Federated brain tumor segmentation: an extensive benchmark

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Manthe, Matthis, Duffner, Stefan, Lartizien, Carole
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866914984441675776
author Manthe, Matthis
Duffner, Stefan
Lartizien, Carole
author_facet Manthe, Matthis
Duffner, Stefan
Lartizien, Carole
contents Recently, federated learning has raised increasing interest in the medical image analysis field due to its ability to aggregate multi-center data with privacy-preserving properties. A large amount of federated training schemes have been published, which we categorize into global (one final model), personalized (one model per institution) or hybrid (one model per cluster of institutions) methods. However, their applicability on the recently published Federated Brain Tumor Segmentation 2022 dataset has not been explored yet. We propose an extensive benchmark of federated learning algorithms from all three classes on this task. While standard FedAvg already performs very well, we show that some methods from each category can bring a slight performance improvement and potentially limit the final model(s) bias toward the predominant data distribution of the federation. Moreover, we provide a deeper understanding of the behaviour of federated learning on this task through alternative ways of distributing the pooled dataset among institutions, namely an Independent and Identical Distributed (IID) setup, and a limited data setup.
format Preprint
id arxiv_https___arxiv_org_abs_2410_17265
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Federated brain tumor segmentation: an extensive benchmark
Manthe, Matthis
Duffner, Stefan
Lartizien, Carole
Computer Vision and Pattern Recognition
Artificial Intelligence
Machine Learning
Image and Video Processing
Recently, federated learning has raised increasing interest in the medical image analysis field due to its ability to aggregate multi-center data with privacy-preserving properties. A large amount of federated training schemes have been published, which we categorize into global (one final model), personalized (one model per institution) or hybrid (one model per cluster of institutions) methods. However, their applicability on the recently published Federated Brain Tumor Segmentation 2022 dataset has not been explored yet. We propose an extensive benchmark of federated learning algorithms from all three classes on this task. While standard FedAvg already performs very well, we show that some methods from each category can bring a slight performance improvement and potentially limit the final model(s) bias toward the predominant data distribution of the federation. Moreover, we provide a deeper understanding of the behaviour of federated learning on this task through alternative ways of distributing the pooled dataset among institutions, namely an Independent and Identical Distributed (IID) setup, and a limited data setup.
title Federated brain tumor segmentation: an extensive benchmark
topic Computer Vision and Pattern Recognition
Artificial Intelligence
Machine Learning
Image and Video Processing
url https://arxiv.org/abs/2410.17265