with eight narrow band lters ranging from 450 nm to 800 nm at the interval of 50 nm. 181 The features are extracted from the image using the principal components analysis. The characterizations of malignant and benign tumors are separated by spatial gray level co-occurrence matrix. 182 Multispectral imaging technique includes in vivo methods in which the images were analyzed at the wavelength between 483 nm and 950 nm at equal wavelength intervals. 183 Multispectral imaging employs the automated diagnosis of skin cancer using the neural network classier. The neural network has the ability to discriminate the malignant and benign tumor with the sensitivity of 80.4%. In the multispectral imaging system, the pigmented skin lesions are separated by using the automatic segmentation algorithm. 184 Usually, in multispectral imaging technique, the images of the skin lesions are analyzed as 2-dimensional and 3-dimensional images at the wavelength range from visible to the infrared spectrum. 185 Multispectral imaging technique facilitates the analysis of skin lesions, vascular depth and sub-cellular pigmentations. 76 It uses the radiometric measurement to analysis the malignant and benign tumors. 186 A multispectral imaging technique employs the multispectral camera to capture the skin lesions images. The images are analyzed based on the spectral ranges between 450 nm to 950 nm which is used to discriminate the melanoma from nevus cells. 187 The self-developed soware is used in multispectral imaging technique to analyze the skin chromophores which is a non-contact method. 188 Multispectral imaging method nds its applications in the detection of melanoma skin cancer by considering the parameters such as melanin index and erythema index. 189 It also provides the decision of undergoing biopsy from the pigmented lesion using multispectral digital skin lesion analysis (MSDSLA) device. 190 Multispectral imaging technique enhances the diagnosis by analyzing the vascular depth of the skin lesion. 191 The vascular depth can be interpreted with the help of 6 layered skin models. Reported the multispectral imaging techniques combined with 3-dimensional imaging sensor to detect the skin cancer. 192 Contemporary reports on MMS methods for skin cancer diagnosis are summarised in Table 9.
Thermography Infrared radiation is emitted by every object which possesses the temperature absolute zero point. 194 The thermal imager determines the temperature of the object's surface based on the intensity of infrared radiation making it visible to the human eye with the thermal image. This process is referred as thermography. Thermal imager translates the wavelength from the infrared to the wavelength which is visible to human eye as shown in Fig. 14. 195 This is the principle which is used to detect the skin cancer using thermography. 196 Thermography has been employed in medicine for various applications. 197 However, thermography overcomes all the shortcomings that other methods had. A study performed by Cila Herman 198 found a signicant nding on recently introduced quantitative dynamic infrared system QUAINT which measures differences in the infrared emission between healthy tissue and lesion during the thermal recovery process aer removal of cooling stress. Thermistors, thermocouple and other advanced sensors are used to detect the variation in temperature. 199 The rise in temperature is a signicant symptom in the skin cancer patient. Nano type sensors are involved in detecting the changes in the biological analyses and the obtained information is processed to the local server in the hospital which is directed to the medical professionals in the hospital.  Table 10.

Electrical bio-impedance
The living cells membrane act as an electrochemical membrane, which is a semi-permeable membrane allows only    217 The measurement was done by using electrodes of a two-point measurement. 218 The measurement of electrical impedance shows the morphological changes which relate to the growth of skin cancer lesion. 219 The multifrequency impedance spectra are used to detect the electrical bioimpedance for different skin lesions. 220,221 The non-invasive probe separates the nevi from BCC with 96% sensitivity and 86% specicity. 222 The impedance is measured by use of impedance spectrometer between 1 kHz and 1 MHz of various types of skin cancers including melanoma. This improves the signal to noise ratio. By applying small alternating voltage and comparing the measured current with voltage, the impedance between two electrodes is measured. 218 The change in impedance is used to detect the skin cancer by using the information about cell shape, structure and orientation, integrity of cell membranes, relative properties of intra and extra-cellular uids and ionic composition. This device consists of the system on chip impedance converter, AD5933, an ultra-low power MCU, MSP430F6638, and electrodes. The electrode used is Ag/AgCl electrodes, which is non-toxic and low cost. 18 A smart phone based early cancer detection using EIT patch electrodes has been demonstrated. 223 Electrical impedance spectroscopy (EIS) algorithm is used in detecting the lesions which have high sensitivity in the detection of malignant melanoma. 224 Electrical impedance helps in differentiating the cancerous cell from noncancerous cell at the range of 1 KHz to 2.5 MHz. 225 Bioimpedance spectroscopy also available in portable form which is used to monitor the physiological system. 226 Latest progress on electrical bio-impedance approaches for skin cancer diagnosis are compiled in Table 11. Fig. 16 shows the different devices used for electrical bio-impedance measurement.

Tape stripping
Tape stripping is the removal stratum corneum layers as demonstrated in Fig. 17. Adhesive tape is used to remove the cell layers of the stratum cornea from the skin area. The number of corneocytes and a considerable amount of penetrated formulation will be present in the tape strips. 228 Tape stripping is also known as Epidermal Genetic Information Retrieval (EGIR). 90 The sample cells from the stratum corneum by the use of adhesive tape stripping from which RNA is isolated, ampli-ed and gene expression is noted. 229 This adhesive tape has the RNA genetic expression prole which is useful in differentiating melanoma from nevi. 230 With the help of 17-gene genomic biomarker, the presence of melanoma is detected from the harvested specimens. This analysis is carried out using ribonuclease protection assay (RPA). This analysis is used to differentiate malignancies on the basis of gene expression proles. DermTech's Epidermal Genetic Information Retrieval (EGIRTM) uses a custom adhesives lm to collect surface skin samples. This method is painless and it can also be virtually applied to the skin surface. Also, this method has 100% sensitivity and 88% of specicity. 231,232 It is a rapid and convenient method 233 to contrast various changes in the cancerous patients. It also differentiate lentigo maligna melanoma from solar lentigo, which is sometimes difficult with histology. 234 This method is widely used as a robust bedside system for subject with numerous dysplastic nevi. 235 Can be performed from the extracted RNA only by freezing it up to À80 C. Only then the frozen sample is treated. 236 Early detection can be achieved if tape stripping toluidine blue (TSTB) 237 is combined with ABCDE rule. In 2011, Wachsman and colleagues showed that EGIR performed on 202 melanocytic lesions identied 312 genes that were capable of differentiating normal skin, nevi and melanoma. 229 Recent progress on tape stripping techniques for skin cancer diagnosis is compiled in Table 12.

Multi photon scanning
Multiphoton scanning is also known as two-photon excitation microscopy, non-linear or two-photon laser scanning microscopy which is an alternative to confocal and deconvolution microscopy that provides distinct advantages for threedimensional imaging. 239 The principle of multiphoton scanning is the nonlinear interactions between photons and matter. Interaction of two photons with the same molecule at the same time occurs in two-photon excitation. This two-photon excitation causes uorescence. Then, the single photon is required to process the uorescence of right energy to interact with the  molecule. Excitation is not produced when the two-photon excitation is used through the specimen, as the photon density is adequately high to cause two-photon absorption. Excitation occurs in the focal spot of the microscope. 240 The complete setup is shown in Fig. 18. In particular, two-photon excitation excels at imaging of living cells, especially within intact tissues such as brain slices, embryos, whole organs, and even entire animals. Two-photon excitation microscopy provides great utility for dynamic imaging of living cells in thick specimens, such as intact tissue. The technique does many possible experiments in which conventional imaging cannot be performed. 241 Researchers like Barry R. Masters, Peter T., C. Enrico Gratton, who worked on multiphoton laser scanning microscopy (MPLSM) demonstrated that the power entering the microscope can be controlled by Glan-Thompson polarizer which reduces power dissipation and increases the efficiency. 242 In order to nd the difference in accuracy for confocal laser scanning microscopy (CLSM) method and multiphoton microscopy, various histological tests were undertaken by Barry et al. and inferred that both methods have similar accuracy. 243 However, resolution in terms of three-dimensional gene expressions and depths of tumor are more accurate in MPLSM. 244 Later, Timothy P. Padera and colleagues found that epiluminescence property in MPLSM showed the better result in terms of gene expression. 245 The major cause of skin cancer is UV rays which makes the skin to lose the collagen and elastin material in the skin are widely affected. 246 Morphological altering due to skin aging is investigated by pathological characterization. 247 Morphological features in 14 people who are non-melanoma are performed with MPLSM. 248 Later 83 person were tested considering the morphological features of non-melanoma and melanoma skin cancer subjects, which showed a major difference and useful in diagnosis. 249 Further to prove the accuracy of both CLSM and MPLCM in differentiating skin layer tumour, Martin Johannes Koehler and his colleagues performed pathological tests. 250 MPLSM provides high-resolution uorescence imaging, allowing visualization of cellular and subcellular structures of the epidermis and upper dermis. 251,252 MPLSM based morphologic features of skin cancer images are comparable to traditional histopathology. 252 Latest updates on multiphoton scanning for skin cancer diagnosis are summed in Table 13.

Conclusion
This article briey summarizes the techniques which deal with the non-invasive approach towards the detection of skin cancer. The different techniques that use various modalities to diagnose the skin cancer are demonstrated with recent updates. However, every system equally shares the advantages and disadvantages in all aspects. We found that the use of TBP surveillance is must in skin lesions and dysplastic nevi which show the result but these results vary according to the experience of the doctor that lead to misinterpretation in some cases while differentiating BCC and melanoma. Also, the age of onset and frequency at which it should be performed remains unclear. This makes limited data to diagnose. Some studies show that there was no evidence in practicing photography as a screening tool in the United States by the dermatologists; instead, they prefer a dermoscopy. A handheld dermoscope is implemented in dermoscopy that provides at least 10-fold magnication of the skin lesions, and almost all primary cares adopt this mode of diagnosing. And no study reported that the survival of melanoma has increased due to dermoscopy evolution. This is mainly due to the same fact that the result varies according to the experience of physician that makes less trust in Review public towards the clinical screening. As far as the reviews from several abstracts, this could be changed by using a combination of both photography and dermoscopy in heterogeneous aspects that resolves all cons of the conventional techniques and provide an accurate result. The efficiency of the results is based on the usage of algorithms, devices, training and practitioner's experience. Nowadays to avoid false positive results dermatologist use confocal laser scanning microscopy which aid in the assessment of skin lesions by providing high resolution, high sensitivity, high specicity images that are similar to the histopathological images. The only downside of this technique is the high cost that could not be undertaken or affordable by economically backward people. Unlike other techniques, this is unique which is not engaged in comparison with other techniques (dermoscopy) trials to prove the accuracy of results. However, the reported outcomes were mainly focused on the lesion characterizations of CSLM and its test accuracy. Also, imaging modalities like high-frequency sonography, confocal microscopy and OCT have limited availability and are usually found in research units and selected dermatology departments. They are predominantly practiced on an experimental and research basis. Despite their limitations, these high-resolution techniques can give us valuable information on supercial BCC. Nevertheless, awareness of their penetration intricacies must be kept in mind because they may leave cancerous lesions that affect the lower dermis and deeper layers out of their elds of view. Many other modalities like high-frequency ultrasound sonography, PDD, Raman spectroscopy, uorescence spectroscopy and THz spectroscopy are still in limited practice in case of skin cancer detection. Techniques like thermography which needs optimum instrumentation for the recording purpose and widely affected by the external noise though it provides a thermogram at high resolution. In case of MPLSM, bio-electrical impedance, OCT, tape stripping, multispectral imaging and fully automated computer-based analysis are considered as investigational that need clinical approval in different usage modalities. Also there are not much evidences or trials undertaken to prove the accuracy with conventional histological determination.

Future directions
The delay of implementing dermoscopy widely in the hospital and other healthcare is due to the lack of training among the physicians. This made an adverse impact in getting such product to the market eld. However, usage of higher end resolution camera and instrumentation set up must provide a better result that is indirectly failed to achieve due to insuf-cient training. Thus proper training, knowledge and experience sharing of practitioners are must for better diagnosis. The use of computer-aided diagnosis in nding the skin lesions could be furtherly enhanced using various pre-processing techniques and segmentation method that would be more informative and satisfying the needs. It is the need for analyzing various attributes and can also be included in dermoscopic techniques so that it is easier to link with computer-aided analysis. The optical methods like OCT and PDD are still under investigation stages; this could be improved by dening the borders of suspicious lesions particularly in non-melanoma skin cancer and the common skin disorders. Further exploration is required in using the laser for detecting melanoma.
The features of the future systems should be (i) low cost, (ii) compact and portable, (iii) less diagnosis time, (iv) comfortable (no biopsy), (v) high sensitivity and accuracy. Combination of multiple techniques should be explored for high-quality output and accuracy. Clinicians, researchers and practitioners should step forward towards arriving at a standard protocol and knowledge sharing database, thus help in providing better healthcare relating to skin cancer.

Conflicts of interest
There are no conicts to declare.