The Comprehensive Investigation of Federated Learning with Its Application in the Medical Image Analysis

Wenxiao Zeng

2024

Abstract

Due to the pandemic in 2019 and the causing medical system crush, it has become necessary for the nations to increase the system capacity of health care services. By combining with machine learning, the newly developed automatic diagnosis can largely save the manpower and time cost required for the current medical systems, thus increase their overall efficiencies. The Federated Learning algorithm, based on the background of rising demand for automatic diagnosis and data privacy, is becoming widely-applied in the medical diagnosis, especially in the sophisticated medical image analysis. This study overviews the current researches of the Federated Learning algorithms in the health care system, including the diagnosis prediction of the Federated Learning-based models towards three specific types of diseases. The study discusses the advantages of Federated Learning in data privacy and heterogeneous massive data processing by the architecture of its workflow. On the other hand, its potential drawbacks are the lack of interpretability and applicability, which can possibly be solved or improved by SHapley Addictive exPlanation (SHAP) algorithm and Dynamic Weighting Translation Transfer Learning (DTTL) algorithm. Its potential safety issue by data transmission, however, though being minimized by the decentralized computation architecture of the Federated Learning, can hardly be fully removed unless the fully distributed algorithm will have developed and replaced its application in the future.

Paper Citation

in Harvard Style

Zeng W. (2024). The Comprehensive Investigation of Federated Learning with Its Application in the Medical Image Analysis. In Proceedings of the 2nd International Conference on Data Analysis and Machine Learning - Volume 1: DAML; ISBN 978-989-758-754-2, SciTePress, pages 532-536. DOI: 10.5220/0013527800004619

in Bibtex Style

@conference{daml24,
author={Wenxiao Zeng},
title={The Comprehensive Investigation of Federated Learning with Its Application in the Medical Image Analysis},
booktitle={Proceedings of the 2nd International Conference on Data Analysis and Machine Learning - Volume 1: DAML},
year={2024},
pages={532-536},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0013527800004619},
isbn={978-989-758-754-2},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 2nd International Conference on Data Analysis and Machine Learning - Volume 1: DAML
TI - The Comprehensive Investigation of Federated Learning with Its Application in the Medical Image Analysis
SN - 978-989-758-754-2
AU - Zeng W.
PY - 2024
SP - 532
EP - 536
DO - 10.5220/0013527800004619
PB - SciTePress