Autonomous Lung Cancer Detection Using Federated Deep Learning: Advancing Scalable Diagnosis in India

Abstract

Lung cancer is still one of the main causes of cancer-related death in India, where inadequate data sharing across institutions, a lack of radiologists, and unequal access to imaging services hinder early detection. The autonomously lung cancer detection method proposed in this paper is based on federation deep learning techniques, allowing several healthcare facilities to work together to train models for diagnosis without sharing private patient information.The solution ensures complete data protection and ethical compliance by integrating automatic CT preprocessing, lung segmentation, nodules detection, and malignancy prediction inside a decentralised architecture. Each hospital uses a combination of transformer-enhanced classifiers and 3D convolutional networks for local training, and a secure aggregation server uses federated optimisation to update the global model.The federated technique greatly enhances model generalisation, lowers false positives, and nearly resembles the performance of centralised training, according to experimental results over heterogeneous, non-IID datasets. In order to assist earlier diagnosis and lessen radiologist burden, the suggested framework provides a scalable, privacy-preserving method for implementing artificial intelligence-driven lung cancer screening systems across various clinical environments in India.