Automatic Detection of Cancerous Lung Tissue

Principal Investigator

Name
Maribeth Cogan

Degrees
B.S. Biomedical Engineering

Institution
The University of Texas at Dallas

Position Title
Student - Electrical Engineering Masters Program

Email
mxr127730@utdallas.edu

About this CDAS Project

Study
NLST (Learn more about this study)

Project ID
NLST-355

Initial CDAS Request Approval
Sep 22, 2017

Title
Automatic Detection of Cancerous Lung Tissue

Summary
My project is to identify cancerous tissue within a CT or x-ray of a lung using a deep neural network. To do this, I first need to train the network on a large dataset of normal and cancerous lungs. Then when the trained network sees a new image of a lung, it would identify if cancer is present, and if so, would draw a box around the unhealthy tissue and provide a confidence score.

Aims

The specific aims of this project are (1) to acquire a large database of annotated lung images for healthy and cancerous lungs, (2) to train a deep convolutional neural network to identify healthy vs. cancerous lungs from images, (3) to test the neural network on a reserved set of unseen data, (4) to assess the accuracy of the network and modify the network parameters until optimal accuracy is achieved,and (5) write a paper describing my results.

Collaborators

Dr. Lakshman Tamil, The University of Texas at Dallas
Timothy Cogan, The University of Texas at Dallas