Conference
Robust Federated Learning Methodology for Automatic Chest X-Ray Diagnosis
Medical Imaging with Deep Learning, 2024
Abstract
Chest X-rays are an important diagnostic tool, but their interpretation by radiologists can be complex and time consuming. Deep Learning (DL) was identified as a possible solution to this problem, but it requires large and well-annotated datasets. This requirement represents a significant limitation due to the difficulties in accessing patient data and ensuring their confidentiality. Federated Learning (FL) emerged as a promising alternative, as it allows hospitals to share local models while maintaining patient data within the hospital. However, the decisions made by the FL models are difficult to interpret and prevent radiologists from validating the results. In this paper we propose a methodology that adapts a lightweight version of the ConvNeXt architecture for automatic diagnosis of chest radiographs within an FL framework, evaluating the impact of adversarial attacks on model training. The evaluation showed that our …
Interactive explainer
Coming soon. An interactive Slidev walkthrough of this paper — with live, explorable figures — will appear here.
Authors: set
slidev: in this page's front matter to the deployed deck URL.Research themes
Lab authors
Cite (BibTeX)
@inproceedings{del2024robust,
author = {Del Cerro, Carlos Fernández and Abella, Mónica and Desco, Manuel and Albarqouni, Shadi},
booktitle = {Medical Imaging with Deep Learning},
title = {Robust Federated Learning Methodology for Automatic Chest X-Ray Diagnosis},
year = {2024}
}