Using generative adversarial neural network to diagnose benign pulmonary nodules in lung cancer screening trial
Principal Investigator
Name
Guangming Lu
Degrees
M.D.
Institution
Department of Medical Imaging, Jinling Hospital
Position Title
Director of Department of Medical Imaging
Email
About this CDAS Project
Study
NLST
(Learn more about this study)
Project ID
NLST-435
Initial CDAS Request Approval
Aug 22, 2018
Title
Using generative adversarial neural network to diagnose benign pulmonary nodules in lung cancer screening trial
Summary
NLST has shown a 20% reduction in lung cancer mortality by low-dose computed tomography screening of high-risk population. However, most of nodules detected in the screening were proved to be benign, which may induce unnecessary follow up and over-diagnosis. Thus, this project aims to develop and validate a computer aided diagnosis (CAD) tool, which can definitely diagnose benign pulmonary nodules in the lung cancer screening and increase the confidence of radiologists. Specifically, we would combine large sample dataset without pathological results and limited data with pathological diagnosis, training and validating a robust and well-generalized CAD tool by using generative adversarial neural network.
Aims
Goal 1: Develop a CAD tool which can definitely diagnose benign pulmonary nodules in the lung cancer screening；
Goal 2: Construct external validation using the NLST CT data.
Collaborators
Longjiang Zhang, Jinling Hospital, Clinical School of Medical College, Nanjing University
Zhiqiang Zhang，Jinling Hospital, Clinical School of Medical College, Nanjing University
Xiuli Li, Deepwise, Beijing, China