Detection of pulmonary nodules and reduction of the false positive rate

Principal Investigator

Name
Hsin-Ming Chen

Degrees
M.D.

Institution
National Taiwan University Hospital

Position Title
Visiting Staff

Email
hsinming@g.ntu.edu.tw

About this CDAS Project

Study
NLST (Learn more about this study)

Project ID
NLST-411

Initial CDAS Request Approval
May 7, 2018

Title
Detection of pulmonary nodules and reduction of the false positive rate

Summary
A good CAD tool can help radiologists detect pulmonary nodules quickly. However, many of the current softwares have high false positive rates, which hinder the efficiency and reduce the willingness to use. As a radiologist, I need a more practical aid to make daily work done smoothly and accurately, so I tried a couple of machine learning models to develop a CAD tool for lung nodule detection. In this custom-made CAD, the false positive rate is 4%, which may be improved with a larger dataset.

Aims

Reducing the false positive rate of CAD system while keep adequately high sensitivity.

Collaborators

1. Prof. Yeun-Chung Chang, Chairman, Department of Radiology, College of Medicine, National Taiwan University, Taipei, Taiwan, ROC

Related Publications

• An improved 3-D attention CNN with hybrid loss and feature fusion for pulmonary nodule classification.
  Huang YS, Wang TC, Huang SZ, Zhang J, Chen HM, Chang YC, Chang RF
  Comput Methods Programs Biomed. 2023 Feb; Volume 229: Pages 107278 PUBMED