Real-time differentiation of adenomatous and hyperplastic diminutive colorectal polyps during analysis of unaltered videos of standard colonoscopy using a deep learning model

Michael F. Byrne, Nicolas Chapados, Florian Soudan, Clemens Oertel, Milagros Linares Pérez, Raymond Kelly, Nadeem Iqbal, Florent Chandelier, Douglas Rex

Research output: Contribution to journalArticle

69 Citations (Scopus)

Abstract

Background: In general, academic but not community endoscopists have demonstrated adequate endoscopic differentiation accuracy to make the € resect and discard' paradigm for diminutive colorectal polyps workable. Computer analysis of video could potentially eliminate the obstacle of interobserver variability in endoscopic polyp interpretation and enable widespread acceptance of € resect and discard'. Study design and methods: We developed an artificial intelligence (AI) model for real-time assessment of endoscopic video images of colorectal polyps. A deep convolutional neural network model was used. Only narrow band imaging video frames were used, split equally between relevant multiclasses. Unaltered videos from routine exams not specifically designed or adapted for AI classification were used to train and validate the model. The model was tested on a separate series of 125 videos of consecutively encountered diminutive polyps that were proven to be adenomas or hyperplastic polyps. Results: The AI model works with a confidence mechanism and did not generate sufficient confidence to predict the histology of 19 polyps in the test set, representing 15% of the polyps. For the remaining 106 diminutive polyps, the accuracy of the model was 94% (95% CI 86% to 97%), the sensitivity for identification of adenomas was 98% (95% CI 92% to 100%), specificity was 83% (95% CI 67% to 93%), negative predictive value 97% and positive predictive value 90%. Conclusions: An AI model trained on endoscopic video can differentiate diminutive adenomas from hyperplastic polyps with high accuracy. Additional study of this programme in a live patient clinical trial setting to address resect and discard is planned.

Original languageEnglish (US)
JournalGut
DOIs
StateAccepted/In press - Nov 9 2017

Fingerprint

Colonoscopy
Polyps
Learning
Artificial Intelligence
Adenoma
Narrow Band Imaging
Neural Networks (Computer)
Observer Variation
Histology
Clinical Trials

Keywords

  • colorectal adenomas
  • endoscopic polypectomy
  • polyp

ASJC Scopus subject areas

  • Gastroenterology

Cite this

Real-time differentiation of adenomatous and hyperplastic diminutive colorectal polyps during analysis of unaltered videos of standard colonoscopy using a deep learning model. / Byrne, Michael F.; Chapados, Nicolas; Soudan, Florian; Oertel, Clemens; Linares Pérez, Milagros; Kelly, Raymond; Iqbal, Nadeem; Chandelier, Florent; Rex, Douglas.

In: Gut, 09.11.2017.

Research output: Contribution to journalArticle

Byrne, Michael F. ; Chapados, Nicolas ; Soudan, Florian ; Oertel, Clemens ; Linares Pérez, Milagros ; Kelly, Raymond ; Iqbal, Nadeem ; Chandelier, Florent ; Rex, Douglas. / Real-time differentiation of adenomatous and hyperplastic diminutive colorectal polyps during analysis of unaltered videos of standard colonoscopy using a deep learning model. In: Gut. 2017.
@article{84aa251722c745c897aebb988918c1b3,
title = "Real-time differentiation of adenomatous and hyperplastic diminutive colorectal polyps during analysis of unaltered videos of standard colonoscopy using a deep learning model",
abstract = "Background: In general, academic but not community endoscopists have demonstrated adequate endoscopic differentiation accuracy to make the € resect and discard' paradigm for diminutive colorectal polyps workable. Computer analysis of video could potentially eliminate the obstacle of interobserver variability in endoscopic polyp interpretation and enable widespread acceptance of € resect and discard'. Study design and methods: We developed an artificial intelligence (AI) model for real-time assessment of endoscopic video images of colorectal polyps. A deep convolutional neural network model was used. Only narrow band imaging video frames were used, split equally between relevant multiclasses. Unaltered videos from routine exams not specifically designed or adapted for AI classification were used to train and validate the model. The model was tested on a separate series of 125 videos of consecutively encountered diminutive polyps that were proven to be adenomas or hyperplastic polyps. Results: The AI model works with a confidence mechanism and did not generate sufficient confidence to predict the histology of 19 polyps in the test set, representing 15{\%} of the polyps. For the remaining 106 diminutive polyps, the accuracy of the model was 94{\%} (95{\%} CI 86{\%} to 97{\%}), the sensitivity for identification of adenomas was 98{\%} (95{\%} CI 92{\%} to 100{\%}), specificity was 83{\%} (95{\%} CI 67{\%} to 93{\%}), negative predictive value 97{\%} and positive predictive value 90{\%}. Conclusions: An AI model trained on endoscopic video can differentiate diminutive adenomas from hyperplastic polyps with high accuracy. Additional study of this programme in a live patient clinical trial setting to address resect and discard is planned.",
keywords = "colorectal adenomas, endoscopic polypectomy, polyp",
author = "Byrne, {Michael F.} and Nicolas Chapados and Florian Soudan and Clemens Oertel and {Linares P{\'e}rez}, Milagros and Raymond Kelly and Nadeem Iqbal and Florent Chandelier and Douglas Rex",
year = "2017",
month = "11",
day = "9",
doi = "10.1136/gutjnl-2017-314547",
language = "English (US)",
journal = "Gut",
issn = "0017-5749",
publisher = "BMJ Publishing Group",

}

TY - JOUR

T1 - Real-time differentiation of adenomatous and hyperplastic diminutive colorectal polyps during analysis of unaltered videos of standard colonoscopy using a deep learning model

AU - Byrne, Michael F.

AU - Chapados, Nicolas

AU - Soudan, Florian

AU - Oertel, Clemens

AU - Linares Pérez, Milagros

AU - Kelly, Raymond

AU - Iqbal, Nadeem

AU - Chandelier, Florent

AU - Rex, Douglas

PY - 2017/11/9

Y1 - 2017/11/9

N2 - Background: In general, academic but not community endoscopists have demonstrated adequate endoscopic differentiation accuracy to make the € resect and discard' paradigm for diminutive colorectal polyps workable. Computer analysis of video could potentially eliminate the obstacle of interobserver variability in endoscopic polyp interpretation and enable widespread acceptance of € resect and discard'. Study design and methods: We developed an artificial intelligence (AI) model for real-time assessment of endoscopic video images of colorectal polyps. A deep convolutional neural network model was used. Only narrow band imaging video frames were used, split equally between relevant multiclasses. Unaltered videos from routine exams not specifically designed or adapted for AI classification were used to train and validate the model. The model was tested on a separate series of 125 videos of consecutively encountered diminutive polyps that were proven to be adenomas or hyperplastic polyps. Results: The AI model works with a confidence mechanism and did not generate sufficient confidence to predict the histology of 19 polyps in the test set, representing 15% of the polyps. For the remaining 106 diminutive polyps, the accuracy of the model was 94% (95% CI 86% to 97%), the sensitivity for identification of adenomas was 98% (95% CI 92% to 100%), specificity was 83% (95% CI 67% to 93%), negative predictive value 97% and positive predictive value 90%. Conclusions: An AI model trained on endoscopic video can differentiate diminutive adenomas from hyperplastic polyps with high accuracy. Additional study of this programme in a live patient clinical trial setting to address resect and discard is planned.

AB - Background: In general, academic but not community endoscopists have demonstrated adequate endoscopic differentiation accuracy to make the € resect and discard' paradigm for diminutive colorectal polyps workable. Computer analysis of video could potentially eliminate the obstacle of interobserver variability in endoscopic polyp interpretation and enable widespread acceptance of € resect and discard'. Study design and methods: We developed an artificial intelligence (AI) model for real-time assessment of endoscopic video images of colorectal polyps. A deep convolutional neural network model was used. Only narrow band imaging video frames were used, split equally between relevant multiclasses. Unaltered videos from routine exams not specifically designed or adapted for AI classification were used to train and validate the model. The model was tested on a separate series of 125 videos of consecutively encountered diminutive polyps that were proven to be adenomas or hyperplastic polyps. Results: The AI model works with a confidence mechanism and did not generate sufficient confidence to predict the histology of 19 polyps in the test set, representing 15% of the polyps. For the remaining 106 diminutive polyps, the accuracy of the model was 94% (95% CI 86% to 97%), the sensitivity for identification of adenomas was 98% (95% CI 92% to 100%), specificity was 83% (95% CI 67% to 93%), negative predictive value 97% and positive predictive value 90%. Conclusions: An AI model trained on endoscopic video can differentiate diminutive adenomas from hyperplastic polyps with high accuracy. Additional study of this programme in a live patient clinical trial setting to address resect and discard is planned.

KW - colorectal adenomas

KW - endoscopic polypectomy

KW - polyp

UR - http://www.scopus.com/inward/record.url?scp=85037665458&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85037665458&partnerID=8YFLogxK

U2 - 10.1136/gutjnl-2017-314547

DO - 10.1136/gutjnl-2017-314547

M3 - Article

C2 - 29066576

AN - SCOPUS:85037665458

JO - Gut

JF - Gut

SN - 0017-5749

ER -