Automated detection of pulmonary nodules via deep neural networks
Lung cancer has been the leading cause of cancer death worldwide. We propose a novel framework with 3D convolutional networks for automated detection of pulmonary nodules from low-dose CT scans, which is a challenging yet crucial task for lung cancer early diagnosis and treatment.
Problem to be solved:
Screening primary lung cancer by automatically detecting lung nodules from low-dose CT images
Lung cancer diagnosis via low-dose CT scan
Radiology department in hospitals
Uniqueness and Competitive Advantages:
Our proposed framework consists of two stages: 1) candidate screening, and 2) false positive reduction. Different from previous standard deep learning based methods, we try to tackle the severe hard/easy sample imbalance problem in medical datasets and explore the benefits of localized annotations to regularize the learning.