While optical spectroscopy may play an important role in many aspects of clinical cancer management such as cancer screening in the breast[16
], oral cavity[20
], and treatment planning and monitoring in the brain[21
] and the prostate[23
], emphasis will be placed on cancer diagnosis due to the following reasons: First, cancer diagnosis appears to be the most frequently cited goal in relevant publications. Second, the optical techniques developed for cancer diagnosis are in general also applicable to cancer screening and treatment planning as well as outcome evaluation.
In this literature review, the preference will be given to in vivo clinical studies. Ex vivo or in vitro studies are sometimes mentioned to highlight specific applications. No preclinical studies will be included. The review will be organized in terms of organ sites, which cover the brain, breast, cervix, lung, stomach, colon, prostate and the skin. For each organ site, a brief description of the state of the art is presented, followed by representative studies.
Diffuse reflectance spectroscopy in the near infra-red (NIR) spectrum, sometimes named NIR spectroscopy, has attracted much attention in brain imaging since the 1990s, because of the relatively large penetration depth of NIR light. It was found feasible to detect brain activities through intact skulls[25
] especially in children. The major physiological parameters extracted from NIR spectroscopy include total hemoglobin concentration and tissue oxygenation. Because of weak scattering in this range, light transport in this region can be described by an analytical expression, i.e. diffuse theory, which greatly facilitates data analysis. This technique is frequently used in an imaging modality, i.e. diffuse optical tomography (DOT). Multiple detectors are used to reconstruct three-dimensional images of cerebral hemodynamics. For a review of DOT imaging in the brain, please refer to Jacobs et al[26
], Gibson et al[27
], and Huppert et al[28
]. However, because most brain cancers in an intact skull are not within the reach of light, this technique is less studied in clinical oncology.
Diffuse reflectance spectroscopy in the visible spectrum and fluorescence spectroscopy have been studied in tumor margin assessment or tissue characterization during neurosurgery procedures. Compared to NIR spectroscopy, diffuse reflectance spectroscopy in the visible spectrum and fluorescence spectroscopy possess higher chemical sensitivity. The problem of small penetration depth associated with ultraviolet and visible light is no longer an issue during surgery, which opens a new venue to apply optical spectroscopy in brain cancer management. For a review on the neuro-oncological applications of optical spectroscopy, please refer to Toms et al[29
]. A few representative studies were briefly reviewed below.
Lin et al[30
] investigated the applicability of combined autofluorescence and diffuse-reflectance spectroscopy for intraoperative detection of infiltrating tumor margins (ITM) in a pilot in vivo
clinical trial consisting of 26 brain tumor patients. A two-step empirical discrimination algorithm yielded a sensitivity and specificity of 100% and 76%, respectively, in differentiating ITM from normal brain tissues.
Antonsson et al[31
] investigated the use of diffuse reflectance spectroscopy for differentiating tissue types to improve intracerebral guidance during deep brain stimulation. Diffuse reflectance spectroscopy measurements in 10 patients were recorded for three different functional targets including the subthalamic nucleus (STN), internal globus pallidus (GPi) and zona incerta (Zi). Significant intensity differences between white and gray matter were found to be at least 14% (P
< 0.05) and 20% (P
< 0.0001) for MRI and spectral-sorted data, respectively.
Lin et al[32
] further investigated the feasibility of using diffuse reflectance and fluorescence spectroscopy to differentiate pediatric neoplastic and epileptogenic brain from normal brain in an in vitro
study. Statistically significant differences (P
< 0.01) were found between (1) neoplastic brain and normal gray matter; (2) epileptogenic brain and normal gray matter; and (3) neoplastic brain and normal white matter.
Krafft et al[33
] explored the use of Raman spectroscopic mapping for distinguishing between normal brain tissue and gliomas and meningiomas. Ex vivo
tissues were examined by a Raman spectrometer with 785 nm excitation coupled to a microscope. Normal brain tissue was found to contain higher levels of lipids, intracranial tumors have more hemoglobin and lower lipid to protein ratios, meningiomas contain more collagen with maximum collagen content in normal meninges.
Breast cancer is perhaps the most extensively studied cancer in the community of biomedical optics. Optical spectroscopy techniques have been explored in a variety of forms in breast cancer diagnosis, including non-invasive breast cancer imaging, tumor characterization for margin assessment during breast surgery and “optical biopsy” measurements in needle biopsy or fine needle aspiration procedures.
Non-invasive breast cancer imaging can be performed in diffuse optical tomography[34
] or diffuse reflectance spectroscopy often in the frequency domain[6
]. Multiple wavelengths are used to achieve spectroscopic measurements and provide functional images of the breast, which include hemoglobin concentration, oxygen saturation, water and lipid content as well as scattering properties. Although this approach sounds attractive, it suffers from low spatial resolution due to multiple light scattering. To tackle this problem, other imaging modalities such as CT and MRI were proposed to provide anatomical images at a high spatial resolution[35
], which is incorporated into the DOT reconstruction algorithm to combine with the functional images obtained in DOT. Because of the advantage in the signal to noise ratio, optical equipment in the frequency domain is preferred in many cases. For a more detailed review of this technique, please refer to publications[35
]. Shah et al[36
] and Tromberg et al[16
] have provided excellent discussion on the potential roles of optical spectroscopy in the clinical management of breast cancer.
Tumor margin assessment during open surgery using optical spectroscopy has been reported by several groups[41
] as summarized in Table . It is worth noting that the accuracy changes with the classification algorithm[41
] which suggests the importance of selecting appropriate data analysis methods. There is a special case[42
] in which Raman spectroscopy detected a grossly invisible cancer that was confirmed by pathologic review. This finding gave the patient a chance for a second surgical procedure to prevent the recurrence of cancer.
Summary of techniques, patient population and accuracy in previous optical spectroscopy studies for intraoperative breast margin assessment
All these previous studies demonstrated that optical spectroscopy could be used in a real-time fashion to guide tissue excision during breast surgery, potentially to reduce the need for repeated surgery resulting from positive margins, and thereby reducing the recurrence rate of breast cancer following mastectomy surgery.
Needle biopsy has become another popular carrier of fiber-optic probes for performing optical spectroscopy, which is technically called “optical biopsy”. The advantages of “optical biopsy” compared to physical biopsy are not only non-invasiveness and high accuracy but also an increased sensing tissue volume, which could greatly reduce the chance of missing hidden cancer sites. Manoharan et al[45
] performed an early study to picture the possibility of incorporating Raman spectroscopy into a biopsy needle for breast cancer examination. Bigio et al[41
] performed transdermal-needle measurement using elastic scattering spectroscopy for instant diagnosis with minimal invasiveness for breast tissue examination. The accuracy of transdermal-needle measurements combined with spectral measurements in open surgery is reported in Table .
van Veen et al[46
] performed differential path-length spectroscopy (DPS), which is essentially a type of diffuse reflectance spectroscopy, on healthy and malignant breast tissue using a fiber-optic needle probe. A special tissue model was used to yield information on the local tissue blood content, the local blood oxygenation, the average micro-vessel diameter, the beta-carotene concentration and the scatter slope. The histological outcome of core-needle biopsies taken from the same location was used as the gold standard. Malignant breast tissue has a smaller tissue oxygenation and a higher blood content compared to normal breast tissue.
Yu et al[47
] developed a side-firing fiber-optic sensor based on near-infrared spectroscopy for guiding core needle biopsy diagnosis of breast cancer. The sensor is inserted into a core biopsy needle to measure diffuse reflectance spectra in the NIR spectrum at the biopsy site through an aperture on the needle before the tissue is removed for histology. Preliminary in vivo
measurements were performed on 10 normal or benign breast tissues from 5 women undergoing stereo- or ultrasound-guided core needle biopsy and showed good correlation with histopathology.
Zhu et al[48
] explored the use of fluorescence spectroscopy for guiding breast biopsies. A total of 121 biopsy samples with accompanying histological diagnosis were obtained clinically. The statistical data analysis provided a cross-validated sensitivity and specificity of up to 81% and 87%, respectively, for discrimination between malignant and fibrous/benign samples, and up to 81% and 81%, respectively, for discriminating between malignant and adipose samples. The corresponding receiver operator curves (ROC) yielded an area under the curve (AUC) of 0.87 and 0.84 in two cases. It is noted that ROC is a graph of sensitivity against (1-specificity) and the AUC is an indicator of the diagnostic performance.
In these applications, quantitative methods have been developed in all optical spectroscopy techniques, which provide extra information to elucidate the biochemical basis of carcinogenesis in the breast in addition to its use for diagnosis. For instance, diffuse reflectance spectroscopy has been used to derive vascular oxygenation and total hemoglobin content in breast cancer[5
]. Raman spectroscopy was used to derive information on cholesterol-like lipid deposits, fat, collagen, and cell nucleus/cytoplasm[42
Because of relatively easy access to the cervix, numerous studies[14
] have been reported on the in vivo
diagnostics of cervical neoplasia. Cardenas-Turanzas et al[52
] presented an excellent review on the clinical effectiveness of diffuse reflectance and fluorescence spectroscopy for the in vivo
diagnosis of cervical intraepithelial neoplasia. According to this review, optical spectroscopy showed a similar performance to colposcopy and can be an effective adjunct to colposcopy to help localize lesions. It also has potential use in cervical screening or to triage patients on Pap smear.
The following review papers may also be helpful. Murali Krishna et al[51
] provided a brief overview on the optical spectroscopic approach to cervical cancer diagnosis as well as on radiation therapy and radiation resistance. Bazant-Hegemark et al[53
] reviewed several tools capable of non-destructive mapping of the cervix at high resolution in a clinical environment including infrared spectroscopy and Raman spectroscopy in terms of clinical performance for diagnosis. Drezek et al[54
] presented an overview of various optical techniques including optical spectroscopy for the detection of precancerous lesions in the uterine cervix presented at the Second International Conference on Cervical Cancer. This review strong recommends the use of the Littenberg method for assessing new techniques to ensure that better technologies will stand out.
Several studies reported on the quantification of physiological parameters based on measured optical spectra, which provided insight into the development of cervical dysplasia at various stages. Chang et al[55
] used an analytical model to estimate the contributions of several optical biomarkers by analyzing spectra from diffuse reflectance spectroscopy and fluorescence spectroscopy measurements. The model was applied to 493 in vivo
fluorescence measurements of cervical tissue acquired from 292 patients. The results show an increase in epithelial flavin adenine dinucleotide (FAD) fluorescence, a decrease in epithelial keratin fluorescence, an increase in epithelial light scattering, a decrease in stromal collagen fluorescence, and an increase in stromal hemoglobin light absorption in dysplastic tissue compared to normal tissue. These changes likely reflect an increase in the metabolic activity and loss of differentiation of epithelial dysplastic cells, and stromal angiogenesis associated with dysplasia.
Chang et al[56
] assessed the capability of a diffuse reflectance spectroscopy technique to identify contrasts in optical biomarkers at different grades of cervical intraepithelial neoplasia (CIN) using a numerical model. In a total of 89 sites examined in 38 patients, there were 46 squamous normal sites, 18 CIN 1 sites, and 15 CIN 2(+) sites. Total hemoglobin was statistically higher in CIN 2(+) compared with normal and CIN 1 sites, which was attributed to stromal angiogenesis. Scattering was significantly reduced in CIN 1 and CIN 2(+) compared with normal sites, which was attributed to breakdown of the collagen network in the cervical stroma.
Typical optical spectroscopy equipment can be readily used for characterization of tissue biopsies from the lung[57
]. Huang et al[59
] used a near-infrared (NIR) Raman spectroscopy system at 785 nm excitation to measure bronchial tissue specimens including 12 normal specimens, 10 squamous cell carcinoma (SCC) and 6 adenocarcinoma specimens obtained from 10 patients. They demonstrated that Raman spectra differed significantly between normal and malignant tumor tissue, with tumors showing higher percentage signals for nucleic acid, tryptophan and phenylalanine and lower percentage signals for phospholipids, proline and valine, compared to normal tissue.
Yamazaki et al[58
] constructed a near-infrared multichannel Raman system with an excitation wavelength at 1064 nm. They collected a total of 210 Raman spectra. The resulting sensitivity of cancer prediction was up to 91% and the specificity was 97% with an error margin of P
< 0.0001 according to Fisher’s exact test.
Aerts et al[57
] successfully related HIF1a, which is one of the hypoxia-related proteins, to in vivo
spectroscopic measurements of tumor blood saturation performed during bronchoscopy in 17 tissue samples. There was a significant difference in the spectroscopically determined saturations between the biopsies with negative expression of HIF1a and the biopsies with positive expression of HIF1a (P
Optical spectroscopy can also be incorporated into a commercial endoscopy system to perform in vivo
examinations of the lung. Zeng et al[60
] developed an integrated endoscopy system for simultaneous imaging and spectroscopy to detect early lung cancers. Zeng et al[61
] proposed to use autofluorescence imaging and white light reflectance imaging to obtain high diagnostic sensitivity, while at the same time using non-contact point reflectance/fluorescence spectroscopy to reduce false positive biopsies. A pilot clinical study involving 22 lung patients demonstrated that the malignant lung lesions can be differentiated from the benign lesions using this system with a sensitivity and specificity of more than 80%.
Numerous studies have reported on the examination of ex vivo
stomach tissue samples using optical spectroscopy. Kawabata et al[62
] performed Raman spectroscopy measurements on 251 fresh biopsy specimens obtained from 49 gastric cancer patients. Fresh specimens were measured with an excitation wavelength of 1064 nm. A sensitivity of 66%, a specificity of 73%, and an overall accuracy of 70% were achieved for the differentiation of gastric carcinoma from normal mucosa. Teh et al[63
] applied near-infrared (NIR) Raman spectroscopy at 785-nm excitation in a total of 73 gastric tissue samples (55 normal, 18 cancer) from 53 patients. The predictive sensitivity and specificity of the independent validation dataset were 88.9% and 92.9%, respectively, for separating cancer from normal samples.
Given the high chemical specificity of Raman spectroscopy, it can be used to find the source of Raman signals contributing to cancer diagnosis. Teh et al[64
] measured Raman spectra of 88 gastric tissue samples from 56 patients. Significant differences in Raman spectra were observed among normal, Helicobacter pylori
-infection) and intestinal metaplasia (IM) gastric tissue, which were attributed to proteins, lipids and porphyrin. Data analysis yielded diagnostic sensitivities of 91.7%, 80.0%, and 80.0%, and specificities of 80.0%, 100%, and 92.7%, respectively, for the classification of normal, Hp-infection and IM gastric tissues. Raman spectroscopy has also been used in the early diagnosis and typing of intestinal and diffuse adenocarcinoma of the stomach[65
], in which predictive accuracies of 88%, 92% and 94% were achieved for normal stomach, and intestinal- and diffuse-type gastric adenocarcinomas, respectively.
Optical spectroscopy has been incorporated into endoscopy systems for in vivo
measurements of stomach cancer. Mayinger et al[66
] evaluated light-induced autofluorescence spectroscopy in a commercial endoscopy system for the in vivo
diagnosis of gastric cancer. A total of 15 patients with pure adenocarcinoma and 16 patients with gastric cancer containing signet-ring cells were recruited into the study. A sensitivity of 84% and a specificity of 87% were obtained for the diagnosis of pure adenocarcinoma of the stomach. The diagnostic performance was found to decrease with increasing numbers of signet-ring cells and tumor grade.
Some optical spectroscopy techniques such as diffuse reflectance spectroscopy have been incorporated into colonoscopy[67
] or flexible sigmoidoscopy[68
] to perform in vivo
measurements. Dhar et al[69
] assessed the diagnostic potential of elastic scattering spectroscopy, which is essentially a particular type of diffuse reflectance spectroscopy, in colonoscopy to differentiate abnormal colon tissues in vivo
. A total of 483 spectra (290 normal, 19 hyperplastic, 69 adenomatous polyps, 74 chronic colitis, and 31 colorectal cancer) were obtained from 138 sites in 45 patients at colonoscopy. The sensitivity and specificity of differentiating adenomas from hyperplastic polyps, cancer from adenomatous polyps, colitis from normal tissue, dysplastic mucosa (from polyps) from colitis were 84% and 84%, 80% and 75%, 77% and 82%, 85% and 88%, respectively.
Zonios et al[70
] first analyzed diffuse reflectance spectra collected from adenomatous colon polyps and normal colonic mucosa of patients undergoing colonoscopy to estimate the following four parameters: hemoglobin concentration, hemoglobin oxygen saturation, effective scatterer density, and effective scatterer size. It was observed that normal and adenomatous tissue sites exhibited differences in hemoglobin concentration and effective scatterer size, which were in agreement with other studies that employed standard methods. A similar method was used by Wang et al[67
] to quantify total hemoglobin concentration (THC) and oxygen saturation (StO2
) in vivo
in 27 patients with colorectal cancer (CRC). Increased hemoglobin concentration and decreased oxygenation were observed from normal sites to premalignant tissues and then to malignant tissues.
Roy et al[68
] incorporated polarization-gated spectroscopy into flexible sigmoidoscopy to detect an early increase in blood supply (EIBS) in the endoscopically normal rectum (n
= 366). The rectal mucosal oxyhemoglobin content in females with advanced proximal neoplasia (n
= 10) was significantly higher than that in the control group. It is worth noting that the addition of rectal oxyhemoglobin information dramatically increased the sensitivity to advanced neoplasia compared to flexible sigmoidoscopy alone. The sensitivity and specificity were 100% and 76.8%, respectively.
The research in Raman spectroscopy has been mostly limited to ex vivo
studies likely due to the technical difficulty in obtaining Raman spectra at a reasonable signal to noise ratio. Stone et al[71
] carried out Raman measurements for optical diagnostics in various organs including colon using an optimized commercial Raman microspectrometer. Both the sensitivity and the specificity were greater than 90% for all tissues. Widjaja et al[72
] combined near-infrared (NIR) Raman spectroscopy at 785 nm excitation with support vector machines (SVM) for the classification of different histopathological groups in colon tissues. A total of 105 colonic tissue specimens from 59 patients including 41 normal, 18 hyperplastic polyps and 46 adenocarcinomas were included in this study. The results showed that the maximum overall diagnostic accuracy ranged from 98.4% to 99.9%. Beljebbar et al[73
] used near-infrared Raman microspectroscopic imaging to investigate the changes in composition from normal colonic tissues to adenocarcinoma ex vivo
. Multivariate statistical analysis was applied to the Raman spectra to identify the molecular composition and distribution of lipids, proteins, mucus and collagens in normal and malignant tissue. The results matched those of conventional histopathological examination.
Blood plasma has been also used as samples for optical measurements. Fluorescence spectroscopy in blood has been exploited by Lualdi et al[74
] to diagnose colorectal cancer. The study involved 341 subjects including 169 normal blood donors. Plasma fluorescence spectrum was measured in all subjects. The fluorescence emission peak around 615-635 nm was assigned to endogenous porphyrin. The peak intensity was significantly different between patients bearing colorectal cancer and normal blood donors. The ROC analysis resulted in an area under the curve of 0.72, close to that reported for the carcinoembryonic antigen (CEA) test, which suggests that this method could be a cost effective alternative screening test to CEA.
Dekker et al[75
] and Wallace et al[76
] reviewed recent advances in colonic imaging, which included various optical imaging techniques. Many of these optical imaging techniques, such as fluorescence imaging and narrow-band imaging, are essentially the extension of optical spectroscopy techniques, in which only the data at a few discrete wavelengths are acquired at many pixels. Moglia et al[77
] reviewed another exciting in vivo
imaging technique, capsule endoscopy, which enables remote diagnostic inspection of the gastrointestinal tract without sedation and with minimal discomfort.
Various optical spectroscopy techniques have been explored for the study of ex vivo
prostatic tissue samples, and these have provided the basis for future in vivo
studies. Because the prostate is a solid organ and light typically cannot penetrate the whole organ, optical spectroscopy techniques would have to be incorporated into biopsy needles or applied during prostatectomy to be useful. Sharma et al[78
] reported the development of a needle like, bifurcated, fiber-optic probe for diffuse reflectance spectroscopy measurements in human prostate cancer. The results from 23 prostate specimens demonstrate that the derived hemodynamic parameters and optical properties can serve as good biomarkers to differentiate tumor tissue from normal tissue in the human prostate.
Crow et al[79
] employed a fiber-optic Raman system to differentiate between benign and malignant bladder and prostate pathologic findings in vitro
, in which a total of 197 Raman spectra were recorded from 38 snap-frozen prostate samples collected at transurethral resection of the prostate. An overall accuracy of 86% was reported for differentiation of benign prostatic hyperplasia and prostatitis from prostate cancer.
The combination of multiple complementary techniques has been reported to improve prostate cancer detection. Salomon et al[80
] combined laser-induced autofluorescence, white-light remission, and high-frequency impedance spectroscopy in an ex vivo
study. Ninety-five frozen tissue samples from 32 patients undergoing radical prostatectomy for clinically localized prostate cancer were thawed for data acquisition. The statistical analysis of laser-induced autofluorescence and white-light remission data demonstrated a differentiation of benign and malignant prostate tissue with a sensitivity of 87.5% and a specificity of 87.3%. By adding the acquired high-frequency impedance data to the statistical analysis, sensitivity and specificity were increased to 93.8% and 92.4%.
Interstitial photodynamic therapy (PDT) in prostate cancer has been intensively studied. This therapy is particularly suited to the prostate because of its capability of precise delivery of light dosage and minimum damage to surrounding vital organs. Optical spectroscopy techniques have been shown to be valuable in monitoring relevant light and tissue parameters for the optimization of PDT outcome. Zhu et al[81
] quantified the distribution of light fluence rate, optical properties, drug concentration, and tissue oxygenation for PDT of prostate cancer using diffuse reflectance spectroscopy and fluorescence spectroscopy before and after PDT treatment. This study shows significant inter- and intra-prostatic variations in tissue optical properties and drug distribution, which suggests that a real-time dosimetry measurement and feedback system is needed for monitoring these values during treatment to ensure the outcome.
Yu et al[82
] reported the development of an optical system, combining diffuse reflectance spectroscopy (DRS) for measurement of tumor blood oxygenation and diffuse correlation spectroscopy (DCS) for measurement of tumor blood flow and its application in real time clinical monitoring during interstitial prostate PDT, which was tested on three patients. Prostrate blood oxygen saturation (StO2) was found to decrease only slightly (approximately 3%) after treatment. Prostate blood flow and total hemoglobin concentration over the course of PDT decreased by 50% and 15%, respectively. Johansson et al[23
] developed an instrument for interstitial PDT on prostate tissue that combines therapeutic light delivery and monitoring of light transmission. They demonstrated this using a system to obtain data on the light distribution within the target tissue and to provide real time treatment feedback based on a light dose threshold model for PDT.
Fluorescence spectroscopy was used by Zaak et al[83
] to study the feasibility of 5-aminolevulinic-acid (5-ALA)-induced photodynamic diagnosis (PDD) for margin evaluation during radical prostatectomy (RP) in patients with prostate cancer. Eight out of ten patients demonstrated negative margins and one positive margin in fluorescence measurements, which were confirmed by histology. One positive margin in fluorescence measurements was not confirmed.
A few review articles may help interested readers to learn more about optical spectroscopy in prostate cancer management. Hanchanale et al[84
] reviewed the use of Raman spectroscopy in urological applications including margin assessment during prostatectomy. Manyak et al[85
] reviewed the advance of medical imaging for prostate cancer including optical techniques such as hyperspectral spectroscopy.
Because of easy accessibility, skin cancer has been extensively studied in the past two decades both ex vivo
and in vivo
. The papers reviewed next were all in vivo
studies to highlight the most recent progress. Zonios et al[86
] developed a method for estimating the absorption spectra of melanin in vivo
based on diffuse reflectance spectroscopy of human skin. They found that the histologic transition from dysplastic nevi to melanoma in situ
and then to malignant melanoma was reflected in the melanin absorption spectra.
Marchesini et al[87
] attempted to determine the role of melanin in the various steps of progression of melanocytic neoplasia using diffuse reflectance spectroscopy. They examined 288 melanomas in different phases of progression, i.e. in situ
, horizontal and vertical growth phase invasive melanomas, 424 dysplastic nevi, and 957 melanocytic lesions. The absorbance spectra in the different groups showed that melanin level was correlated with the progression from dysplastic nevi to vertical growth phase melanomas. In addition, it was observed that in vivo
diffuse reflectance spectroscopy can be used to differentiate eumelanin and pheomelanin in lesions.
Sterenborg et al[88
] examined the feasibility of using in vivo
autofluorescence at an excitation wavelength of 375 nm for the diagnosis of skin cancer in 1995. They did not observe any significant differences in the shape of fluorescence spectra or spatial distribution of fluorescence intensity between tumors and the corresponding control sites, which was likely due to the choice of the excitation wavelength.
As technology has advanced since then, several groups have reported the effective diagnosis of skin cancer using fluorescence spectroscopy. Brancaleon et al[89
] examined the autofluorescence of normal skin and nonmelanoma skin cancers (NMSC) in vivo
excited by UV light in 18 patients. They observed that the endogenous fluorescence due to tryptophan residues in both basal cell carcinomas (BCC) and squamous cell carcinomas (SCC) was stronger than in normal tissue, probably due to epidermal thickening and/or hyperproliferation. In contrast, the fluorescence intensity associated with dermal collagen crosslinks was generally lower in tumors than in the surrounding normal tissue, probably because of degradation or erosion of the connective tissue due to enzymes released by the tumor.
Panjehpour et al[90
] used laser-induced fluorescence spectroscopy at the visible excitation wavelength of 410 nm to detect NMSC in vivo
. Two hundred and seventy nine measurements were performed in 49 patients. Patients were classified as having either skin types I, II, or III. Cancers were classified correctly in 93%, 89%, and 78% of patients with skin types I, II, and III, respectively. Normal tissues were classified correctly in 93%, 88%, and 50% of patients with skin types I, II, and III, respectively. Using the same threshold, pre-cancerous spectra were classified correctly in 78% and 100% of patients with skin types I and III, respectively. Benign lesions were classified correctly in 100%, 46%, and 27% of patients with skin types I, II, and III, respectively.
In contrast to UV or visible excitation in conventional fluorescence spectroscopy, Han et al[91
] developed an NIR autofluorescence and reflectance imaging system excited at 785 nm aiming to characterize cutaneous melanins in vivo
. Their preliminary results show that cutaneous melanin in pigmented skin disorders emits higher NIR autofluorescence than surrounding normal tissue. Because NIR light penetrates deeper in the skin, this technique is expected to examine a larger volume of the skin tissue, which may be useful for clinical evaluation and diagnosis of pigmented skin lesions.
Raman spectroscopy also appears to be an effective optical technique for skin cancer diagnosis. Lieber et al[92
] used a portable confocal Raman system to measure Raman spectra from 21 suspected NMSC in 19 patients. A 100% (21/21) sensitivity and 91% (19/21) specificity for abnormality, with a 95% (40/42) overall classification accuracy were achieved.
Zhao et al[93
] reported the development of a rapid real-time Raman spectrometer system with measurement times of less than 1 s in a preliminary study. In total, 289 skin cancers and benign skin lesions were measured. Skin cancers could be well differentiated from benign skin lesions (sensitivity 91% and specificity 75%) and malignant melanoma from benign pigmented lesions (sensitivity 97% and specificity 78%).
For a review on the application of Raman spectroscopy in skin cancer, please refer to Eikje et al[94
], which includes a survey of introduced sampling methods for IR and Raman spectroscopy in dermatology, and describes the differences between microscopic, macroscopic and fiber-optic measurements of skin cancer. The authors are optimistic about the potential role of vibrational spectroscopy including Raman spectroscopy as a rapid screening tool in dermatology. Krafft et al[95
] also reviewed the application of Raman spectroscopy in the recognition of a variety of diseases including skin tumors. Mogensen et al[96
] reviewed the diagnostic accuracy of nonmelanoma skin cancer diagnostic tests and technologies including optical spectroscopy techniques such as spectroscopy and fluorescence imaging. They pointed out the need for larger scale trials despite the promising diagnostic accuracy using optical techniques.