Weakly Supervised Pulmonary Nodule Detection
Principal Investigator
Name
Yan Yan
Degrees
Ph.D.
Institution
Xiamen University
Position Title
Professor
Email
About this CDAS Project
Study
NLST
(Learn more about this study)
Project ID
NLST-1041
Initial CDAS Request Approval
Apr 13, 2023
Title
Weakly Supervised Pulmonary Nodule Detection
Summary
Project is only used for academic research purpose. Recently, a lot of pulmonary nodule detection methods have been proposed, but they mainly work on the fully-supervised setting. Large-scale labeled data is needed to train models. To address this problem, we aim to propose a weakly-supervised pulmonary detection method to detect pulmonary nodules in lung computed tomography images. Specifically, we want to train a model by only using NLST weak labels (such as central slice, lobe location and nodule number, etc.), and fine tune the model with a fully-supervised dataset (such as LIDC-IDRI). Compared with state-of-the-art methods, the performance of our model is expected to be improved by effectively exploiting weak labels.
Aims
Achieve better detection performance compared with state-of-the-art methods.
Collaborators
Guangyu Huang