ABNORMAL DETECTION OF CHEST X- RAY IMAGES USING FEDERATED LEARNING

Authors

  • Hafidzul Abdullah Computational and Theoretical Sciences, International Islamic University Malaysia, Malaysia

Abstract

Nowadays, data has become a valuable asset to the world. Even though the values of these data can help people to analyse information with accuracy and precision, it may lead to privacy concerns for sensitive information such as patient's chest X-rays (CXR) images. This study is focusing on maintaining the privacy of users by applying the Federated Learning system using the TensorFlow library as a method of developing machine learning algorithms that can detect any abnormalities in the CXR images without requiring any client to share their CXR images outside of the hospitals. This method showed a better score in the precision-recall curve graph compared to the conventional methods that used locally trained machine learning.

Acknowledgement: I would like to express my very great appreciation to my supervisor, Asst. Prof. Dr. Mohd Adli Bin Md. Ali and my co-supervisor, Assoc. Prof. Dr. Mohd. Zulfaezal Bin Che Azemin for guiding me in this research.

Downloads

Download data is not yet available.

Downloads

Published

2021-04-08

How to Cite

Abdullah, H. . (2021). ABNORMAL DETECTION OF CHEST X- RAY IMAGES USING FEDERATED LEARNING . INTERNATIONAL JOURNAL OF ALLIED HEALTH SCIENCES, 5(1), 2111. Retrieved from https://journals.iium.edu.my/ijahs/index.php/IJAHS/article/view/548