Training a computer-aided polyp detection system to detect sessile serrated adenomas using public domain colonoscopy videos

Abstract

Background Colorectal cancer (CRC) is a major public health burden worldwide, and colonoscopy is the most commonly used CRC screening tool. Still, there is variability in adenoma detection rate (ADR) among endoscopists. Recent studies have reported improved ADR using deep learning models trained on videos curated largely from private in-house datasets. Few have focused on the detection of sessile serrated adenomas (SSAs), which are the most challenging target clinically.

Methods We identified 23 colonoscopy videos available in the public domain and for which pathology data were provided, totaling 390 minutes of footage. Expert endoscopists annotated segments of video with adenomatous polyps, from which we captured 509 polyp-positive and 6,875 polyp-free frames. Via data augmentation, we generated 15,270 adenomatous polyp-positive images, of which 2,310 were SSAs, and 20,625 polyp-negative images. We used the CNN AlexNet and fine-tuned its parameters using 90 % of the images, before testing its performance on the remaining 10 % of images unseen by the model.

Results We trained the model on 32,305 images and tested performance on 3,590 images with the same proportion of SSA, non-SSA polyp-positive, and polyp-negative images. The overall accuracy of the model was 0.86, with a sensitivity of 0.73 and a specificity of 0.96. Positive predictive value was 0.93 and negative predictive value was 0.96. The area under the curve was 0.94. SSAs were detected in 93 % of SSA-positive images.

Conclusions Using a relatively small set of publicly-available colonoscopy data, we obtained sizable training and validation sets of endoscopic images using data augmentation, and achieved an excellent performance in adenomatous polyp detection.

Publikationsverlauf

Artikel online veröffentlicht:
07. Oktober 2020

• References

• 1 Arnold M, Sierra MS, Laversanne M. et al. Global patterns and trends in colorectal cancer incidence and mortality. Gut 2017; 66: 683-691
• 2 Brody H. Colorectal cancer. Nature 2015; 521: S1
• 3 Center MM, Jemal A, Ward E. International trends in colorectal cancer incidence rates. Cancer Epidemiol Biomarkers Prev 2009; 18: 1688-1694
• 4 Kahi CJ. Imperiale TF, Juliar BE et al. Effect of screening colonoscopy on colorectal cancer incidence and mortality. Clin Gastroenterol Hepatol 2009; 7: 770-775
• 5 Pan J, Xin L, Ma Y-F. et al. Colonoscopy reduces colorectal cancer incidence and mortality in patients with non-malignant findings: a meta-analysis. Am J Gastroenterol 2016; 111: 355-365
• 6 Jacob BJ, Moineddin R, Sutradhar R. et al. Effect of colonoscopy on colorectal cancer incidence and mortality: an instrumental variable analysis. Gastrointest. Endosc 2012; 76: 355-64.e1
• 7 Ahn SB, Han DS, Bae JH. et al. The miss rate for colorectal adenoma determined by quality-adjusted: back-to-back colonoscopies. Gut Liver 2012; 6: 64-70
• 8 Corley DA, Levin TR, Doubeni CA. Adenoma detection rate and risk of colorectal cancer and death. N Engl J Med 2014; 370: 1298-1306
• 9 Wang C-L, Huang Z-P, Chen K. et al. Adenoma miss rate determined by very shortly repeated colonoscopy: Retrospective analysis of data from a single tertiary medical center in China. Medicine (Baltimore) 2018; 97: e12297
• 10 Rex DK, Ahnen DJ, Baron JA. et al. Serrated lesions of the colorectum: review and recommendations from an expert panel. Am J Gastroenterol 2012; 107: 1315-1329 ; quiz 1314
• 11 Bouwens MW, van Herwaarden YJ, Winkens B. et al. Endoscopic characterization of sessile serrated adenomas/polyps with and without dysplasia. Endoscopy 2014; 46: 225-235
• 12 Ma MX, Bourke MJ. Sessile serrated adenomas: how to detect: characterize and resect. Gut Liver 2017; 11: 747-760
• 13 Bettington M, Walker N, Rosty C. et al. Critical appraisal of the diagnosis of the sessile serrated adenoma. Am J Surg Pathol 2014; 38: 158-166
• 14 Benedict M, Galvao Neto A, Zhang X. Interval colorectal carcinoma: An unsolved debate. World J. Gastroenterol 2014; 21: 12735-12741
• 15 Alagappan M, Brown JRG, Mori Y. et al. Artificial intelligence in gastrointestinal endoscopy: The future is almost here. World J Gastrointest Endosc 2018; 10: 239-249
• 16 Topol EJ. High-performance medicine: the convergence of human and artificial intelligence. Nat Med 2019; 25: 44-56
• 17 Chen P-J, Lin M-C, Lai M-J. et al. Accurate Classification of diminutive colorectal polyps using computer-aided analysis. Gastroenterology 2018; 154: 568-575
• 18 Misawa M, Kudo S-E, Mori Y. et al. Artificial intelligence-assisted polyp detection for colonoscopy: initial experience. Gastroenterology 2018; 154: 2027-2029.e3
• 19 Wang P, Berzin TM, Glissen Brown JR. et al. Real-time automatic detection system increases colonoscopic polyp and adenoma detection rates: a prospective randomised controlled study. Gut 2019; 68: 1813-1819
• 20 Krizhevsky A, Sutskever I, Hinton GE. ImageNet classification with deep convolutional neural networks. In: Pereira F, Burges CJ, Bottou L. et al. (eds.) Advances in Neural Information Processing Systems 25. Curran Associates: Inc; 2012: 1097-1105
• 21 Perez L, Wang J. The effectiveness of data augmentation in image classification using deep learning. arXiv:1712.04621 [cs.CV]. 2017
• 22 Huertas A, Medioni G. Detection of intensity changes with subpixel accuracy using Laplacian-Gaussian masks. IEEE Trans. Pattern Anal Mach Intell PAMI 1986; 8: 651-664
• 23 Jia Y, Shelhamer E, Donahue J. et al. Caffe: convolutional architecture for fast feature embedding. arXiv:1408.5093 [cs.CV]. 2014
• 24 Obuch JC, Pigott CM, Ahnen DJ. Sessile serrated polyps: detection: eradication: and prevention of the evil twin. Curr Treat Options Gastroenterol 2015; 13: 156-170
• 25 Kim NH, Jung YS, Jeong WS. et al. Miss rate of colorectal neoplastic polyps and risk factors for missed polyps in consecutive colonoscopies. Intest Res 2017; 15: 411-418
• 26 Rajaratnam R, Purchiaroni F, Wilson A. et al. Mo1075 high levels of presumed polyp miss rate at 1 and 3 years following index screening colonoscopy: no room for complacency. Gastrointest Endosc 2017; 85: AB417
• 27 Mori Y, Kudo S-E, Misawa M. et al. Real-Time use of artificial intelligence in identification of diminutive polyps during colonoscopy: a prospective study. Ann Intern Med 2018; 169: 357-366