Paper published in a book (Scientific congresses and symposiums)
Relieving pixel-wise labeling effort for pathology image segmentation with self-training
Mormont, Romain; Testouri, Mehdi; Marée, Raphaël et al.
2022In Lecture Notes in Computer Science
Peer reviewed
 

Files


Full Text
070.pdf
Author preprint (988.76 kB) Creative Commons License - Attribution, ShareAlike
Download
Annexes
070-supp.pdf
(12.43 MB) Creative Commons License - Attribution, ShareAlike
Download

All documents in ORBi are protected by a user license.

Send to



Details



Keywords :
deep learning, image segmentation, self-training, data scarcity, digital pathology
Abstract :
[en] Data scarcity is a common issue when training deep learning models for digital pathology, as large exhaustively-annotated image datasets are difficult to obtain. In this paper, we propose a self-training based approach that can exploit both (few) exhaustively annotated images and (very) sparsely-annotated images to improve the training of deep learning models for image segmentation tasks. The approach is evaluated on three public and one in-house datasets, representing a diverse set of segmentation tasks in digital pathology. The experimental results show that self-training allows to bring significant model improvement by incorporating sparsely annotated images and proves to be a good strategy to relieve labeling effort in the digital pathology domain.
Disciplines :
Electrical & electronics engineering
Author, co-author :
Mormont, Romain  ;  Université de Liège - ULiège > Montefiore Institute of Electrical Engineering and Computer Science
Testouri, Mehdi;  Unilu - Université du Luxembourg [LU] > Interdisciplinary Centre for Security, Reliability and Trust > Services and Data Management (SEDAN)
Marée, Raphaël  ;  Université de Liège - ULiège > Montefiore Institute of Electrical Engineering and Computer Science
Geurts, Pierre ;  Université de Liège - ULiège > Montefiore Institute of Electrical Engineering and Computer Science
Language :
English
Title :
Relieving pixel-wise labeling effort for pathology image segmentation with self-training
Alternative titles :
[fr] Utilisation de l'auto-apprentissage pour réduire le coût d'annotation pour la segmentation d'image en pathology digitale
Publication date :
October 2022
Event name :
European Conference in Compter Vision (ECCV2022)
Event organizer :
European Computer Vision Association
Event place :
Tel Aviv, Israel
Event date :
October 23, 2022
Audience :
International
Main work title :
Lecture Notes in Computer Science
Publisher :
Springer Cham, Genève, Switzerland
Peer reviewed :
Peer reviewed
Available on ORBi :
since 17 August 2022

Statistics


Number of views
180 (33 by ULiège)
Number of downloads
213 (21 by ULiège)

Scopus citations®
 
0
Scopus citations®
without self-citations
0
OpenCitations
 
0

Bibliography


Similar publications



Contact ORBi