Université de Reims Champagne-Ardenne, MéDIAN-Biophotonique et Technologies pour la Santé, UFR de Pharmacie, 51 rue Cognacq-Jay, 51096 REIMS cedex, France; CNRS UMR7369, Matrice Extracellulaire et Dynamique Cellulaire, MEDyC, Reims, France.
J Biophotonics. 2014 Apr;7(3-4):241-53. doi: 10.1002/jbio.201300166. Epub 2014 Mar 28.
This study aimed at determining whether FTIR spectroscopy is able to distinguish bile samples from patients with and without malignant biliary strictures. Bile samples were collected in 19 patients with malignant biliary strictures and 38 with benign biliary diseases during endoscopic procedures. FTIR spectra were acquired on dried drops of whole bile, aqueous and organic phases obtained after lipid extraction. Data were analyzed by principal component analysis and by the support vector machine classification using a leave-n-out cross validation procedure. This was applied to the whole set of spectra and the mean and median spectra of each patient. By leaving one patient out, the classifier allowed discriminating patients with and without malignant biliary strictures with a sensitivity between 82% and 95% and a specificity between 85% and 100%. Using a randomized leave-n -out cross-validation with n = 2, 5 and 10 patients, the sensitivity decreased slightly by about 5 to 10% while the specificity remained stable, suggesting the robustness of the classifier. FTIR spectroscopy combined with chemometrics therefore shows potential to differentiate bile from patients with and without malignant biliary strictures. Although promising, the results of this pilot study cannot be generalized and needs to be confirmed in a larger population.
本研究旨在确定傅里叶变换红外(FTIR)光谱是否能够区分恶性胆道狭窄患者和无恶性胆道狭窄患者的胆汁样本。在内镜检查过程中,收集了 19 例恶性胆道狭窄患者和 38 例良性胆道疾病患者的胆汁样本。对干燥的全胆汁滴、脂质提取后获得的水相和有机相进行 FTIR 光谱采集。通过主成分分析和支持向量机分类,使用留一交叉验证程序对数据进行分析。这适用于整个光谱集以及每位患者的平均和中位数光谱。通过留出一位患者,分类器可以区分恶性胆道狭窄患者和无恶性胆道狭窄患者,其敏感性在 82%至 95%之间,特异性在 85%至 100%之间。使用随机留一交叉验证(n = 2、5 和 10),敏感性略有下降约 5%至 10%,而特异性保持稳定,表明分类器的稳健性。因此,FTIR 光谱结合化学计量学具有区分恶性胆道狭窄患者和无恶性胆道狭窄患者胆汁的潜力。尽管有希望,但这项初步研究的结果不能推广,需要在更大的人群中得到证实。