Reducing segmentation failures in cardiac MRI via late feature fusion and GAN-based augmentation
Computers in Biology and Medicine · 2023-07 · 161 · 106973
DOI: 10.1016/j.compbiomed.2023.106973 ↗Kurz gefasst
Fachartikel in der Zeitschrift Computers in Biology and Medicine, 2023-07. Verfasst von Yasmina Al Khalil und 5 weiteren Autor:innen. Beteiligte Einrichtung: Department of Biomedical Engineering, Eindhoven University of Technology, Eindhoven, The Netherlands. Electronic address: y.al.khalil@tue.nl. Hauptthemen laut MEDLINE: Algorithms, Heart Diseases. Die Arbeit wurde bisher 34-mal zitiert (Crossref) und nennt 75 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: European Commission, Horizon 2020.
Abstract (Originalsprache)
Cardiac magnetic resonance (CMR) image segmentation is an integral step in the analysis of cardiac function and diagnosis of heart related diseases. While recent deep learning-based approaches in automatic segmentation have shown great promise to alleviate the need for manual segmentation, most of these are not applicable to realistic clinical scenarios. This is largely due to training on mainly homogeneous datasets, without variation in acquisition, which typically occurs in multi-vendor and multi-site settings, as well as pathological data. Such approaches frequently exhibit a degradation in prediction performance, particularly on outlier cases commonly associated with difficult pathologies, artifacts and extensive changes in tissue shape and appearance. In this work, we present a model aimed at segmenting all three cardiac structures in a multi-center, multi-disease and multi-view scenario. We propose a pipeline, addressing different challenges with segmentation of such heterogeneous data, consisting of heart region detection, augmentation through image synthesis and a late-fusion segmentation approach. Extensive experiments and analysis demonstrate the ability of the proposed approach to tackle the presence of outlier cases during both training and testing, allowing for better adaptation to unseen and difficult examples. Overall, we show that the effective reduction of segmentation failures on outlier cases has a positive impact on not only the average segmentation performance, but also on the estimation of clinical parameters, leading to a better consistency in derived metrics.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Department of Biomedical Engineering, Eindhoven University of Technology, Eindhoven, The Netherlands. Electronic address: y.al.khalil@tue.nl.
- Department of Biomedical Engineering, Eindhoven University of Technology, Eindhoven, The Netherlands.
- Philips Research Laboratories, Hamburg, Germany.
- Department of Biomedical Engineering, Eindhoven University of Technology, Eindhoven, The Netherlands; Philips Healthcare, MR R&D - Clinical Science, Best, The Netherlands.
Förderung
- European Commission
- Horizon 2020
Zitieren
APA
Al Khalil, Y., Amirrajab, S., Lorenz, C., Weese, J., Pluim, J., & Breeuwer, M. (2023). Reducing segmentation failures in cardiac MRI via late feature fusion and GAN-based augmentation. Computers in Biology and Medicine, 161, 106973. https://doi.org/10.1016/j.compbiomed.2023.106973
Vancouver
Al Khalil Y, Amirrajab S, Lorenz C, Weese J, Pluim J, Breeuwer M. Reducing segmentation failures in cardiac MRI via late feature fusion and GAN-based augmentation. Computers in Biology and Medicine. 2023;161:106973. doi:10.1016/j.compbiomed.2023.106973
BibTeX
@article{AlKhalil2023,
author = {Al Khalil, Yasmina and Amirrajab, Sina and Lorenz, Cristian and Weese, Jürgen and Pluim, Josien and Breeuwer, Marcel},
title = {Reducing segmentation failures in cardiac MRI via late feature fusion and GAN-based augmentation},
journal = {Computers in Biology and Medicine},
year = {2023},
volume = {161},
pages = {106973},
doi = {10.1016/j.compbiomed.2023.106973}
}
EUVIMED