Design of Convolutional Neural Network for Lung Cancer Diagnosis

Abstract

Computed Tomography(CT) lung screening is becoming more commonly used around the world. However, radiologists face a significant challenge in processing these videos. In a CT scan, there can be up to 600 slices. As a result, computer-aided-detection (CAD) systems are critical for a faster and more reliable data analysis. To minimize false positives, a system for CT lung screenings using convolutional neural networks (CNNs) is proposed in this study. When compared to conventional methods, the framework shows that even simple architectures are very effective at classifying 3D nodule data. We tested our model with various volume sizes and found that volume size has a significant impact on efficiency.