Hofstra Northwell School of Medicine, Northwell Health System, Division of Gastroenterology, Department of Medicine, Long Island Jewish Medical Center, New Hyde Park, New York, USA.
Lewis Katz School of Medicine at Temple University, Philadelphia, Pennsylvania, USA.
Gastrointest Endosc. 2017 Jul;86(1):133-139. doi: 10.1016/j.gie.2016.11.026. Epub 2016 Nov 27.
Targeting neoplasia in Barrett's esophagus (BE) is challenging. Volumetric laser endomicroscopy (VLE) is a new imaging technique that allows for real time cross-sectional microstructure imaging that can detect BE neoplasia. The interobserver agreement among users in practice is unknown.
Eight high-volume users of VLE from different academic centers in the United States evaluated 120 stored VLE images blinded to the endoscopic and clinical findings. There were 30 images for each tissue type: gastric cardia, esophageal squamous mucosa, nonneoplastic BE, and neoplastic BE. Each image with BE had corresponding histology confirming the tissue diagnosis. Each normal esophagus and gastric cardia had matching endoscopic images confirming normal mucosa. These were considered the criterion standard. Respondents were asked to classify the images into 1 of each category. Agreement among the users was measured.
The overall agreement among users was almost perfect (kappa = 0.81; 95% confidence interval [CI], 0.79-0.83). For esophageal squamous and gastric cardia, the agreement was almost perfect (kappa 0.95 and 0.86, respectively [95% CI, 0.92-0.98 and 0.83-0.89]). For nonneoplastic BE and neoplastic BE, the agreement was strong (kappa 0.66 [95% CI, 0.63-0.69] and 0.79 [95% CI, 0.75-0.82], respectively). When compared with the criterion standard, the median accuracy for identifying normal squamous mucosa, normal gastric mucosa, nonneoplastic BE, neoplastic BE, and all tissue types was 99% (95% CI, 98%-100%), 97% (95% CI, 95%-99%), 93% (95% CI, 88%-98%), 95% (95% CI, 91%-99%), and 96% (95% CI, 94%-99%), respectively.
VLE has a high level of agreement and accuracy among high-volume users.
靶向 Barrett 食管(BE)中的肿瘤具有挑战性。容积激光内窥技术(VLE)是一种新的成像技术,允许实时进行横截面微观结构成像,从而能够检测 BE 中的肿瘤。目前尚不清楚该技术在实践中不同使用者之间的观察一致性。
来自美国 8 个不同学术中心的 8 名 VLE 高用量使用者对 120 个存储的 VLE 图像进行评估,这些图像对内镜和临床发现均为盲法。每个组织类型有 30 个图像:胃贲门、食管鳞状黏膜、非肿瘤性 BE 和肿瘤性 BE。每个具有 BE 的图像均有相应的组织学证实其组织诊断。每个正常食管和胃贲门都有相应的内镜图像证实正常黏膜。这些被视为标准。回答者被要求将图像分类为每一类中的 1 个。测量用户之间的一致性。
用户之间的总体一致性几乎是完美的(kappa=0.81;95%置信区间[CI],0.79-0.83)。对于食管鳞状和胃贲门,一致性几乎是完美的(kappa 值分别为 0.95 和 0.86 [95%CI,0.92-0.98 和 0.83-0.89])。对于非肿瘤性 BE 和肿瘤性 BE,一致性是强的(kappa 值分别为 0.66 [95%CI,0.63-0.69]和 0.79 [95%CI,0.75-0.82])。与标准相比,识别正常鳞状黏膜、正常胃黏膜、非肿瘤性 BE、肿瘤性 BE 和所有组织类型的中位数准确性分别为 99%(95%CI,98%-100%)、97%(95%CI,95%-99%)、93%(95%CI,88%-98%)、95%(95%CI,91%-99%)和 96%(95%CI,94%-99%)。
VLE 在高用量使用者中具有高度的一致性和准确性。