Automated Detection of Lung Nodules in CT Images with 3D Convolutional Neural Networks

Accurate lung nodule detection in computed tomography (CT) images is a critical step in diagnosing lung cancer. This paper proposes an efficient model to build a Computer-Aided Detection (CADe) system for lung nodule detection, which points out a way to enhance a patient's chance of survival. We first introduce a specially designed three-dimension convolutional neural networks (3D CNN) for candidate detection that takes into account of nodules with various sizes. Then, another 3D CNN is presented for the subsequent false positive reduction. Experimental results of the Tianchi Medical AI Challenge demonstrate the superior detection performance of the proposed approach.