Longitudinal analysis of lung diseases using CT and pathology images

Principal Investigator

Name
Vaishnavi Subramanian

Degrees
M.S.

Institution
University of Illinois at Urbana-Champaign

Position Title
Graduate Research Assistant

Email
vs5@illinois.edu

About this CDAS Project

Study
NLST (Learn more about this study)

Project ID
NLST-512

Initial CDAS Request Approval
May 22, 2019

Title
Longitudinal analysis of lung diseases using CT and pathology images

Summary
The lung screening trial provides a large collection of longitudinal data corresponding to lung diseases in thousands of patients. While most previous models were built based on a single time point, the inclusion of time-varying information in patients will enable us to monitor and understand the disease progression in finer detail.

We plan to make use of the longitudinal CT data to predict the onset, stage progression and mortality events of lung diseases in patients. Our specific focus will be on lung cancer. We will develop machine learning and deep learning models for these tasks, making use of temporal modeling ideas, including LSTMs and GRUs. We will also develop a multi-modal model to predict patient prognosis and survival using fused information from CT and pathology images - these correspond to different scales and could provide useful complementary information for more accurate prognosis and survival prediction.

Aims

- Develop a longitudinal model to predict the onset of lung cancer using CT images
- Develop a longitudinal model to predict the progression of lung cancer using CT images
- Build a multi-modal fusion network that combines information from CT and pathology images to predict the prognosis and survival variables of lung cancer patients

Collaborators

Mary Kathleen Dasso, University of Illinois at Urbana-Champaign
Minh Do, University of Illinois at Urbana-Champaign
Joseph R. Evans, OSF HealthCare, Peoria
Sugandhi Mahajan, Carle Foundation Hospital