Detecting Lung Cancer with Federated and Transfer Learning

Lung cancer is a disease that affects and causes abnormalities in the lungs. The current methods to find and treat lung cancer require precise and timely detection to improve patient outcomes and survival rates. However, traditional approaches to lung cancer detection face challenges due to the extensive patient information spread across different medical institutions and research centers. Concerns about data privacy and the need to keep data controlled have prevented the consolidation of this valuable data into a central repository for analysis. As a result, the development of effective and accurate detection models has been limited by restricted access to diverse and comprehensive datasets. To overcome these challenges, federated learning (FL) has emerged as a promising approach in the healthcare field. It has great potential in the healthcare or medical sector where we can get a better Machine learning model while ensuring patient data privacy. This paper presents an FL approach for detecting lung cancer in medical images after setting an initial weight using Transfer Learning. Using this approach, we achieved a break a hand accuracy of 91.03% in detecting lung cancer. This demonstrates the potential of FL for accurate and privacy-preserving medical diagnosis.