A neural 
network algorithm for detection of GI angiectasia during 
small-bowel capsule endoscopy - Archive ouverte HAL Accéder directement au contenu
Article Dans Une Revue Gastrointestinal Endoscopy Année : 2019

A neural 
network algorithm for detection of GI angiectasia during 
small-bowel capsule endoscopy

Pauline Vasseur
  • Fonction : Auteur
  • PersonId : 1297589
Cynthia Li
  • Fonction : Auteur
Franck Cholet
  • Fonction : Auteur
  • PersonId : 929637
Jean-Christophe Saurin
  • Fonction : Auteur
  • PersonId : 869009
Xavier Amiot
  • Fonction : Auteur
Michel Delvaux
  • Fonction : Auteur
Clotilde Duburque
  • Fonction : Auteur
Geoffroy Vanbiervliet
  • Fonction : Auteur
  • PersonId : 887034
Romain Gerard
  • Fonction : Auteur
Jean-Philippe Le Mouel
  • Fonction : Auteur
Chloé Leandri
  • Fonction : Auteur
Stéphane Lecleire
  • Fonction : Auteur
  • PersonId : 895636
Farida Mesli
  • Fonction : Auteur
Isabelle Nion-Larmurier
  • Fonction : Auteur
Sylvie Sacher-Huvelin
  • Fonction : Auteur
  • PersonId : 895626
Philippe Marteau
  • Fonction : Auteur
  • PersonId : 914789
Camille Chane Simon
Pierre Jacob
Olivier Romain

Résumé

Background and Aims Gastrointestinal angiectasia (GIA) is the most common small bowel (SB) vascular lesion, with an inherent risk of bleeding. SB Capsule Endoscopy (SB-CE) is the currently accepted diagnostic procedure. The aim of this study was to develop a computer-assisted diagnosis (CAD) tool for the detection of GIA. Methods Deidentified SB-CE still frames featuring annotated typical GIA and normal control still frames, were selected from a database. A semantic segmentation images approach associated with a convolutional neural network (CNN) was used for deep feature extractions and classification. Two datasets of still frames were created and used for machine-learning and for algorithm testing. Results The GIA detection algorithm yielded a sensitivity of 100%, a specificity of 96%, a positive predictive value of 96%, and a negative predictive value of 100%. Reproducibility was optimal. The reading process for an entire SB-CE video would take 2340 seconds (39 minutes). Conclusion The developed CNN-based algorithm had high diagnostic performances allowing detection of GIA in SB-CE still frames. This study paves the way for future automated CNN-based SB-CE reading softwares.
Fichier non déposé

Dates et versions

hal-01835422 , version 1 (11-07-2018)

Identifiants

Citer

Romain Leenhardt, Pauline Vasseur, Cynthia Li, Gabriel Rahmi, Franck Cholet, et al.. A neural 
network algorithm for detection of GI angiectasia during 
small-bowel capsule endoscopy. Gastrointestinal Endoscopy, 2019, 89 (1), pp.189-194. ⟨10.1016/j.gie.2018.06.036⟩. ⟨hal-01835422⟩
252 Consultations
0 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More