Development and validation of a deep-learning algorithm for the detection of polyps during colonoscopy

There is no author summary for this article yet. Authors can add summaries to their articles on ScienceOpen to make them more accessible to a non-specialist audience.

Abstract

The detection and removal of precancerous polyps via colonoscopy is the gold standard for the prevention of colon cancer. However, the detection rate of adenomatous polyps can vary significantly among endoscopists. Here, we show that a machine-learning algorithm can detect polyps in clinical colonoscopies, in real time and with high sensitivity and specificity. We developed the deep-learning algorithm by using data from 1,290 patients, and validated it on newly collected 27,113 colonoscopy images from 1,138 patients with at least one detected polyp (per-image-sensitivity, 94.38%; per-image-specificity, 95.92%; area under the receiver operating characteristic curve, 0.984), on a public database of 612 polyp-containing images (per-image-sensitivity, 88.24%), on 138 colonoscopy videos with histologically confirmed polyps (per-image-sensitivity of 91.64%; per-polyp-sensitivity, 100%), and on 54 unaltered full-range colonoscopy videos without polyps (per-image-specificity, 95.40%). By using a multi-threaded processing system, the algorithm can process at least 25 frames per second with a latency of 76.80 ± 5.60 ms in real-time video analysis. The software may aid endoscopists while performing colonoscopies, and help assess differences in polyp and adenoma detection performance among endoscopists.

Related collections

Most cited references 22

Record: found
Abstract: found
Article: not found

Protection from colorectal cancer after colonoscopy: a population-based, case-control study.

Hermann Brenner, Jenny Chang-Claude, Christoph Seiler … (2011)

Colonoscopy with detection and removal of adenomas is considered a powerful tool to reduce colorectal cancer (CRC) incidence. However, the degree of protection achievable in a population setting with high-quality colonoscopy resources remains to be quantified. To assess the association between previous colonoscopy and risk for CRC. Population-based case-control study. Rhine-Neckar region of Germany. A total of 1688 case patients with colorectal cancer and 1932 control participants aged 50 years or older. A detailed lifetime history of CRC risk factors and preventive factors, including history and results of previous colonoscopies, and of medical data obtained by self-reports and medical records. Odds ratios of CRC associated with colonoscopy in the preceding 10 years were estimated, after adjustment for sex, age, education level, participation in a general health screening examination, family history of CRC, smoking status, body mass index, and use of nonsteroidal anti-inflammatory drugs or hormone replacement therapy. Overall, colonoscopy in the preceding 10 years was associated with 77% lower risk for CRC. Adjusted odds ratios for any CRC, right-sided CRC, and left-sided CRC were 0.23 (95% CI, 0.19 to 0.27), 0.44 (CI, 0.35 to 0.55), and 0.16 (CI, 0.12 to 0.20), respectively. Strong risk reduction was observed for all cancer stages and all ages, except for right-sided cancer in persons aged 50 to 59 years. Risk reduction increased over the years in both the right and the left colon. The study was observational, with potential for residual confounding and selection bias. Colonoscopy with polypectomy can be associated with strongly reduced risk for CRC in the population setting. Aside from strong risk reduction with respect to left-sided CRC, risk reduction of more than 50% was also seen for right-sided colon cancer. German Research Council and German Federal Ministry of Education and Research.