Vigneswaran
Narayanamurthy
*ab,
P.
Padmapriya
c,
A.
Noorasafrin
c,
B.
Pooja
c,
K.
Hema
c,
Al'aina Yuhainis
Firus Khan
d,
K.
Nithyakalyani
c and
Fahmi
Samsuri
b
aInnoFuTech, No: 42/12, 7th Street, Vallalar Nagar, Pattabiram, Chennai, Tamil Nadu 600072, India. E-mail: PEL13006@stdmail.ump.edu.my
bFaculty of Electrical and Electronics Engineering, University Malaysia Pahang, Pekan 26600, Malaysia
cDepartment of Biomedical Engineering, Veltech Multitech Dr. RR & Dr. SR Engineering College, Chennai 600 062, India
dDepartment of Biomedical Science, Faculty of Allied Health Sciences, International Islamic University Malaysia, 25200 Kuantan, Pahang, Malaysia
First published on 6th August 2018
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.
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 87110 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 132000 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
Fig. 1 shows (a) recent stats and (b) stages of skin cancer. (Data Source: NCI SEER Cancer Stat Facts).7 |
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 after-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 often 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.
Fig. 5 (a) Illustrates the total body imaging using photographic devices and (b) shows general steps in image processing technique for skin cancer diagnosis. Reproduced under CC license form ref. 43. |
Lynn T. Dengel, Gina R. Petroni and Craig L. Slingluff et al. had a significant finding 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 classified 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 fitted.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 filters 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 quantified and used for classification.28 In the classification part, bringing out the algorithm in the examination process combines visual processing with deep learning.29 In deep learning, the computer is trained to figure 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-shift 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 identification of skin lesion area. Using TBP, different shape, texture and colour features are extracted and malignant melanomas have been classified 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 significantly decreases when melanoma images are processed. Geometrical and colour features were extracted from the binary masks generated by the algorithm for classification, 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 artificial 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 qualification 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, specificity 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 field of image processing, pre-processing, feature extraction, segmentation and classification of demographics images.34 Computer-aided melanoma diagnosis is used to provide the support to the physicians in enhancing the sensitivity and specificity 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.
S. no. | Description | Upside | Downside | Accuracy, sensitivity, specificity | Ref. |
---|---|---|---|---|---|
1 | Optical threshold is determined in region based segmentation, whereas an approximate closed elastic curve between the recognized neural network edge patterns is determined in neural network edge detection | Best performance for lesions with a range of different border irregularity | Iterative thresholding technique is required | — | 25 |
2 | Different digital images have been analyzed based on unsupervised segmentation techniques followed by feature extraction techniques | — | Provides erroneous results if not all analysis carried out accordingly | Accuracy (Acc): 0.9748, sensitivity (Se): 0.8624, specificity (Sp): 0.9857 | 31 |
3 | The mathematical morphology is used for segmenting the skin lesion in this method | Identification of skin lesion is clear | High cost and consume more time | Se: 85.71% | 32 |
4 | Delaunay triangulation is an accurate process for detecting the benign lesions, doesn't require training. Extraction of binary mask of the lesion is made | Accuracy level is high | Time consumption is more | Se: 93.5%, Sp: 87.1% | 33 |
5 | Polarized light is used for illumination and found that if plane of polarization is parallel then enhanced image is obtained if the plane of polarization is perpendicular then wrinkles or surface details disappear | Patient worry is reduced | Consumes more time | — | 44 |
6 | Smallest lesion of 1 mm is detected using photography technique and differentiation of melanoma and non-melanoma is detected using high resolution camera | Thinner melanomas are detected | Cost is high | — | 24 |
7 | Lower melanoma site region is detected using digital photography | Patient self-screening can be done | Cost is high | — | 45 |
8 | Pigmented lesion in the human body is detected using TBP which is useful in detecting the early stages of melanoma | Minute lesions can be differentiated using this technique | Instrumentation set up covers larger area | — | 46 |
9 | Incipient melanoma is detected using combination of both dermoscopy and photography in which accuracy of result is obtained | Not only detects the macroscopic changes and also the detection of dermatoscopic changes over time | Considerable proportion of melanomas misclassified as benign | — | 47 |
10 | Melanoma is detected in the initial growth stage itself with diameter of >6 mm and baseline images were obtained | — | Cost is high | — | 48 |
11 | Found that combination of total body examination, TBP and dermoscopy provides accurate result in detecting the early stage of melanoma | High resolution and sensitivity | Cost is high | — | 49 |
Fig. 6 (a) and (b) shows different types of dermoscopes. (a) Reproduced under CC license from ref. 71. |
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 reflection. In contrast, naked eye examination limits visualization of certain skin structures because the stratum corneum has reflective 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 specificity54 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 malinga55 based on logistic regression method. Dermoscopy plays a crucial role in distinguishing the melanoma cells from benign cells with the help of various diagnostic tools56 such as pattern analysis, ABCD rule, Menzies method, and 7-point checklist. The dermoscopic studies were made on 20 pigmented skin lesions57 based on the pattern analysis and the results were found to be more accurate. Dermoscopy techniques are used to analyze the vascular structure58 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 differential59 diagnosis of pigmented lesions of the skin by in vivo methods. The study was made on 2522 skin cancer subjects60 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 high-resolution camera to capture the images of skin lesions.
Boundary detection technique which uses gradient vector flow 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 modified version of JSEG algorithm for detecting the border is used. The border detection error is quantified 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 quantified 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 first 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 filtered 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 women66 and the results showed that the pregnancy leads to significant modifications in pigmented skin lesions. Border detection67 method was used in dermoscopy techniques to enhance the accuracy level of the diagnosis of skin cancer. Mostly for non-melanoma68 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 classification 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.
S. no. | Description | Upside | Downside | Acc, Se, Sp | Ref. |
---|---|---|---|---|---|
1 | Dermoscopy significantly has higher discriminating power than the clinical analysis. The sensitivity and the specificity ranges obtained from diagnosis of melanoma were found to be 0.75 to 0.96 and 0.79 to 0.98 respectively | More accurate than clinical examination for the diagnosis of melanoma in a pigmented skin lesion | Requires experience for better diagnosis | Se: 0.75–0.96 and Sp: 0.79–0.9 | 54 |
2 | Compared the dermatoscopic features of lento senilis and lentigo malinga on the face by using logistic regression analysis | Analysis is easy | Resolution is low | Se: 93.8% and Sp: 52.3% | 55 |
3 | Dermoscopic images of the skin lesions were analyzed using 2-step diagnostic methods | Good computational capability | Various algorithms are needed | Se: 64.8% and Sp: 72.8% | 56 |
4 | Dermoscopic images of 20 pigmented skin lesions were evaluated based upon the menzie's method and ABCD rule and pattern analysis. It was found that results of pattern analysis were comparatively more accurate than any other methods | Web-based training is an effective tool for teaching dermoscopy | Involves training of practioners | Acc: 62.8% | 57 |
5 | The vascular structure of melanocytic and non-melanocytic skin tumours were evaluated based upon the morphological features | High resolution | Analysis of distinctive vascular structures is required | Se: 81.1% | 58 |
6 | Dermoscopy helps in early diagnosis of melanoma cancer by in vivo methods | Early diagnosis of melanoma | Resolution is low | Acc: 5% to 30% | 59 |
7 | Primary physicians made study on 2522 skin cancer subjects and the accuracy level were compared with the existing system | Improves the ability of physicians to triage lesions suggestive of skin cancer | Involves physician training, algorithm and expert consultation | Se: 54.1%, Sp: 71.3% | 60 |
8 | The studies were made on 35 healthy pregnant women and 35 age-matched female controls. The analysis showed that the pregnancy leads to significant modifications in PSL, especially with respect to globules, pigment network, and architectural order or disorder | Local intensity variant is done | Consumes more time to diagnosis the cancer | Se: 79.3%, Sp: 93.18% | 66 |
9 | In this method, the dermoscopic images obtained from the subject were analyzed based on border detection | Fast and accurate border detection | Requires a series of process algorithms | — | 62 |
10 | Through the dermosccpy technique the non-melanoma skin cancer were detected and the accuracy was high | Identification of skin lesion is clear | Only detects the non-melanoma cancer. | — | 68 |
11 | In dermoscopic images, the diagnosis of melanoma can be easily identified by irregular streaks | Accuracy level is high | Only includes the analysis of irregular streaks | Acc: 76.1% | 69 |
12 | Demonstrated the classification of skin lesions using a single deep convolutional neural network (CNN), trained end-to-end from images directly, using only pixels and disease labels as inputs | The CNN achieves performance on par with all tested experts | Involves CNN computing platform | Acc: 72.1 ± 0.9% | 70 |
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. 7.80–82 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 identification of hyper-sonographic spots. This study concludes the conflicts 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-frequency sonography which is coupled with color Doppler gives 100% specificity 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 reflex 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 pre-surgical examinations.95 Fine-needle aspiration cytology is also used as a minimally invasive technique of ultrasound in detecting melanoma.96–98 Recent summarization of ultrasound techniques for skin cancer diagnosis is presented in Table 3.
S. no. | Signal used | Description | Upside | Downside | Acc, Se, Sp | Ref. |
---|---|---|---|---|---|---|
1 | 20 MHz | Sonography with 20 MHz frequency is used as an pre-operative diagnosis of malignant melanoma | Hypo-echoic is identified easily with greater tumour thickness | Definite differential diagnosis is not possible | — | 17 |
2 | 40 MHz | It is the superior, non-invasive method for the detection of BCC | High resolution | Microscopic examination is still required | — | 85 |
3 | 7.5–20 MHz | The probe which is used has a high frequency which can detect the three layers of the skin | It is used to find the depth of melanoma | Resolution is less | — | 84 |
4 | 100 MHz | It uses imaging technique which uses the transducer of frequency 95 MHz, whose resolution is high | Detailed visualization of the upper skin layers as compared with 20–40 MHz sonography | Image acquisition takes much time | Resolution (res): 8.5 μm | 86 |
5 | 20 MHz | B-scanning ultrasound with 20 MHz frequency is interfaced to the computer and assessed acoustic shadowing and entry echo line enhancement for 29 basal cell papillomas (BCPs) and 25 melanomas | High-performance screening tool to assist in the discrimination between BCP | Not good in discriminating between benign naevi, and melanoma. Accuracy of detecting melanoma is not that accurate | Se: 100%, Sp: 93% | 74 |
6 | 7.5 MHz | Fine-needle aspiration cytology procedures were performed in melanoma patients with palpable tumors or non-palpable, ultrasonically suspicious lesions. Device was interposed between needle and syringe which was vacuum valued | The procedure is painless and well tolerated by patients | Ultrasound guides the process and is minimally invasive | Se: 94.6% and Sp: 100.0% | 96 |
7 | 7.5–15 MHz | Ultrasound B-scan is a technique used for detection of regional melanoma | Quick and reliable | Only in combination with cytologic examination of fine-needle aspirates, allows definite diagnosis | Se: 89.2%, Sp: 99.7% | 97 |
8 | 13 and 20 MHz | Intralesional vascularization is assessed by the use of coupled high-frequency sonography and color Doppler | Simple and reliable. | Requires experienced operators | Se: 100% and Sp: 100% | 87 |
9 | Less than 15 MHz | Ultrasound is used as an imaging technique which uses transducer of superior designs | Parameters are quantified. | Photo damage occurs | — | 83 |
10 | 7.5–10 MHz | Ultrasonography is used to produce a 3-dimensional size and outline of the lesions | Used to distinguish benign and malignant tumors | The quality of information depends heavily on the examiner's skill and experience | Se: 99.2%, Sp: 99.7% | 89 |
11 | 7.5–13 MHz | Evaluated the distribution of metastases from scar of primary to the regional lymph node with subjects of cutaneous melanoma | — | Suitable only for follow up cases | — | 99 |
12 | 20 MHz | Ultrasound technique distinguishes the nevi from melanoma using probes of different frequencies | — | Frequency of probes differs for various types of lesions | Se: 100%, Sp: 79% | 90 |
13 | 20, 75, 100 MHz | Thickness of melanoma is assessed using different frequency in the transducer probe. Among all the frequency, using 7.5 MHz in transducer proves to be more versatile in detecting the thickness of melanoma | Information about the dermatology is provided completely | Sometimes may be overestimated due to infiltrate process | Se: 99%, Sp: 100% | 88 |
14 | 10–100 MHz | Asymmetry of the tumour is determined by applying the gel on the skin surface and examining with the ultrasound with different frequencies | Deepest tissues are also analysed. | Pigments (melanin) of the skin cannot be detected | Se:92% | 91 |
15 | 20–100 MHz | The sonographic findings in common benign and malignant skin tumors, inflammatory dermatologic diseases, and ungual and cosmetic conditions, among others, are considered and compared with various systems | Discrimination is better for different cutaneous layers | Visualization problem arises | — | 93 |
16 | 15–22 MHz | Demonstrated use of sonography in facial BCC diagnosis. A comparative discussion was reported with other methods of assessing tumours and invasion besides sonography | Doesn't have any radiations | Provides only preliminary detection | — | 94 |
17 | 20 MHz | High resolution ultrasound is used to detect metastases in melanoma subject by monitoring lymph nodes. | Early detection of recurrences can be made. | Not clinically used due to high cost | Se: 65%, Sp: 99% | 98 |
18 | 20 MHz | Hyperechoic spots of BCC tumors are analysed using retrospective analysis of ultrasound examinations | Less expensive | Quality depends upon operator's skill | Se: 79%, Sp: 53% | 95 |
19 | 7.5–100 MHz | Reported that sonography was effective at determining lesion thickness, defining lesion borders, and aids to identify whether lymph nodes are metastatic | Can be used to obtain measurements of skin cancer lesion thickness and can define lesion boarders. It can further delineate metastatic lymph nodes | Applicable only for screening method and not used widely by physicians due to lack of data's proving its accuracy level | Se: 90.74%, Sp: 98.89% | 92 |
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.
Fig. 8 Confocal microscopy imaging technique. In image (i) (A and B) showing hand held RCM probes and (C and D) showing wide probe RCM, reprinted from ref. 110 with permission from Elsevier, and (ii) showing its mechanism, reproduced under CC license from ref. 111. |
Confocal microscopy provides the capacity for direct, non-invasive, serial optical screening for thick, living specimens with least of sample preparation as well as a minimal improvement in lateral resolution, commonly called reflectance 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 vivascope107 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.
S. no. | Signal used | Description | Merits | Demerits | Acc, Se, Sp | Ref. |
---|---|---|---|---|---|---|
1 | 633 nm | Imaging is based on the detection of singly backscattered photons from the optical section and contrast is due to the relative variations in refractive indices and sizes of organelles and microstructures | RCM imaging with optical sectioning of 2–5 μm and resolution of 0.5–1.0 μm can be performed | Results vary according to the expertise in the clinical field | For melanoma, Se: 93% and Sp: 78% and for BCC, Se: 92% and Sp: 91% | 112 |
2 | 820 nm | Light emitted by a laser diode (a safe wavelength for patient and operator) is reflected, directed through a small pinhole, and arrives at the detector, where an image is formed of horizontal sections of the different layers of an area of the skin depicting intracellular structures | Used in combination as complementary imaging tools, all those techniques will provide detailed, higher sensitivity, information about field cancerization, with increased specificity for higher accuracy of diagnosis and delivery of optimal treatment | High cost | — | 101 |
3 | 450–650 nm | Testing the feasibility of a new mosaicing algorithm for perioperative RCM imaging of non-melanoma skin cancer margins on patients during Mohs micrographic surgery (MMS) | Video acquisition and video-mosaicing have improved RCM as an emerging tool | Cost is high | — | 113 |
4 | 450 nm | Described dermoscopic and confocal features of 2 couples of similar lesions in order to achieve the correct diagnosis and the best therapeutic approach | Simple technique to implement | Difficult to diagnose lesions such as pink tumors | — | 114 |
5 | — | Reported that in three children RCM examination of cutaneous lesions of Langerhans cell histiocytosis showed specific signs correlated to histopathology | Aids in decision making for biopsy | Cost is high | — | 115 |
6 | — | Confocal microscopy a non-invasive tool for detection of various skin disorders such as melasma and vitiligo or melanocytic lesions such nevi and melanoma | Both hypopigmentary and hyperpigmentary disorders can be evaluated using confocal microscopy | Cost of equipment is expensive | — | 105 |
7 | 720–950 nm | Development of integrated system of both confocal microscopy and Raman spectroscopy in the evaluation of melanoma | In depth microstructure analysis is performed using this microscopic technique | Integration of this system cost is high | — | 106 |
8 | 830 nm | Multimodality imaging technique using high resolution microscope, different cellular level features are observed | Among the non-invasive techniques, reflectance confocal microscopy imaging is unique to provide cellular details | Difficulty in distinguishing dendritic melanocytes in pagetoid pattern from Langerhans cells | Se: 96.6% Sp: 89.2% | 107 |
9 | 488 to 700 nm | Point source of light emitted by laser chamber falls on the tissue layer to provide the refractive property of normal and melanocytic skin | Real time imaging can be performed | Detection of tumor in the superficial layer with RCM is limited | Se: 97.3%, Sp: 72.3% | 108 |
10 | 630, 660 and 830 nm | Combination of confocal microscopy and multispectral polarized light imaging (MSPLI) provide imaging in cellular level | Good correlation of result equal to histology | Small field of view | — | 109 |
Fig. 9 Raman spectroscopy technique. In image (a) showing hand held probe, (Courtesy of Verisante Technology Inc., Canada; with permission) and (b) showing the Raman spectroscopy mechanism. |
One of the article deals with the classification of skin lesion using Raman spectra by a non-linear neural classifier.120 The classification model is highly probabilistic and automated. This model involves the feature extraction for Raman spectra and a fully adaptive robust feed-forward neural classifier. 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 non-invasive, 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 system1 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 specified 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 significant 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 fitting 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 shifting the excitation energy of Raman spectroscopy from the visible to NIR regions, the fluorescence components present within the normal cells can be minimized. The software 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 significant role in the classification 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.
S. no. | Signal used | Description | Merits | Demerits | Acc, Se, Sp | Ref. |
---|---|---|---|---|---|---|
1 | 1064 nm | Using the near infrared Raman spectroscopy, the molecular structures of normal and abnormal tissues are analysed | Skin lesions produced reproducible and unique spectra | Spectra are complex and an objective method for their interpretation remains to be developed | Acc: 94.8% | 120 |
2 | 1064 nm | Near-infrared Fourier transform (NIR-FT) Raman spectroscopy was used to study the molecular alterations in the most common skin cancer, BCC | Can differentiate the normal cells from BCC | Needs neural network for classification | Acc: 95% | 121 |
3 | 1064 nm | Raman spectra of normal tissues from benign and malignant can be differentiated by cluster analysis | Can differentiate the normal and abnormal tissues | Complex analysis is required | — | 123 |
4 | 1064 nm | Nonlinear neural network is used to perform the Raman Spectra on the skin lesions to classify the cancer cell | The framework is highly automated | Complex algorithms are used | Acc: 80–95% | 119 |
5 | 1064 nm | Basal carcinoma can be diagnosed by using the Nd:Yag laser at the wavelength of 1064 nm | This technique is applied to all types of tissues | Only analysis the non-melanoma cancer | Se: 83% | 124 |
Sp: 100% | ||||||
6 | 785 nm | Demonstrated an integrated real-time Raman spectroscopy system for skin evaluation and characterization, which combines customized hardware features and software implementation | Improved the signal-to-noise ratio by 16-fold | Involves a series of processing stages | Acc: 100% | 125 |
7 | 825 nm | Demonstrated the capability of Raman micro spectroscopy to provide differential diagnosis of BCC, SCC, inflamed scar tissue, and normal tissue in vivo | High accuracy | Only diagnosis the non-melanoma skin cancer | Acc: 95% | 126 |
8 | 1064 nm | Discriminate analysis and partial least square method is used to classify the cancer tissue | Good diagnostic accuracy | Detection needs optimum instrumentation setup | Se: 95–99%, Sp: 15–54% | 116 |
9 | 830 nm | Linear least square fitting models is used to determine the biomolecules contributions | The distribution of lipids, protein, elastin, collagen compounds in normal and the abnormal tissues can be diagnosed | Samples becomes over heat due to high intense laser | Acc: 91.7% (actin vs. nucleic acid), Acc: 93.1% (collagen vs. elastin vs. melanin) | 127 |
Fig. 10 Fluorescence spectroscopy. It is used to detect the skin cancer with the help of fluorophores. Reprinted from ref. 145 with permission from Elsevier. |
A study was conducted by E. Borisova and his team on fluorescence spectroscopy; they noticed that excitation spectrum are usually identical to the absorption spectrum, as the fluorescence intensity is directly proportional to the absorption.132 The non-melanoma skin cancer can be diagnosed by the fluorescence 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 fluorophores were retained in tumours. Hyperspectral fluorescence 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-fluorescence between the bio-molecules that resides in the skin tissue.134 Auto-fluorescence technique is used to capture the images of the pigmented skin lesions, in which both the fluorescence and reflectance were studied.135 The excitation source used in this method is fibre coupled laser operated at the wavelength of 785 nm and the illumination of a light source is filtered by bandpass filter in order to acquire the fluorescence imaging and the reflectance imaging were obtained by the selection of components of the long pass filter. Auto-fluorescence technique is used to select the sampling points for the Raman spectroscopy, based on that the classification of various skin lesions was performed.136 Fluorescence spectroscopy is used to determine the distribution of biological substances by employing the fluorescent agent.137 The excitation source that causes fluorescence has been integrated with fiber 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 fluorescence 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 fluorescence were emitted at the wavelength of 340 nm and 440 nm. The difference in intensity between the two fluorescence will decide whether the tissue is a malignant or a benign tumour.140
PDD of doubtful skin lesions relies on the fluorescent 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 specific 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 fluorescence spectroscopy by both in vivo and ex vivo method. Auto-fluorescence is used to detect the presence of an abnormality in malignancy. The fluorescence intensity of dermal collagen is generally lower in tumour than in the normal tissues.143 Synchronous fluorescence method with time-resolved fluorescence technique can also be used for this applications.144 Modern reports on fluorescence spectroscopy methods for skin cancer diagnosis are summarised in Table 6.
S. no. | Signal used | Description | Merits | Demerits | Acc, Se, Sp | Ref. |
---|---|---|---|---|---|---|
1 | 300 nm | The tumours are differentiated by the fluorescence intensity | High sensitivity | Samples become darker if proper dying agent is not provided | — | 133 |
2 | 442 nm | Fluorescent agent is employed to analyze the bio molecule distribution | Distribution of bio molecule can be analyzed easily | Applicable only for detecting non-melanoma | Se: 96.6% | 134 |
3 | 360–600 nm | Fluorescence spectra is used to classify the tumors at the wavelength of 436 nm and 440 nm | Short excitation source is enough | Less feasible | — | 135 |
4 | 630 nm | Suppression of cytokine production is recorded | Cytokine production can be analysed | Two common side effects of PDT are cutaneous photosensitization and systemic immunosuppression | Acc: 83% | 136 |
5 | 442 nm | A non-invasive diagnostic tool to identify diseased tissue and normal tissue from the absorption property of the chromophores | High resolution | Need high and optimum level of instrumentation setup | — | 137 |
6 | 410 nm | Laser-induced fluorescence spectroscopies are used to detect the skin cancer by in vivo methods without biopsy | Can able to differentiate the normal tissue from malignant tissue | Highly engaged to external noise that disturbs the measurement | — | 138 |
7 | 300 nm for epidermal tryptophan and 330 to 400 nm for dermal collagen crosslink | The fluorescence intensity of dermal collagen is generally lower in tumour than in the normal tissues | Abnormality in malignancy condition can be estimated easily | It requires endogenous fluorescence for diagnosis | — | 139 |
8 | 650–750 nm | The tumours can be detected by using a tunable light source along with CCD camera | Fluorophore can be retained between the normal and tumour cells | High cost | — | 140 |
9 | 355 nm and 440 nm | Two ultrafast lasers were used at 355 nm that excites the auto fluorescence | High resolution | External probe is required | — | 141 |
10 | 785 nm | From the images of the pigmented skin lesions, both the fluorescence and reflectance were studied | High sensitivity | Cost is high | — | 142 |
11 | 292 nm for tryptophan and 377 nm for collagen | The sampling points for Raman spectroscopy were chosen by auto fluorescence segmentation. It's function is to diagnose the BCC in sampled tissue during the Mohs micrographic surgery which is much faster than frozen section histopathology | Classification of skin lesions can be made easy | Segmentation is required to analyse whether the surgery is needed or not | Se: 100%, Sp: 92% | 143 |
12 | 400 and 2400 nm | Cancer tissue can be diagnosed by synchronous fluorescence (SF) imaging | Highly efficient | It is a weekly emitting component. | Se: 82–97% | 144 |
Fig. 11 Tetrahertz Spectroscopy. In this technique the properties of matters are probed with short pulses of THz radiations. In the above image (i) showing TPI handheld probe system, (A) main unit with computer monitor, handheld imaging probe and black umbilical cord (visible on the right), (B) close up of the handheld imaging probe, (C) close up of the head of the imaging probe showing the black quartz window. The probe scans an area of 15 × 2 mm, and acquires data from 26 pixels (red), reproduced under CC license from ref. 156; and (ii) showing the THz mechanism, reproduced under CC license from ref. 157. |
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 wavelength – limited 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 identifications, space exploration and THz cancer detections.155 Contemporary updates on THz spectroscopy approaches for skin cancer diagnosis are compiled in Table 7.
S. no. | Signal used | Description | Merits | Demerits | Ref. |
---|---|---|---|---|---|
1 | 0.1 to 2.5 THz | Dielectric properties of the human skin is analysed to differentiate the normal cell from abnormal cells | Acts as on effective tool to diagnose the skin neoplasm | Longer acquisition time | 147 |
2 | 14 THz | Monte Carlo simulation and Mueller matrices are used to diagnosis the skin cancer based upon the polarization property of the skin | Sensitivity is high | Injection of nano particles is required | 148 |
3 | 0.1–10 THz | Applied in the field of bio molecules of the cells and tissues | Cutting edge technology is addressed | It uses less lens array which results in less accuracy | 154 |
4 | 0.1–2.5 THz | Involves the detection of tumours for skin cancer diagnosis | Sensing application can be done easily | It requires less lens array | 155 |
5 | 0.3–4.3 THz | Used to image the neo plastic and the non-neo plastic colon tissue | Resolution is high | — | 149 |
6 | 0.1–1.1 THz | Plasmonic photoconductive antenna uses the T wave to diagnose the skin cancer in both electrical and optical methods | Plasmonic photoconductive antenna element are available in single component with optical frequency | Cost is high | 150 |
7 | 0.1–2.7 THz | Using the TPI technique the contrast between normal and tumour cells can be examined by time domain and frequency domain analysis | It requires only minimum impulse function | Consume more time | 146 |
8 | 76 MHz | High sensitive of THz detection of nanoparticles can be provided by plasmonic photoconductive antenna | High resolution | High cost | 151 |
9 | 4.2 THz | The biological tissue can be visualized by integrating the IR camera with the IR detector by exposing the human skin 4.2 THz | High resolution | Very expensive | 152 |
10 | 1.39–1.63 THz | The contrast between the normal and cancerous cells are demonstrated using THz imaging technique | Easily differentiate the cells | Consume more time | 153 |
Fig. 12 Optical coherence tomography. In the above image (a) showing a hand held OCT device, reproduced under CC license from ref. 176 and (b) showing the OCT mechanism. |
A report was made by Michael R. Hee, Joseph A. Izatt, Joseph M. Jacobson, James G. Fujimoto, and Eric A. Swanson, who made a significant finding on OCT by introducing a trans-illuminance 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 field 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 influence 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 inflammation 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 conflicts 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.
S. no. | Signal used | Description | Merits | Demerits | Acc, Se, Sp | Ref. |
---|---|---|---|---|---|---|
1 | 830 nm | With the help of fiber optic interferometer which releases a femtosecond pulses and coherent heterodyne detection to attain 130 dB dynamic range and time gating is performed to attain the image. Time gating is performed with the scattering medium | Monochromatic and coherent in nature | Cost is high | — | 161 |
2 | 830 nm | A compactible OCT system is presented which uses fiber interferometer with integrated longitudinal scanning of tissue. The dynamic range 60 dB allows observing structure of human skin in vivo up to 1.5 mm in depth | Non-contact method | Consumes more time to analyse the structure | — | 162 |
3 | 10 MHz | A slow-scan CCD-camera is able to detect very little numbers of photons. First measurements were made on models consisting of very small glass spheres embedded in polyester resin. It was possible to prove coherent photons from a depth of 2 mm | Accuracy is high in the obtained image | Consumes time since the measurement is performed with little number of photons so the field of view is less | — | 163 |
4 | 830 nm | Based on the principle of low-coherence interferometry, cross-sectional images of the human skin can be obtained in vivo with a high spatial resolution of about 15 μm | In vivo investigation of pharmacologic effects | Detection depth is much lower | — | 164 |
5 | 830 nm and 1285 nm | The effects of both instrumentation parameters and the dynamic characteristics of living tissue on image contrast and resolution and on speckle reduction are discussed. Both theoretical predictions and experimental results in human skin imaging show that longer wavelength can minimize the influence of multiple scattering on image contrast and resolution and thus increase the effective penetration depth of OCT | High resolution | Very expensive | — | 165 |
6 | 1300 nm | OCT and ODT system that uses phase information derived from a Hilbert transformation to image blood flow in human skin with fast scanning speed and high velocity sensitivity. This technique decouples spatial resolution and velocity sensitivity in low images and increases imaging speed by more than 2 orders of magnitude | High scanning speed | Cost is high | — | 166 |
7 | 1300 nm | A review was made on OCT method in a healthy skin, tumor region and inflammatory regions of skin and concluded that because of its non-invasive character, the technique allows monitoring of inflammatory diseases over time easily | Resolution is high | Data analysis takes times for investigating various skin disease conditions | — | 167 |
8 | 1310 nm | Here polarization-sensitive OCT images of human skin in vivo, demonstrating the ability of the technique to visualize and quantify the birefringent properties of skin | Used to determine the polarization properties of the skin in real time | Results show poor categorization | — | 168 |
9 | 1300 nm | Study demonstrates the clinical diagnostic potential of MPT/OCT for pre-screening relatively to obtain sub-cellular level information of the respective regions. | System design is simple and high speed image acquisition. | Cost is high | — | 169 |
10 | 1318 nm | Deep margins of skin tumors are greatly assessed using OCT. Inflammatory skin diseases can also be detected | Differentiates premalignant from malignant lesions | Resolution is low | Sp: 83–100%, Se: 79–86% | 170 |
11 | 1300 nm | The magnitude of the thermal coefficient of attenuation coefficient is greater in epidermis than in dermis. Thus it can be used to detect the optical attenuation using optical probe with the help of temperature control module | Can be used as a confirmation technique without physicians | Lack large scale clinical trials | — | 171 |
12 | 1305 nm | While considering OCT base on density of the skin for detecting BCC, normal skin has higher densities whereas the lesion cells have lower densities than the normal skin | Non-contact approach | It requires much time for imaging | Se: 79–94%, Sp: 85–96% | 172 |
13 | — | To determine the margins of BCCs with OCT, prior to MMS (Mohs micrographic surgery), to reduce the number of surgical steps. Scans of the center and entire margin were performed. If parts of the BCC were visible outside the margin, another 2 mm were added and the scan was repeated until the tissue outside the labelling looked tumor free | Stages of MMS were reduced by OCT method | Expensive set up | — | 173 |
14 | 1310 nm | To perform epithelialization of suction blister lesions by OCT and to find epidermal thickness (ET) easily as the primary outcome from histology test takes time. Results at earlier shows discrimination for neoepithelization but later shows better result | Fast speed scanning of image | This method not able to distinguish blood vessels from dilated lymphatic vessels. 20 experienced similar difficulties when evaluating OCT images of normal human skin | — | 174 |
15 | 1300 nm | Combining OCT and optoacoustic modalities to provide precise tumor depth determination also with Raman spectroscopy to determine the tumor depth, volumetric imaging. Results found penetration depth is high in optoacoustic method | High resolution | Consume more time | — | 175 |
Fig. 13 Multi-modal spectroscopy. In image (a) showing the setup of MMS and hand held probe and (b) showing exploded view with optical elements such as the filters and front lens identified along with the collection and delivery fibers for all three modalities. Reprinted from ref. 193 with the permission of AIP publishing. |
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 classifier. 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 software is used in multispectral imaging technique to analyze the skin chromophores which is a non-contact method.188 Multispectral imaging method finds 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.
S. no. | Signal used | Description | Merits | Demerits | Acc, Se, Sp | Ref. |
---|---|---|---|---|---|---|
1 | 430–950 nm | In this technique the skin tissue are characterized by texture, asymmetry, blotchiness and boarder | Easily differentiate the malignant and benign tumors | High cost | Se: 100%, Sp: 97% | 181 |
2 | 430–950 nm | The images of the moles are collected using charge coupled camera and the classifications of tumors are made by spatial gray co-occurrence matrix | Have high potential to classify the skin cancer | Texture information is needed | — | 182 |
3 | 483–950 nm | The skin lesions were imaged by in vivo imaging system operated at the wavelength between 483 nm and 959 nm at the equal wavelength intervals | High sensitivity | Classifier is required. | Se: 80.4% Sp: 75.6% | 183 |
4 | 483–951 nm | The pigmented lesions are categorized by automatic segmentation algorithm | Applicable only for non-melanoma skin cancer | Cost is high | Acc: 97.1% | 184 |
5 | 405–910 nm | In this method, the 2-dimensional and 3-dimensional images were analyzed at the wavelength ranges from visible to infrared spectrum | High resolution | Only for the particular wavelength results were optimum | — | 185 |
6 | 510–610 nm | In this method, the multispectral nevoscope trans illumination technique is used to compare the radiometric measurement of malignant and benign tumors | Differentiate the malignant and benign tumor | Vascular depth of the skin is hard to interpret | — | 186 |
7 | 450–950 nm | Using the multi spectral camera the images of the skin lesions were captured and analyzed based upon the spectral ranges | Novel image processing algorithm to differentiate of melanoma from pigmented nevi | High cost | — | 187 |
8 | 450–950 nm | The clinical trials were made on 266 pigmented lesions and 49 vascular lesions. The software system was used to mapping the skin chromophores | High clinical information from the image | Various algorithms are required for mapping process | — | 188 |
9 | 450–950 nm | The melanoma images were analyzed based upon the melanin index and erythema index | High resolution images are obtained | Cost of instrumentation set up is high | — | 189 |
10 | 400–720 nm and 650–1100 nm | MSDLA device is used to determine whether the biopsy is required or not | High resolution | High cost | Se: 94% | 190 |
11 | 400–720 nm and 650–1100 nm | The six layered skin model was made to analyze the optical parameters | Accuracy level is high. | Doesn't classify the melanoma and non-melanoma skin cancer | — | 191 |
12 | 414–995 nm | The skin cancer can be detected based on the spectral parameters which involve the reflectance and the color of skin lesions | High quality | High cost | Se: 93%, Sp: 54% | 192 |
Fig. 14 (a) The infrared imaging system, (b) photograph of the body surface area with a cluster of pigmented lesions, (c) reference infrared image of the region at ambient temperature, (d) the same area after cooling and (e) magnified section of the melanoma lesion and surroundings. Reproduced under CC license from ref. 19. |
A study performed by Cila Herman198 found a significant finding 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 after 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 significant 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.200,201 Prateek Mathur et al. studied on temperature variation of thermal images obtained from skin cancer patient and introduced a device named sKan as shown in Fig. 15. It is a new technique which maps the affected area with the help of sensor and monitors the patient's stage in the skin cancer.202,203 This is achieved by having a template of temperature values on which the obtained value is mapped and the data are sent to the doctor for the analysis.201 Viktor Gruev et al. made a report on an imaging sensor capable of recording the optical properties of partially polarized light by monolithically integrating aluminium nanowire optical filters with a CCD imaging array. The imaging sensor consists of 1000 by 1000 imaging elements with 7.4 μm pixel pitch. The polarization imaging sensor captures intensity, angle, and degree of linear polarization in the visible spectrum.204 Current updates on thermography techniques for skin cancer diagnosis are presented in Table 10.
Fig. 15 sKan device. This device detects skin cancer using thermal images that are obtained from affected areas of the skin surface using thermistors. Reproduced from ref. 202. |
S. no | Device used | Description | Merits | Demerits | Acc, Se, Sp | Ref. |
---|---|---|---|---|---|---|
1 | Image sensing device 340–380 nm | Using image sensing device illumination of ultraviolet light in the wavelength range of 340–380 nm, fluorescence property of the different skin anomalies were detected | Real time recording can be done and non-invasive | Highly expensive | — | 205 |
2 | Infrared camera | Infrared imaging takes place in different region of the tissues, which is caused due to micro calcification or calcification in the benign lesion | High resolution | Skin temperature varies for external noise that affects the result | — | 206 |
3 | Infrared thermometer | Thermal sensor that detect the affected area in a thermal map and calibrated using the processor (30) and corresponding output signals taken over a predetermined range of ambient temperatures | High resolution | Cost is high | — | 207 |
4 | Infrared detectors | Temperature distribution of the body surface is performed using infrared thermography which produces thermogram that further used for diagnosing procedure | High resolution | Emission from other surfaces affects the measurement accuracy | — | 208 |
5 | Amber radiance 1 T IR camera | Differentiation of melanoma and benign cutaneous lesions is performed using infrared thermography and found the sensitivity and specificity for different depths of the lesion | Real time recording and non-invasive | Need optimal instrumentation setup | Se: 39% Sp: 100% (0–5 mm lesions), Se: 58% Sp: 98% (lesions >5–15 mm), Se: 95% Sp: 100% (lesions >15–30 mm), Se: 78% Sp: 89% (lesions >30 mm) | 209 |
6 | Infrared thermal camera operated at 3 μm −5 μm | Thermal responses for healthy and malignant tissue were compared and concluded that increased metabolic activity of melanoma lesion can be detected using infrared imaging | High resolution | Camera has only 2% of accuracy | — | 19 |
7 | Infrared thermal camera operated at 14000 nm | Symptoms of illness are detected using embedded devices which can be in the form of watch, glass, camera etc. Communication interface is attached to the embedded device so that the captured data is transferred into the local server of the hospital or other healthcare station | Automatically the captured data's are transmitted. | Cost is high. | — | 203 |
8 | Infrared camera | Thermal imaging for early stage detection of cutaneous melanoma is performed and resulting surface temperature oscillations were recorded using infrared camera | Real time recording | Lower rate transmission | — | 210 |
9 | Electrochemical sensor, optical sensor | Biosensors are used to detect specific biological analyze later by converting a biological entity into an electrical signal and analyzed. Biosensor has vast potential in detecting the cancer cells | Have more resolution comparing to other application sensor | Specific chemical analyzer is random in nature so difficult to interpret | — | 201 |
10 | Nd:YAG laser at 532 nm | The imaging sensor is use to capture the affected area which has 7.4 μm pixel pitch, composed of 1000 by 1000 imaging elements with a nanowire optical filter | The signal to noise ratio is high and consume less power | Cost is high since nanowire fabrication is implemented | — | 204 |
11 | Anon contact IR temperature sensor at 8–14 μm | The emitted IR range extends from 800 nm to few hundred micrometers which are detected for the temperature variation | Mapping of temperature variation is perfect | Resolution is less | — | 211 |
12 | Infrared detectors | The surface the sample is sensed with the small change in the motion. This motion is detected by bending of cantilever in the sensor system. Such system provides information about the position of the subject and the temperature value during displacement | Nanoparticles are synthesized in the sensor which makes the accuracy to be good | Cost is high | — | 212 |
13 | IR absorbance sensor at 3300 nm and 3570 nm | Infrared absorbance sensor used to detect the malignant cells in the biopsy. It takes 8 seconds to identify whether it is tumor or normal cell | This sensor is used in diagnosing the melanoma during histopathological analysis and also decreases the wrong interpretation | Supports shorter range and its performance degrade with longer distances | — | 213 |
Electrical impedance tomography (EIT)216 is being used for various biomedical applications. A report given by Beetner D. G. et al. shows the difference in electrical impedance with the frequency range of 1 KHz to 1 MHz for the lesion which has a diameter of 2–15 mm.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 bio-impedance for different skin lesions.220,221 The non-invasive probe separates the nevi from BCC with 96% sensitivity and 86% specificity.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 fluids 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 non-cancerous cell at the range of 1 KHz to 2.5 MHz.225 Bio-impedance 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.
S. no. | Signal used | Description | Merits | Demerits | Acc, Se, Sp | Ref. |
---|---|---|---|---|---|---|
1 | 1 kHz to 1 MHz | Used to distinguish BCC from benign lesions based on magnitude, phase, real part and imaginary part of the impedance calculated | It provides the rapid differentiation of tumors | High cost | Se: 90%, Sp: 50% | 217 |
2 | 2 MHz with a sampling rate of 50 Hz | To determine the state of accuracy to distinguish benign from malignant lesions by measuring with electrical impedance | Reflects morphological changes when there is a growth in tumor cells | This method is not applicable for humans | Se: 92%, Sp: 67% | 219 |
3 | 1 kHz and 1000 kHz | Distinguishes the skin cancer from the benign lesions using multi-frequency impedance spectra | The result obtained is better accurate than conventional methods | Distinguishing the tumors takes more time and false result may also be obtained | Se: 75–87%, Sp: 100% | 220 |
4 | 1–1000 kHz | Compares the detection of skin cancer by non-invasive probe and micro invasive electrode system, whose surface is furnished with tiny spikes which gets penetrated to the stratum corneum | Electrode system produces better result | Minimally invasive technique | Se: 92–96%, Sp: 80–86% | 222 |
5 | 1 kHz and 1 MHz | Describes the method for detecting the skin cancer using electric impedance. Electric impedance of the biological system decreases with the increase in frequency | High resolution | Multivariate and the impedance is complex | — | 218 |
6 | 1 kHz to 2.5 MHz | Accuracy of electrical impedance to classify malignant melanoma from benign tumour by automated classification algorithm | Accuracy is high | Various algorithm is needed for classification of skin cancer | Se: 95%, Sp: 49% | 221 |
7 | 1–100 kHz | Non-invasive approach for detecting the presence of skin lesions by measuring the impedance change | Low cost and portable | Electrodes are used which cause discomforts | — | 18 |
8 | 1 kHz to 2.5 MHz | EIS algorithm is used on lesions to differentiate normal skin from the abnormal lesions | High resolution | Experienced physician is required | Se: 100% (non-melanoma cancer) | 224 |
9 | 1 kHz to 2.5 MHz | 1300 lesions were collected from multi centre and examined by EIS at 1 kHz to 2.5 MHz in order to differentiate the melanoma from non-melanoma cancer | Classification is easy | High cost | Se: 98.1% (melanoma), Se: 100% (non-melanoma) | 225 |
10 | 20 kHz to 1 MHz | Portable bio-impedance system is used to diagnose the skin cancer based upon the magnitude ratio and phase detection method. | Act as a great tool for monitoring the physiological conditions of the biological system | High cost | — | 226 |
Fig. 16 Electrical bio-impedance measurement setup. In the above image shown (i) Nevisense device including hand piece and touch screen monitor. (Courtesy of SciBase, Stockholm, Sweden; with permission); (ii) Electrical impedance mammograph meik, reproduced under CC license from ref. 227; (iii) Smart phone based diagnosis using EIT patch. |
Fig. 17 Tape stripping technique. Method of tape stripping (a) application of the formulation on a marked skin area; (b) homogeneous distribution; (c) the adhesive tape is pressed with a roller on the skin, an empty sheet of paper avoids the transfer of the formulation onto the back side of the tape; (d) removal of the tape. Reprinted from ref. 228, with permission from Elsevier. |
S. no | Description | Merits | Demerits | Acc, Se, Sp | Ref. |
---|---|---|---|---|---|
1 | mRNA from the cells attached with tapes are collected for gene investigation towards the differentiation of melanoma and non-melanoma | Rapid and easy to perform | Test results need large genetic profile for differentiation process | Se: 69%, Sp: 75% | 90 |
2 | Cells from the stratum corneum is taken by means of tape stripping and detects the melanoma from nevi | Less resource consumption procedure | Produce itching effect after the procedure | Se: 100%, Sp: 88% | 229 |
3 | Different non-invasive techniques involved in detecting melanoma in which tape stripping act as excellent clinical information provider for melanoma detection using mRNA profiles | Easy procedure to perform | Investigation needs large set gene profile datasets | — | 230 |
4 | The best condition for tape stripping procedure was validated through the evaluation of the distribution of corneocytes, mass of stratum corneum (SC) removed and amount of protein removed using finger pressure, a 2 kg weight and a roller | Cost is less | Not used clinically due to conflicts in testing results | — | 232 |
5 | RNA from the stratum corneum is isolated to give 312 gene expressions for the detection of melanoma | Rapid and convenient method and patient friendly | Lack of standardized protocol and heterogeneous sampling | Se: 100%, Sp: 88% | 233 |
6 | Differentiation between lentigo maligna and lentigo maligna melanoma from solar lentigo, which is sometimes difficult with histology is easily performed by the extracted genetic information | Sensitive and specific for melanoma and can be able to detect genomic changes before morphological changes appear | Only few studies are available to prove the accuracy, sensitivity and specificity features of this technique | Se: 100%, Sp: 88% | 234 |
7 | RNA isolation, RNA amplification and array hybridization were performed to differentiate melanoma from dysplastic nevi | Accurate results are obtained at the preliminary stage of the tests | Need extra 15 to 20 melanoma sample for robust operation which causes local irritation to the patient | Acc: 100% | 235 |
8 | RNA extraction, synthesis of cDNA, probe labelling, array hybridization, quantitation of hybridization signals, reverse transcription polymerase chain reaction (RT-PCR), confirmation of markers were performed to differentiate the types of skin cancer | Earlier diagnosis of melanoma is performed | Can be performed only when the extracted particles are frozen deeply | — | 236 |
9 | One hundred and fifty pigmented lesions were tested using tape stripping toluidine blue (TSTB) method in the diagnosis of malignant melanoma and found the sensitivity and specificity | Early detection of malignant melanoma is performed in addition with ABCDE rule | This investigation takes time to collect data | Se: 68.7%, Sp: 74.5% | 237 |
10 | Preliminary investigation was designed to compare the gene expression profiles of melanoma and benign nevi | Epidermal genetic information retrieval (EGIR) has the potential to identify future biomarkers which would be useful in targeted therapy | Only few studies are available to prove the accuracy level | Se: 100%, Sp: 88% | 238 |
Fig. 18 Multiphoton scanning technique. In the above image shown setup used two-photon excitation that allows imaging of living tissue up to one millimetre in depth. |
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 find 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 fluorescence 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.
S. no | Signal used | Description | Merits | Demerits | Acc | Ref. |
---|---|---|---|---|---|---|
1 | 730 nm and 960 nm | Multiphoton excitation microscopy was used to image in vivo human skin auto fluorescence from the surface to a depth of −200 μm | Deepest layer of the tissue is observed and displayed in a 3D image | Photo damage occurs | — | 242 |
2 | 80 MHz at 780 nm | Compared the tandem scanning reflected light confocal microscopy and multi-photon excitation microscopy for the observation of human skin in vivo. Both techniques gave similar results from the skin surface to the epidermal–dermal junction | Excitation can only occur in the focal plane, so out-of-focus bleaching is avoided | Minimal photo damage occurs | — | 243 |
3 | 810–850 nm | Presented that the multiphoton laser-scanning microscope can provide high three dimensional resolution of gene expression and function in deeper regions of tumors | Lower photo toxicity due to the longer wavelengths | Cell damage occur due to three-photon excitation | — | 244 |
4 | 780 nm | Intravital epifluorescence microscopy has provided powerful insight into gene expression, tissue pH, tissue pO2, angiogenesis, blood vessel permeability, leukocyte–endothelial interaction, molecular diffusion, convection and binding | 1. Depth of imaging is increased | Cost is high | — | 245 |
2. Projection of three-dimensional structures onto a two-dimensional plane | ||||||
3. Photo toxicity is reduced | ||||||
5 | 820 nm | Major features of aged skin include dryness, flaccidness, wrinkling, multiple, mostly benign neoplasms, signs of chronic UV damage, carcinogenesis, and functional deficiency. Based on collagen and elastin content in the skin, skin aging is defined | Second harmonic finding correlated with the histological findings in an ex vivo manner | Movement artefacts occur which affects the resulting image | — | 246 |
6 | 750 nm and 850 nm | Pathological and physiological characterization conditions by MPLSM in vivo are performed to investigate and identify skin morphological alterations related to ageing | Useful for diagnostics of pathological conditions and their differentiation from ageing effects | Consumes time for cellular differentiation | — | 247 |
7 | Femtosecond pulsed laser at 780 nm | Describes about the morphologic features of human non-melanoma skin cancer obtained using MPLSM on freshly excised specimens from 14 patients. Emission from the skin was detected in the range of 450–530 nm, corresponding to the auto fluorescence of nicotinamide adenine dinucleotide, NADPH, keratin, elastin, collagen, and melanin, allowing us to visualize the cellular structures within the epidermis and upper dermis of the specimens | Deeper biological tissue can be viewed | Cost is high | — | 248 |
8 | Mode lock at 80 MHz, and tuning range at 750–850 nm | Laser radiation in the near infrared spectrum was used to image endogenous fluorophores by multiphoton excitation. Eighty-three melanocytic skin lesions have been investigated. The results showed distinct morphological differences in melanoma compared with melanocytic nevi | High accuracy | Cost is high | Acc: 97% | 249 |
9 | Tunable range at 750–850 nm and excited at 760 nm | Comparison of confocal microscopy and multiphoton tomography were performed. Both methods, CLSM and MPT, were found to be suitable for in vivo imaging of superficial skin layers | Both methods have high resolution and near-real time acquisition speed of the image | It was not possible to use same optical medium for all the methods | — | 250 |
10 | 532 nm | Multimodal imaging, the combination of two-photon excited fluorescence (TPEF), and second harmonic generation (SHG), which have high diagnostic potential. Multimodal imaging generates molecular contrast, but to use this technique in clinical practice, the optical signals must be translated into diagnostic relevant information | Increased reliability | Cost is high | — | 251 |
11 | Femtosecond pulsed laser at 780 nm | Demonstrated MPLSM as a diagnostic tool for superficial skin cancers, such as BCC, SCC in situ, and skin. The morphologic features observed in MPLSM images of skin tumors are comparable to traditional histopathology | Imaging depth is increased here | Loss of signal occur during focusing | — | 252 |
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 specificity 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 superficial 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 fields of view. Many other modalities like high-frequency ultrasound sonography, PDD, Raman spectroscopy, fluorescence 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.
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.
Acc | Accuracy |
BCC | Basal cell skin cancer |
CNN | Convolutional neural network |
MSOT | Multi-spectral optoacoustic tomography |
MPT | Multi photon tomography |
OCT | Optical coherence tomography |
ODT | Optical doppler tomography |
PDT | Photodynamic therapy |
PAI | Photoacoustic imaging |
PDD | Photodynamic diagnosis |
RCM | Reflectance confocal microscopy |
Res | Resolution |
Se | Sensitivity |
Sp | Specificity |
SCC | Squamous cell carcinoma |
THz | Tetrahertz |
TBP | Total body photography |
TPI | Tetrahertz pulsed imaging |
This journal is © The Royal Society of Chemistry 2018 |