A transfer learning-based computer-aided lung cancer detection system in smart healthcare

Rapid urbanization has resulted in the emergence of smart cities, incorporating technology across various sectors, such as healthcare, to improve the quality of life and increase life expectancy. Lung cancer, a leading cause of cancer-related deaths, requires innovative diagnostic approaches due to the limitations of existing screening methods. Despite progress in medical imaging, accurately identifying lung cancer at its earliest stages is still a clinical challenge. To address this, we proposed a novel transfer learning-based system that uses CT scans to enhance lung cancer detection. Our research aims to evaluate the system’s performance in improving diagnostic accuracy, particularly early-stage detection, and assess its potential clinical impact. To achieve this, we collected data and preprocessed images, including 1097 images of CT scans, encompassing three categories, scanner types, and imaging protocols. Feature extraction techniques focused on capturing texture, shape, intensity, and spatial relationships to support model training and prediction. The transfer learning models adapted knowledge from a large-scale pretrained dataset, effectively identifying complex patterns and outperforming traditional methods. Specifically, our Xception model achieved the highest accuracy and f1 score of 98.32% and 97.96%, respectively. Additionally, we developed a user-friendly web application based on the Xception model for practical implementation. However, further research is needed to address dataset diversity and realworld validation challenges. This study contributes to AI-driven medical imaging, presenting a promising solution to enhance lung cancer screening and patient outcomes within smart healthcare systems.