Deep Learning for Lung Cancer Detection

Principal Investigator

Name
Ryan Sherman

Degrees
B.S.

Institution
Deep Analytics

Position Title
Machine Learning Engineer

Email
rsml7@outlook.com

About this CDAS Project

Study
NLST (Learn more about this study)

Project ID
NLST-341

Initial CDAS Request Approval
Sep 27, 2017

Title
Deep Learning for Lung Cancer Detection

Summary
Using deep learning techniques such as convolutional neural networks, segmentation algorithms like UNet, gradient boosting algorithms, as well as computer vision algorithms, lung cancer can be diagnosed given a CT scan. Lung cancer is a deadly and difficult cancer to detect, so deep learning fits the problem very well as it can be given lots of data, learn, and perform well. Using machine learning libraries, like Tensorflow or Keras, deep learning software can be built to diagnose lung cancer.

Aims

The goal of this project is to save lives, time, and money. Lung cancer takes the lives of around 1.7 million people every year, and early detection of lung cancer doubles the chance of survival. Also, radiologists have limited time to look at CT scans and diagnose patients, making it easy for misdiagnoses. By building intelligent software that can diagnose cancer in the blink of an eye, it will save lots of time for radiologists. Additionally, it is very costly to visit several doctors and radiologists to get a diagnosis, so this project will ensure that the patient will only have to get one diagnosis without having to worry about visiting multiple doctors to make sure the diagnosis is correct. This project also seeks to increase accessibility of this technology to everyone, so they have a greater chance at surviving lung cancer, and I hope the project will inspire others to take part in the fight against cancer.

Collaborators

Ryan Sherman
Deep Analytics