Evaluation of Deep Convolutional Neural Networks for Automated Detection of Lung Nodules in CT Scans Across Multi-Institutional Radiology Datasets

Abstract

Early detection of lung nodules in computed tomography (CT) scans significantly enhances patient outcomes in lung cancer diagnosis. Deep Convolutional Neural Networks (CNNs) have emerged as a promising approach for automating nodule detection with high accuracy. This study evaluates the performance of multiple CNN architectures across heterogeneous, multi-institutional CT scan datasets. We aim to understand the generalizability of these models across varied clinical environments and data sources. Using metrics such as sensitivity, specificity, and the area under the receiver operating characteristic curve (AUC), we assess model robustness and effectiveness. The findings suggest that while CNNs offer high diagnostic potential, their performance varies with data heterogeneity and institutional differences, indicating a need for domain adaptation techniques and standardized imaging protocols.