1 May 2010 In vivo early diagnosis of gastric dysplasia using narrow-band image-guided Raman endoscopy
Author Affiliations +
We first report on the implementation of a novel narrow-band image-guided Raman endoscopy technique for in vivo diagnosis of gastric dysplasia. High-quality in vivo Raman spectra can be acquired from normal and dysplastic gastric mucosal tissue within 0.5 sec under narrow-band image (NBI) guidance at gastroscopy. Significant differences are observed in in vivo Raman spectra between normal (n=54) and dysplastic (n=18) gastric tissue from 30 gastric patients, particularly in the spectral ranges of 825 to 950, 1000 to 1100, 1250 to 1500, and 1600 to 1800 cm−1, which primarily contain signals related to proteins, nucleic acids, and lipids. The multivariate analysis [i.e., principal components analysis (PCA) and linear discriminant analysis (LDA)], together with the leave-one tissue site-out, cross validation on in vivo gastric Raman spectra yields a diagnostic sensitivity of 94.4% (17/18) and specificity of 96.3% (52/54) for distinction of gastric dysplastic tissue. This study suggests that narrowband image-guided Raman endoscopy associated with PCA-LDA diagnostic algorithms has potential for the noninvasive, in vivo early diagnosis and detection of gastric precancer during clinical gastroscopic examination.



Gastric cancer is the second leading cause of cancer-associated death, accounting for approximately 600,000 annual deaths worldwide.1 Early diagnosis and localization with appropriate curative treatments (e.g., endoscopic submucosal dissection and gastrectomy) is critical to decreasing mortality.2 However, identification of early cancer and precancer can be difficult, as conventional white-light reflectance (WLR) endoscopy heavily relies on visual identification of morphological tissue changes. Thus, subtle changes of gastric precancer (i.e., dysplasia) and early cancer may not be apparent, limiting diagnostic accuracy. Positive endoscopic biopsy is the standard criterion for gastric precancer and cancer diagnosis, but it is invasive and impractical for screening high-risk patients who may have multiple suspicious lesions.1, 2 Very recently, narrow-band imaging (NBI) that enhances visualization of irregular mucosal and vascular patterns has shown promise for improving in vivo diagnosis of intraepithelial neoplastic lesions in gastric tissue.3 Although the NBI technique provides good detection sensitivities, this wide-field endoscopic imaging modality still suffers from moderate diagnostic specificity due to the deficiency of revealing specific biochemical information of tissue. Hence, a noninvasive optical diagnostic technique providing a direct assessment of biochemical information of suspicious lesions in vivo will represent a significant advance in endoscopic detection of precancer and early cancer in gastric tissue.

In the past decade, Raman spectroscopy, which measures inelastic light scattering processes and is capable of providing spectroscopic fingerprints of specific molecular structures and compositions of biological tissues, has been comprehensively investigated for precancer and cancer diagnosis and evaluation in a number of organs, including the stomach.4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14 These investigations show that specific spectral features of Raman spectra could be used to correlate with the molecular and structural changes of tissue associated with neoplastic transformation. Diagnostic sensitivities of 85 to 95% and specificities of 90 to 98% have been reported for differentiation between normal and pathological (e.g., dysplasia, adenocarcinoma) gastric tissues in vitro using Raman spectroscopic technique.12, 13, 14 However, to date, in vivo Raman studies for early diagnosis of gastric precancer and cancer have not been reported. In vivo Raman endoscopic applications have been limited not only by the difficulty of capturing inherently weak tissue Raman signals, but also by the relatively high speed of spectral acquisitions required in clinical settings.4, 5, 6 The fabrication of a millimeter-scaled or even smaller flexible fiber optic Raman probe with abilities of efficient fiber fluorescence/Raman background rejections while having high tissue Raman signal collection efficiency renders another technical challenge for clinical Raman endoscopic examinations.6, 8, 15 Very recently, we have successfully developed a high throughput Raman endoscopy system integrated with trimodal wide-field endoscopic imaging for rapid in vivo gastric tissue Raman measurements at gastroscopy.16 With the unique image-guided Raman endoscopy technique developed, we aim, in this work, to evaluate the clinical utility of Raman spectroscopy for in vivo diagnosis and detection of gastric dysplasia during clinical gastroscopic examination.


Materials and Methods

The Raman endoscopy system developed16 consists of a spectrum-stabilized 785-nm diode laser (maximum output: 300mW , B&W TEK Incorporated, Newark, Delaware), a transmissive imaging spectrograph (Holospec f/1.8, Kaiser Optical Systems, Ann Arbor, Michigan), a liquid nitrogen-cooled, near-infrared (NIR)-optimized, back-illuminated deep depletion charge-coupled device (CCD) camera (Spec-10: 400BR/LN, Princeton Instruments, Trenton, New Jersey), and a specially designed Raman endoscopic probe for both laser light delivery and in vivo tissue Raman signal collection. The 1.8-mm fiber optic Raman endoscopic probe, which can fit into the biopsy channel of medical endoscopes, consists of 32 Raman collection fibers (core diameter of 200μm ) surrounding the central laser light delivery fiber (core diameter of 200μm ), with two stages of optical filtering incorporated at the proximal and distal ends of the probe for maximizing the collection of tissue Raman signals while reducing the interference of Rayleigh scattered light, fiber fluorescence, and silica Raman signals.16 A personal computer controls the system using a custom-designed software that triggers real-time data acquisition and online data analysis (e.g., CCD dark-noise subtraction, wavelength calibration, system spectral response calibration, signal saturation detection, cosmic ray rejection, tissue autofluorescence background subtraction, etc.). The system acquires in vivo tissue Raman spectra in the wave number range of 800to1800cm1 , and the spectral resolution of the system is 9cm1 . Each spectrum was acquired within 0.5sec using the 785-nm laser light irradiance of 1.5Wcm2 , which is less than the American National Standards Institute (ANSI) maximum permissible skin exposure limit set out for a 785-nm laser beam.17 Our further calculations based on the finite difference thermal model18, 19 and the optical properties of stomach tissue20 indicate that even without consideration of other cooling effects (e.g., perfusion and evaporation in tissue), the maximum tissue temperature rise is only about 0.15°C after 1min of 785-nm laser radiation with an incident power of 30mW on a tissue spot size of 200μm during tissue Raman measurements. This temperature rise estimated is far below the level to generate cytotoxicity in tissue and cells,21 suggesting that the laser power density used in this study is safe for in vivo tissue Raman measurements. Note that the raw spectra acquired from in vivo gastric tissue represent a combination of intense tissue autofluorescence, weak tissue Raman scattering signals, and noise. These raw spectra are preprocessed by a first-order Savitsky–Golay filter (window width of 3pixels , corresponding to the system spectral resolution) to reduce noise. A fifth-order polynomial5 is found to be optimal for fitting the broad autofluorescence background in the noise-smoothed spectrum, and this polynomial is then subtracted from the raw spectrum to yield the tissue Raman spectrum alone. Each background-subtracted Raman spectrum is also normalized to the integrated area under the curve from 800to1800cm1 , enabling better comparison of the spectral shapes and relative peak intensities among different gastric tissues.5 All of the spectra preprocessing is completed online and the Raman spectrum can be displayed in real time during clinical Raman measurements at gastroscopy. The trimodal endoscope imaging system primarily comprises a 300-W dedicated short-arc xenon light source, a gastrointestinal (GI) video-endoscope (GIF-FQ260Z, Olympus), and a video system processor (CV-260SL, Olympus) for white-light reflectance (WLR) imaging, autofluorescence imaging (AFI), and narrow-band imaging (NBI) during gastroscopic examination. Both the wide-field endoscopic image (WLR/NBI/AFI) and the point-wise in vivo Raman spectra of the tissue imaged can be simultaneously acquired, stored, and displayed in real time on a color video monitor and computer screen, respectively.

In this work, all patients signed an informed consent permitting the in vivo Raman endoscopic measurements of gastric tissue in the Endoscope Centre at the National University Hospital (NUH), Singapore. This study was approved by the Institutional Review Board (IRB) of the National Healthcare Group (NHG) of Singapore. We have acquired in vivo Raman spectra [Fig. 1 ] of 72 gastric mucosal tissue sites in 30 gastric patients (16 men and 14 women, with a median age of 67) under the guidance of wide-field endoscopic imaging [e.g., NBI and WLR imaging in Fig. 1] during gastroscopic examination. The Raman endopscopic probe was placed in gentle contact with the gastric mucosa surface, and the positioning against the tissue sites was verified on the endoscopy monitor by the endoscopists in charge during gastroscopic examinations. Immediately after all Raman acquisitions, the biopsy samples were taken from the tissue sites measured (with suction markings) and fixed in 10% formalin solution for histopathological examinations by a senior gastrointestinal pathologist. For the assessment of diagnostic sensitivity and specificity of Raman endoscopy for gastric tissue classification, histopathological results served as the golden standard.

Fig. 1

In vivo mean Raman spectra±1 standard deviations (SD) of normal (n=54) and dysplastic (n=18) gastric tissue. The difference spectrum (i.e., the mean Raman spectrum of dysplastic tissue minus the mean Raman spectrum of normal tissue) reveals the Raman spectral changes, particularly in the regions of 825 to 950, 1000 to 1100, 1250 to 1500, and 1600to1800cm1 , associated with dysplastic transformation. Note that in vivo Raman spectra of both normal and dysplastic gastric tissues are vertically shifted for better visualization. The shaded areas in tissue Raman spectra stand for the respective standard deviations. (b) Narrow-band image (NBI) and white-light reflectance (WLR) image of dysplastic gastric lesions in the antrum acquired during clinical gastroscopy. NBI enhances the observation of the irregular mucosal glandular patterns and vascular patterns of dysplastic gastric mucosal lesions.



Results and Discussion

Figure 1 shows the in vivo mean Raman spectra±1 standard deviations (SD) and the corresponding Raman difference spectrum of normal (n=54) and dysplastic (n=18) gastric tissue. The representative NBI and WLR image of dysplastic gastric tissue acquired under Raman endoscopic measurements are shown in Fig. 1. All in vivo tissue Raman spectra are acquired within 0.1to0.5sec (depending on the autofluorescence background level of different gastric tissues) with the 785-nm light irradiance power of 1.5Wcm2 . Prominent Raman bands are observed in both normal and dysplastic gastric tissue at the following peak positions with tentative biochemical assignments:4, 5, 6, 7, 9, 10, 11, 12, 13, 14 875cm1 [ v(CC) hydroxyproline], 1004cm1 [ v(CC) ring breathing of phenylalanine], 1078cm1 [ v(CC) or v(CO) of phospholipids], 1265cm1 [amide III v(CN) and δ(NH) of proteins], 1302cm1 ( CH2CH3 twisting of proteins and nucleic acids), 1450cm1 [ δ(CH2) of proteins and lipids], 1655cm1 [amide I v(CO) of proteins], and 1745cm1 [ v(CO) of phospholipids]. The difference spectrum [Fig. 1] reveals the changes of relative percentages of distinctive biomolecules in dysplastic tissue, particularly in the spectral ranges of 825 to 950, 1000 to 1100, 1250 to 1500, and 1600to1800cm1 , which contain signals related to proteins, nucleic acids, and lipids, respectively. Dysplasia tissues show lower intensities at 875, 1004, 1078, 1210, and 1745cm1 , while higher intensities at 1265, 1302, 1450, and 1655cm1 compared with normal tissue. This indicates that there is an increase or decrease in the percentage of a certain type of biomolecules relative to the total Raman-active constituents in gastric tissue associated with dysplastic transformation. More specifically, the decrease in Raman signals at 875cm1 (hydroxyproline of collagen) may reflect cytoplasmic mucin depletion and the elevated concentration of metalloproteinase that cleaves collagen in the stroma layer of gastric dysplasia tissue.13, 14, 22, 23 In addition, the thickening of the epithelium associated with dysplastic progression can attenuate the excitation laser power and also obscure the collagen Raman emission from the deep basal membrane,24, 25 thereby resulting in an overall decrease of Raman intensity at 875cm1 from dysplasia tissue. The increase of Raman signals attributed to proteins (e.g., 1265, 1302, 1450, and 1655cm1 ) in gastric dysplasia may be associated with an increase in the metabolic activities (e.g., increased mitotic activities that include enzymes, hormones, etc.) as well as the increased hyperchromatism and the nucleic acids-to-cytoplasm ratio of dysplastic cells.23, 24 The decrease in Raman signals in the region 1745cm1 suggests a reduction of phospholipids in dysplastic tissue that also has been found in other epithelial tissues with malignancies.5, 10 Further, there are also obvious changes of Raman peak shifts and bandwidths in the spectral ranges of 1200to1500cm1 and 1600to1800cm1 , which are related to the amide III and amide I of proteins, CH3CH2 twisting of proteins/nucleic acids, and CO stretching of phospholipids for dysplasia. The current results of in vivo gastric tissue Raman spectra are also similar to our previous observations of in vitro gastric tissue Raman spectra.12, 13, 14 Therefore, the distinctive differences in in vivo Raman spectra between normal and dysplastic tissue confirm the potential role of Raman endoscopy for in vivo gastric precancer diagnosis at the molecular level.

We have employed the multivariate statistical technique [i.e., principal components analysis (PCA) and linear discriminant analysis (LDA)] by utilizing the entire Raman spectra to determine the most diagnostically significant Raman features for gastric tissue classification. The unpaired, two-sided Student’s t-test on the obtained principal components (PCs) shows that the three PCs (PC1, PC2, and PC4) (Fig. 2 ) accounting for 70.5% of the total variance contain the most diagnostically significant Raman features (p<0.05) for discriminating dysplasia from normal gastric tissue. The first PC accounts for the largest variance ( 52.6% of the total variance), whereas the successive PCs describe the spectral features that contribute progressively smaller variances (e.g., PC213.2% , PC44.7% ). Some PC features, such as the peaks, troughs, and spectral shapes in Fig. 2, are similar to those of tissue Raman spectra (Fig. 1).

Fig. 2

The three significant principal components (PCs) ( PC152.6% , PC213.2% , and PC44.7% ) accounting for 70.5% of the total variance calculated from in vivo gastric Raman spectra.


All three significant PCs were loaded into the LDA model to develop effective diagnostic algorithms for gastric tissue diagnosis. Figure 3 shows the classification results of in vivo Raman spectra of normal and precancer gastric tissue using PCA-LDA together with the leave-one tissue site-out, cross-validation techniques. The PCA-LDA algorithm based on in vivo gastric Raman spectra provides a diagnostic sensitivity of 94.4% (1718) and specificity of 96.3% (5254) for distinguishing dysplasia from normal gastric tissue in vivo. To further evaluate the performance of the PCA-LDA-based diagnostic algorithms derived from all three significant PCs of in vivo gastric Raman datasets, the receiver operating characteristic (ROC) curve (Fig. 4 ) is also generated from the scatter plot in Fig. 3 at different threshold levels, displaying the discrimination results using PCA-LDA diagnostic algorithms together with the leave-one tissue site-out, cross-validation method. The integration area under the ROC curve is 0.997, further demonstrating the diagnostic efficacy of Raman endoscopy for in vivo diagnosis of gastric precancer during clinical gastroscopic examination.

Fig. 3

Scatter plot of the linear discriminant scores belonging to normal and dysplasia gastric tissue using the PCA-LDA algorithms, together with the leave-one tissue site-out, cross-validation method. The separate line yields a diagnostic sensitivity of 94.4% (1718) and specificity of 96.3% (5254) for distinguishing dysplasia from normal gastric tissue in vivo.


Fig. 4

Receiver operating characteristic (ROC) curve of discrimination results of in vivo Raman spectra of normal and dysplastic gastric tissue using the PCA-LDA algorithms, together with the leave-one tissue site-out, cross-validation method.


In summary, we have acquired, for the first time, high-quality in vivo Raman spectra from normal and dysplastic gastric tissue within 0.5sec under narrow-band imaging guidance during clinical gastroscopy. We have observed the significant differences in in vivo Raman spectra between normal and dysplastic gastric tissue. Good differentiation between normal and dysplastic gastric tissues can be achieved using PCA-LDA diagnostic algorithms, indicating the potential of Raman endoscopy for in vivo diagnosis of gastric precancer. Currently, we are conducting in vivo Raman measurements on a larger series of gastric patients at National University Hospital, Singapore, to further evaluate the clinical merits of Raman endoscopy techniques for prospective prediction of gastric precancer and early cancer in vivo during gastroscopic examination.


This research was supported by the National Medical Research Council, the Biomedical Research Council, and the Faculty Research Fund from the National University of Singapore.


1.  B. J. Dicken, D. L. Bigam, C. Cass, J. R. Mackey, A. A. Joy, and S. M. Hamilton, “Gastric adenocarcinoma: review and considerations for future directions,” Ann. Surg.0003-4932 241, 27–39 (2005). Google Scholar

2.  C. J. Clark, R. C. Thirlby, V. Jr. Picozzi, D. B. Schembre, F. P. Cummings, and E. Lin, “Current problems in surgery: gastric cancer,” Curr. Probl Surg.0011-3840 43, 566–570 (2006). 10.1067/j.cpsurg.2006.06.003 Google Scholar

3.  A. J. Overhiser and P. Sharma, “Advances in endoscopic imaging: narrow band imaging,” Rev. Gastroenterol. Disord. 8, 186–193 (2008). Google Scholar

4.  N. Stone, P. Stavroulaki, C. Kendall, M. Birchall, and H. Barr, “Raman spectroscopy for early detection of laryngeal malignancy: preliminary results,” Laryngoscope0023-852X 110, 1756–1763 (2000). 10.1097/00005537-200010000-00037 Google Scholar

5.  Z. Huang, A. McWilliams, H. Lui, D. I. McLean, S. Lam, and H. Zeng, “Near-infrared Raman spectroscopy for optical diagnosis of lung cancer,” Int. J. Cancer0020-7136 107, 1047–1052 (2003). 10.1002/ijc.11500 Google Scholar

6.  M. G. Shim, L. M. Song, N. E. Marcon, and B. C. Wilson, “In vivo near-infrared Raman spectroscopy: demonstration of feasibility during clinical gastrointestinal endoscopy,” Photochem. Photobiol.0031-8655 72, 146–150 (2000). 10.1562/0031-8655(2000)072<0146:IVNIRS>2.0.CO;2 Google Scholar

7.  E. Widjaja, W. Zheng, and Z. Huang, “Classification of colonic tissues using near-infrared Raman spectroscopy and support vector machines,” Int. J. Oncol.1019-6439 32, 653–662 (2008). Google Scholar

8.  M. A. Short, S. Lam, A. McWilliams, J. Zhao, H. Lui, and H. Zeng, “Development and preliminary results of an endoscopic Raman probe for potential in vivo diagnosis of lung cancers,” Opt. Lett.0146-9592 33, 711–713 (2008). 10.1364/OL.33.000711 Google Scholar

9.  D. P. Lau, Z. Huang, H. Lui, C. S. Man, K. Berean, M. D. Morrison, and H. Zeng, “Raman spectroscopy for optical diagnosis in normal and cancerous tissue of the nasopharynx-preliminary findings,” Lasers Surg. Med.0196-8092 32, 210–214 (2003). 10.1002/lsm.10084 Google Scholar

10.  S. K. Teh, W. Zheng, D. P. Lau, and Z. Huang, “Spectroscopic diagnosis of laryngeal carcinoma using near-infrared Raman spectroscopy and random recursive partitioning ensemble techniques,” Analyst (Cambridge, U.K.)0003-2654 134, 1232–1239 (2009). 10.1039/b811008e Google Scholar

11.  A. Nijssen, T. C. Bakker, F. Heule, P. J. Caspers, D. P. Hayes, M. H. Neumann, and G. J. Puppels, “Discriminating basal cell carcinoma from its surrounding tissue by Raman spectroscopy,” J. Invest. Dermatol.0022-202X 119, 64–69 (2002). 10.1046/j.1523-1747.2002.01807.x Google Scholar

12.  S. K. Teh, W. Zheng, K. Y. Ho, M. Teh, K. G. Yeoh, and Z. Huang, “Diagnosis of gastric cancer using near-infrared Raman spectroscopy and classification and regression tree techniques,” J. Biomed. Opt.1083-3668 13, 034013 (2008). 10.1117/1.2939406 Google Scholar

13.  S. K. Teh, W. Zheng, K. Y. Ho, M. Teh, K. G. Yeoh, and Z. Huang, “Near-infrared Raman spectroscopy for gastric precancer diagnosis,” J. Raman Spectrosc.0377-0486 40, 908–914 (2009). 10.1002/jrs.2197 Google Scholar

14.  S. K. Teh, W. Zheng, K. Y. Ho, M. Teh, K. G. Yeoh, and Z. Huang, “Diagnostic potential of near-infrared Raman spectroscopy in the stomach: differentiating dysplasia from normal tissue,” Br. J. Cancer0007-0920 98, 457–465 (2008). 10.1038/sj.bjc.6604176 Google Scholar

15.  Y. Hattori, Y. Komachi, T. Asakura, T. Shimosegawa, G. Kanai, H. Tashiro, and H. Sato, “In vivo Raman study of the living rat esophagus and stomach using a micro-Raman probe under an endoscope,” Appl. Spectrosc.0003-7028 61, 579–584 (2007). 10.1366/000370207781269747 Google Scholar

16.  Z. Huang, S. K. Teh, W. Zheng, J. Mo, K. Lin, X. Shao, K. Y. Ho, M. Teh, and K. G. Yeoh, “Integrated Raman spectroscopy and trimodal wide-field imaging techniques for real-time in vivo tissue Raman measurements at endoscopy,” Opt. Lett.0146-9592 34, 758–760 (2009). 10.1364/OL.34.000758 Google Scholar

17. “American National Standard for the Safe Use of Lasers,” ANSI Standard 2136.1-1986, American National Standards Institute, Washington, D.C. (1986). Google Scholar

18.  F. P. Incropera and D. P. D. Witt, Fundamentals of Heat and Mass Transfer, John Wiley and Sons, New York (1990). Google Scholar

19.  J. H. Torres, M. Motamedi, J. A. Pearce, and A. J. Welch, “Experimental evaluation of mathematical models for predicting the thermal response of tissue to laser irradiation,” Appl. Opt.0003-6935 32, 597–606 (1993). 10.1364/AO.32.000597 Google Scholar

20.  A. N. Bashkatov, E. A. Genina, V. I. Kochubey, A. A. Gavrilova, S. V. Kapralov, V. A. Grishaev, and V. V. Tuchin, “Optical properties of human stomach mucosa in the spectral range from 400to2000nm: Prognosis for gastroenterology,” Med. Laser Appl.1615-1615 22, 95–104 (2007). 10.1016/j.mla.2007.07.003 Google Scholar

21.  S. Thomsen, “Pathologic analysis of photothermal and photomechanical effects of laser-tissue interactions,” Photochem. Photobiol.0031-8655 53, 825–835 (1991). Google Scholar

22.  P. Correa, “A human model of gastric carcinogenisis,” Cancer Res.0008-5472 48, 3554–3560 (1988). Google Scholar

23.  I. Georgakoudi, B. C. Jacobson, M. G. Müller, E. E. Sheets, K. Badizadegan, D. L. Carr Locke, C. P. Crum, C. W. Boone, R. R. Dasari, J. Van Dam, and M. S. Feld, “NAD(P)H and collagen as an in vivo quantitative fluorescence biomarkers of epithelial precancerous changes,” Cancer Res.0008-5472 62, 682–687 (2002). Google Scholar

24.  J. H. Hughes, C. J. Leigh, S. S. Raab, S. Y. Hook, M. B. Cohen, and M. J. Suhrland, “Cytologic criteria for the brush diagnosis of gastric adenocarcinoma,” Cancer0008-543X 84, 289–294 (1998). 10.1002/(SICI)1097-0142(19981025)84:5<289::AID-CNCR4>3.0.CO;2-L Google Scholar

25.  Z. Huang, W. Zheng, S. Xie, R. Chen, H. Zeng, D. I. McLean, and H. Lui, “Laser-induced autofluorescence microscopy of normal and tumor human colonic tissue,” Int. J. Oncol.1019-6439 24, 59–64 (2004). Google Scholar

© (2010) Society of Photo-Optical Instrumentation Engineers (SPIE)
Zhiwei Huang, Zhiwei Huang, Mads Sylvest Bergholt, Mads Sylvest Bergholt, Wei Zheng, Wei Zheng, Kan Lin, Kan Lin, Khek-Yu Ho, Khek-Yu Ho, Ming The, Ming The, Khay-Guan Yeoh, Khay-Guan Yeoh, } "In vivo early diagnosis of gastric dysplasia using narrow-band image-guided Raman endoscopy," Journal of Biomedical Optics 15(3), 037017 (1 May 2010). https://doi.org/10.1117/1.3420115 . Submission:

Back to Top