Fiber-optic Raman spectroscopy for in vivo diagnosis of gastric dysplasia

Abstract

This study aims to assess the clinical utility of a rapid fiber-optic Raman spectroscopy technique developed for enhancing in vivo diagnosis of gastric precancer during endoscopic examination. We have developed a real-time fiber-optic Raman spectroscopy system capable of simultaneously acquiring both fingerprint (FP) (i.e., 800-1800 cm(-1)) and high-wavenumber (HW) (i.e., 2800-3600 cm(-1)) Raman spectra from gastric tissue in vivo at endoscopy. A total of 5792 high-quality in vivo FP/HW Raman spectra (normal (n = 5160); dysplasia (n = 155), and adenocarcinoma (n = 477)) were acquired in real-time from 441 tissue sites (normal (n = 396); dysplasia (n = 11), and adenocarcinoma (n = 34)) of 191 gastric patients (normal (n = 172); dysplasia (n = 6), and adenocarcinoma (n = 13)) undergoing routine endoscopic examinations. Partial least squares discriminant analysis (PLS-DA) together with leave-one-patient-out cross validation (LOPCV) were implemented to develop robust spectral diagnostic models. The FP/HW Raman spectra differ significantly between normal, dysplasia and adenocarcinoma of the stomach, which can be attributed to changes in proteins, lipids, nucleic acids, and the bound water content. PLS-DA and LOPCV show that the fiber-optic FP/HW Raman spectroscopy provides diagnostic sensitivities of 96.0%, 81.8% and 88.2%, and specificities of 86.7%, 95.3% and 95.6%, respectively, for the classification of normal, dysplastic and cancerous gastric tissue, superior to either the FP or HW Raman techniques alone. Further dichotomous PLS-DA analysis yields a sensitivity of 90.9% (10/11) and specificity of 95.9% (380/396) for the detection of gastric dysplasia using FP/HW Raman spectroscopy, substantiating its clinical advantages over white light reflectance endoscopy (sensitivity: 90.9% (10/11), and specificity: 51.0% (202/396)). This work demonstrates that the fiber-optic FP/HW Raman spectroscopy technique has great promise for enhancing in vivo diagnosis of gastric precancer during routine endoscopic examination.

Integrated Mueller-matrix near-infrared imaging and point-wise spectroscopy improves colonic cancer detection

“…The enhanced metabolic rate in colonic cancer [36] contributes to the increased water content as water provides the conversion of mechanical energy developed by contractile proteins into the chemical energy useful in cell process [37]. The increased water for the cancerous colonic tissue has also been observed in other cancer tissues (e.g., esophagus [28], stomach [38], cervix [39][40][41] and brain [42]) by using Raman spectroscopy [4,28,[38][39][40][41][42]. We have also found the decreased DR of hemoglobin band (940 nm) for the colonic cancer, signifying the increased hemoglobin content associated with colonic cancer tissue.…”

“…6) shows that colon cancer was identified with an accuracy of 90.0%, 91.7%, and 80.0%, respectively, by using diattentuation, depolarization, and retardance metrics (Table 1). Remarkably, the combination of the three polarization metrics with majority voting [38] provides an enhanced colonic cancer detection with an accuracy of 95.0% (sensitivity of 93.3%, and specificity of 96.7%), superior to using either of the three polarization metrics alone (Table 1). …”

Characterizing biochemical and morphological variations of clinically relevant anatomical locations of oral tissue in vivo with hybrid Raman spectroscopy and optical coherence tomography technique

“…Prominent FP Raman peaks were identified in the oral cavity as follows: 853 cm −1 ( v (C─C) proteins), 1004 cm −1 ( ν s (C─C) ring breathing of phenylalanine), 960 cm −1 ( ν s (P─O) of hydroxyapatite), 1078 cm −1 ( ν (C─C) of lipids), 1265 cm −1 (amide III v (C─N) and δ(N─H) of proteins), 1302 cm −1 (CH 2 twisting and wagging of lipids), 1445 cm −1 ( δ (CH 2 ) deformation of proteins and lipids) and 1655 cm −1 (amide I v (C═O) of proteins) in the FP range . In the HW range, intense Raman peaks are also observed: that is, 2850 and 2885 cm −1 (symmetric and asymmetric CH 2 stretching of lipids), 2940 cm −1 (CH 3 stretching of proteins) and the broad Raman band of water (OH stretching vibrations peaking at ~3250 and ~3400 cm −1 ) that are related to the local conformation and interactions of OH‐bonds in the intracellular and extracellular space of oral tissue . The difference Raman spectra (ie, lateral tongue − floor of mouth, alveolar process − floor of mouth, and alveolar process − lateral tongue) ± 1SD (shaded area) are shown in Figure B.…”

“…The Raman path is launched from the fiber‐optic Raman probe, focused by the converging lens, and deflected by 90 ° using the LPDM, and then focused onto the tissue surface through the glass window . In vivo oral Raman spectra were acquired within 0.5 s by using the RS‐OCT technique coupled with a simultaneous fingerprint (FP: 800‐1800 cm −1 ) and high wavenumber (HW: 2800‐3600 cm −1 ) RS system . The FP/HW RS system comprises a 785 nm diode laser (maximum output: 300 mW, B&W TEK Inc.), a high‐throughput reflective imaging spectrograph (Acton LS‐785 f/2, Newark, DE; Princeton Instruments Inc., Trenton, NJ) equipped with a customized 830 g/mm gold‐coated grating that has a ~85% diffraction efficiency in the NIR range of 800‐1200 nm.…”

“…The one‐way analysis of variance (ANOVA) with Fisher post hoc least significant difference (LSD) analysis was used to elucidate the statistically different Raman‐active components among the different sites (ie, alveolar process, lateral tongue and floor of the mouth) in the oral cavity. Partial least squares (PLS)‐discriminant analysis (DA) and leave‐one‐tissue site out, cross‐validation (LOOCV) were applied on both Raman spectra and OCT images to classify the different sites of the oral cavity . The multivariate statistical analysis was performed using in‐house written scripts in the Matlab programming environment (Mathworks, Inc., Natick, MA).…”