Numerical Analysis of Autofluorescence Intensity of Neoplastic and Non-Neoplastic Colorectal Lesions Using a Novel Videoendoscopy System

Abstract

Background and Aim: Recently, a novel high-resolution video autofluorescence endoscopy (AFE) system was developed. This study aimed to numerically evaluate the color tones in AFE images, in order to determine whether this innovative AFE system has the potential to distinguish between neoplastic and non-neoplastic colorectal lesions. Patients and Methods: Patients with colorectal lesions more than 10 mm in diameter which underwent AFE and were treated endoscopically or surgically from September 2005 to October 2008 were included in this study. The red (R), green (G), and blue (B) values of recorded digital AFE images were analyzed with special color-analysis software to calculate two indexes - the autofluorescence index (G/R ratio) for each lesion, and the color contrast index (ΔE∗ab value) for each lesion and corresponding normal region. Based on the final pathological diagnosis, lesions were classified into four categories - submucosal invasive cancer, intramucosal cancer, tubular or villous adenoma, and hyperplastic lesion. Results: During the study period, a total of 168 lesions from 156 patients were examined by AFE. The lesions consisted of 119 neoplastic lesions, including 34 submucosal invasive cancers, 43 intramucosal cancers, and 42 adenomas, and 49 non-neoplastic lesions which were all hyperplastic lesions. The mean G/R ratio was significantly higher in non-neoplastic lesions (1.11 [95% CI, 1.07-1.15], n = 49) than neoplastic lesions (0.67 [95% CI, 0.65-0.70], n = 119; P < .001). The mean ΔE∗ab value was significantly lower in non-neoplastic lesions (8.82 [95% CI, 7.75-9.89], n = 49) than neoplastic lesions (34.15 [95% CI, 32.16-36.14], n = 119; P < .001). Also, the mean G/R ratio and ΔE∗ab value for adenoma (0.77 [95% CI, 0.73-0.82], and 22.95 [95% CI, 19.98-25.92] respectively) were significantly different to those of intramucosal cancer (0.64 [95% CI, 0.61-0.67], and 39.60 [95% CI, 37.69-41.52] respectively) and submucosal invasive cancer (0.60 [95% CI, 0.56-0.63], and 41.09 [95% CI, 39.17-43.00] respectively). Conclusions: The numerical color-tone analysis of digital AFE images obtained from the novel high-resolution videoendoscopy system revealed that autofluorescence intensity was significantly different between non-neoplastic and neoplastic lesions, and AFE color-tone may represent the histopathological characteristics of the lesion.