Machine Learning based lung nodule detection and malignancy prediction from low dose computed tomography (LDCT) images

Principal Investigator

Name
Tina Kapur

Degrees
Ph.D

Institution
Brigham and Women's Hospital

Position Title
Executive Director, Image-guided therapy

Email
tkapur@bwh.harvard.edu

About this CDAS Project

Study
NLST (Learn more about this study)

Project ID
NLST-408

Initial CDAS Request Approval
Apr 26, 2018

Title
Machine Learning based lung nodule detection and malignancy prediction from low dose computed tomography (LDCT) images

Summary
It is widely recognized that early detection of lung cancer can reduce the associated morbidity and mortality. An influential study in NEJM 2013 demonstrated how predictive tools based on patient and nodule characteristics can be used to accurately estimate the malignance probability of lung nodules detected on baseline screening low-dose CT scans. The purpose of this study is to further to further refine tools we have created for detection and prediction of lung malignancies in the context of biopsy proven diagnosis and baseline and sequential imaging scans, when available.

Aims

Aim 1: Refine automated machine learning algorithms for malignancy detection and prediction that incorporate imaging and pathology data into the training and testing process

Aim 2: Investigate improvements to detection and predictive strength of using longitudinal data

Collaborators

1. Alireza Mehrtash, Brigham and Women's Hospital, Boston, USA
2. William Wells, Brigham and Women's Hospital, Boston, USA
3. Roya Khajavi, Brigham and Women's Hospital, Boston, USA
4. Purang Abolmaesumi, University of British Columbia, Vancouver, Canada
5. Bernardo Bizzo, MGH and BWH Center for Clinical Data Science, Boston, USA