Nodule malignancy prediction using deep learning and advanced statistics
Principal Investigator
Name
Miguel Angel Gonzalez Ballester
Degrees
Ph.D.
Institution
UPF
Position Title
Research Professor
Email
About this CDAS Project
Study
NLST
(Learn more about this study)
Project ID
NLST-412
Initial CDAS Request Approval
May 15, 2018
Title
Nodule malignancy prediction using deep learning and advanced statistics
Summary
The aim of this project is to leverage state-of-the-art deep learning techniques and advanced statistics to predict malignancy of lung nodules on sequential CT scans. The lines of research to be pursued are threefold: 1) develop a nodule segmentation tool using U-Net architectures based on CT scans; 2) develop a model to predict malignancy on nodules using DL; and (3) assess the evolution of lung nodule malignancy through sequential CT scans using advanced statistics, as well as the enhanced performance in prediction by using sequential CT scans as opposed to one. We thus request access to the NLST datasets “Spiral CT comparison read abnormalities” and “Lung cancer”.
Aims
Nodule segmentation
Malignancy prediction using one CT scan
Malignancy prediction using sequential CT scans
Collaborators
Gabriel Bernardino, UPF - DTIC / Philips France Commercial (Medisys)
Enric Cosp Arqué, UPF
Ricard Delgado Gonzalo, Swiss Center for Electronics and Microtechnology (CSEM)
Josep Marc Mingot, Independent researcher