Endoscopic detection and differentiation of esophageal lesions using a deep neural network

Masayasu Ohmori, Ryu Ishihara, Kazuharu Aoyama, Kentaro Nakagawa, Hiroyoshi Iwagami, Noriko Matsuura, Satoki Shichijo, Katsumi Yamamoto, Koji Nagaike, Masanori Nakahara, Takuya Inoue, Kenji Aoi, Hiroyuki Okada, Tomohiro Tada

Research output: Contribution to journalArticle

Abstract

Background and Aims: Diagnosing esophageal squamous cell carcinoma (SCC) depends on individual physician expertise and may be subject to interobserver variability. Therefore, we developed a computerized image-analysis system to detect and differentiate esophageal SCC. Methods: A total of 9591 nonmagnified endoscopy (non-ME) and 7844 ME images of pathologically confirmed superficial esophageal SCCs and 1692 non-ME and 3435 ME images from noncancerous lesions or normal esophagus were used as training image data. Validation was performed using 255 non-ME white-light images, 268 non-ME narrow-band images/blue-laser images, and 204 ME narrow-band images/blue-laser images from 135 patients. The same validation test data were diagnosed by 15 board-certified specialists (experienced endoscopists). Results: Regarding diagnosis by non-ME with narrow-band imaging/blue-laser imaging, the sensitivity, specificity, and accuracy were 100%, 63%, and 77%, respectively, for the artificial intelligence (AI) system and 92%, 69%, and 78%, respectively, for the experienced endoscopists. Regarding diagnosis by non-ME with white-light imaging, the sensitivity, specificity, and accuracy were 90%, 76%, and 81%, respectively, for the AI system and 87%, 67%, and 75%, respectively, for the experienced endoscopists. Regarding diagnosis by ME, the sensitivity, specificity, and accuracy were 98%, 56%, and 77%, respectively, for the AI system and 83%, 70%, and 76%, respectively, for the experienced endoscopists. There was no significant difference in the diagnostic performance between the AI system and the experienced endoscopists. Conclusions: Our AI system showed high sensitivity for detecting SCC by non-ME and high accuracy for differentiating SCC from noncancerous lesions by ME.

Original languageEnglish
JournalGastrointestinal Endoscopy
DOIs
Publication statusAccepted/In press - Jan 1 2019

Fingerprint

Endoscopy
Artificial Intelligence
Lasers
Sensitivity and Specificity
Squamous Cell Carcinoma
Narrow Band Imaging
Photophobia
Observer Variation
Esophagus
Physicians
Light

ASJC Scopus subject areas

  • Radiology Nuclear Medicine and imaging
  • Gastroenterology

Cite this

Ohmori, M., Ishihara, R., Aoyama, K., Nakagawa, K., Iwagami, H., Matsuura, N., ... Tada, T. (Accepted/In press). Endoscopic detection and differentiation of esophageal lesions using a deep neural network. Gastrointestinal Endoscopy. https://doi.org/10.1016/j.gie.2019.09.034

Endoscopic detection and differentiation of esophageal lesions using a deep neural network. / Ohmori, Masayasu; Ishihara, Ryu; Aoyama, Kazuharu; Nakagawa, Kentaro; Iwagami, Hiroyoshi; Matsuura, Noriko; Shichijo, Satoki; Yamamoto, Katsumi; Nagaike, Koji; Nakahara, Masanori; Inoue, Takuya; Aoi, Kenji; Okada, Hiroyuki; Tada, Tomohiro.

In: Gastrointestinal Endoscopy, 01.01.2019.

Research output: Contribution to journalArticle

Ohmori, M, Ishihara, R, Aoyama, K, Nakagawa, K, Iwagami, H, Matsuura, N, Shichijo, S, Yamamoto, K, Nagaike, K, Nakahara, M, Inoue, T, Aoi, K, Okada, H & Tada, T 2019, 'Endoscopic detection and differentiation of esophageal lesions using a deep neural network', Gastrointestinal Endoscopy. https://doi.org/10.1016/j.gie.2019.09.034
Ohmori, Masayasu ; Ishihara, Ryu ; Aoyama, Kazuharu ; Nakagawa, Kentaro ; Iwagami, Hiroyoshi ; Matsuura, Noriko ; Shichijo, Satoki ; Yamamoto, Katsumi ; Nagaike, Koji ; Nakahara, Masanori ; Inoue, Takuya ; Aoi, Kenji ; Okada, Hiroyuki ; Tada, Tomohiro. / Endoscopic detection and differentiation of esophageal lesions using a deep neural network. In: Gastrointestinal Endoscopy. 2019.
@article{5d9acc6b1eaa48c5ba652b1a79629c9a,
title = "Endoscopic detection and differentiation of esophageal lesions using a deep neural network",
abstract = "Background and Aims: Diagnosing esophageal squamous cell carcinoma (SCC) depends on individual physician expertise and may be subject to interobserver variability. Therefore, we developed a computerized image-analysis system to detect and differentiate esophageal SCC. Methods: A total of 9591 nonmagnified endoscopy (non-ME) and 7844 ME images of pathologically confirmed superficial esophageal SCCs and 1692 non-ME and 3435 ME images from noncancerous lesions or normal esophagus were used as training image data. Validation was performed using 255 non-ME white-light images, 268 non-ME narrow-band images/blue-laser images, and 204 ME narrow-band images/blue-laser images from 135 patients. The same validation test data were diagnosed by 15 board-certified specialists (experienced endoscopists). Results: Regarding diagnosis by non-ME with narrow-band imaging/blue-laser imaging, the sensitivity, specificity, and accuracy were 100{\%}, 63{\%}, and 77{\%}, respectively, for the artificial intelligence (AI) system and 92{\%}, 69{\%}, and 78{\%}, respectively, for the experienced endoscopists. Regarding diagnosis by non-ME with white-light imaging, the sensitivity, specificity, and accuracy were 90{\%}, 76{\%}, and 81{\%}, respectively, for the AI system and 87{\%}, 67{\%}, and 75{\%}, respectively, for the experienced endoscopists. Regarding diagnosis by ME, the sensitivity, specificity, and accuracy were 98{\%}, 56{\%}, and 77{\%}, respectively, for the AI system and 83{\%}, 70{\%}, and 76{\%}, respectively, for the experienced endoscopists. There was no significant difference in the diagnostic performance between the AI system and the experienced endoscopists. Conclusions: Our AI system showed high sensitivity for detecting SCC by non-ME and high accuracy for differentiating SCC from noncancerous lesions by ME.",
author = "Masayasu Ohmori and Ryu Ishihara and Kazuharu Aoyama and Kentaro Nakagawa and Hiroyoshi Iwagami and Noriko Matsuura and Satoki Shichijo and Katsumi Yamamoto and Koji Nagaike and Masanori Nakahara and Takuya Inoue and Kenji Aoi and Hiroyuki Okada and Tomohiro Tada",
year = "2019",
month = "1",
day = "1",
doi = "10.1016/j.gie.2019.09.034",
language = "English",
journal = "Gastrointestinal Endoscopy",
issn = "0016-5107",
publisher = "Mosby Inc.",

}

TY - JOUR

T1 - Endoscopic detection and differentiation of esophageal lesions using a deep neural network

AU - Ohmori, Masayasu

AU - Ishihara, Ryu

AU - Aoyama, Kazuharu

AU - Nakagawa, Kentaro

AU - Iwagami, Hiroyoshi

AU - Matsuura, Noriko

AU - Shichijo, Satoki

AU - Yamamoto, Katsumi

AU - Nagaike, Koji

AU - Nakahara, Masanori

AU - Inoue, Takuya

AU - Aoi, Kenji

AU - Okada, Hiroyuki

AU - Tada, Tomohiro

PY - 2019/1/1

Y1 - 2019/1/1

N2 - Background and Aims: Diagnosing esophageal squamous cell carcinoma (SCC) depends on individual physician expertise and may be subject to interobserver variability. Therefore, we developed a computerized image-analysis system to detect and differentiate esophageal SCC. Methods: A total of 9591 nonmagnified endoscopy (non-ME) and 7844 ME images of pathologically confirmed superficial esophageal SCCs and 1692 non-ME and 3435 ME images from noncancerous lesions or normal esophagus were used as training image data. Validation was performed using 255 non-ME white-light images, 268 non-ME narrow-band images/blue-laser images, and 204 ME narrow-band images/blue-laser images from 135 patients. The same validation test data were diagnosed by 15 board-certified specialists (experienced endoscopists). Results: Regarding diagnosis by non-ME with narrow-band imaging/blue-laser imaging, the sensitivity, specificity, and accuracy were 100%, 63%, and 77%, respectively, for the artificial intelligence (AI) system and 92%, 69%, and 78%, respectively, for the experienced endoscopists. Regarding diagnosis by non-ME with white-light imaging, the sensitivity, specificity, and accuracy were 90%, 76%, and 81%, respectively, for the AI system and 87%, 67%, and 75%, respectively, for the experienced endoscopists. Regarding diagnosis by ME, the sensitivity, specificity, and accuracy were 98%, 56%, and 77%, respectively, for the AI system and 83%, 70%, and 76%, respectively, for the experienced endoscopists. There was no significant difference in the diagnostic performance between the AI system and the experienced endoscopists. Conclusions: Our AI system showed high sensitivity for detecting SCC by non-ME and high accuracy for differentiating SCC from noncancerous lesions by ME.

AB - Background and Aims: Diagnosing esophageal squamous cell carcinoma (SCC) depends on individual physician expertise and may be subject to interobserver variability. Therefore, we developed a computerized image-analysis system to detect and differentiate esophageal SCC. Methods: A total of 9591 nonmagnified endoscopy (non-ME) and 7844 ME images of pathologically confirmed superficial esophageal SCCs and 1692 non-ME and 3435 ME images from noncancerous lesions or normal esophagus were used as training image data. Validation was performed using 255 non-ME white-light images, 268 non-ME narrow-band images/blue-laser images, and 204 ME narrow-band images/blue-laser images from 135 patients. The same validation test data were diagnosed by 15 board-certified specialists (experienced endoscopists). Results: Regarding diagnosis by non-ME with narrow-band imaging/blue-laser imaging, the sensitivity, specificity, and accuracy were 100%, 63%, and 77%, respectively, for the artificial intelligence (AI) system and 92%, 69%, and 78%, respectively, for the experienced endoscopists. Regarding diagnosis by non-ME with white-light imaging, the sensitivity, specificity, and accuracy were 90%, 76%, and 81%, respectively, for the AI system and 87%, 67%, and 75%, respectively, for the experienced endoscopists. Regarding diagnosis by ME, the sensitivity, specificity, and accuracy were 98%, 56%, and 77%, respectively, for the AI system and 83%, 70%, and 76%, respectively, for the experienced endoscopists. There was no significant difference in the diagnostic performance between the AI system and the experienced endoscopists. Conclusions: Our AI system showed high sensitivity for detecting SCC by non-ME and high accuracy for differentiating SCC from noncancerous lesions by ME.

UR - http://www.scopus.com/inward/record.url?scp=85075999272&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85075999272&partnerID=8YFLogxK

U2 - 10.1016/j.gie.2019.09.034

DO - 10.1016/j.gie.2019.09.034

M3 - Article

C2 - 31585124

AN - SCOPUS:85075999272

JO - Gastrointestinal Endoscopy

JF - Gastrointestinal Endoscopy

SN - 0016-5107

ER -