Lung cancer diagnosis by using deep learning
Principal Investigator
Name
Qinming He
Degrees
Ph.D
Institution
College of Computer Science and Technology, Zhejiang University
Position Title
Prof. He
Email
About this CDAS Project
Study
NLST
(Learn more about this study)
Project ID
NLST-298
Initial CDAS Request Approval
May 4, 2017
Title
Lung cancer diagnosis by using deep learning
Summary
Lung cancer is a leading cause of cancer-related death in China, with a death rate of 85 percent. Due to the severe air pollution in China, the incidence rate of lung cancer continues to grow rapidly.
Early diagnosis of lung cancer from lung CT images can reduce its death rate. However manual evaluation of lung CT images requires expensive human labors, and can easily cause the misdiagnose of lung cancer. To this end, we focus on automatically diagnosing lung cancer by using the deep learning technique and developing an auxiliary lung cancer diagnostic system.
Deep learning technology uses massive tagged data to train a neural network model. This model can later be used to perform specific tasks. We plan to use tagged datasets, such as LIDC-IDRI, to train our model, and use NSLT datasets (mainly spiral CT datasets) to verify training results. Our team has the first-hand exiperience in developing neural network model to identify different types of lupus erythematosus and diagnose lung cancer. We believe we can develop a system that can help radiologists to identify lung nodules.
Early diagnosis of lung cancer from lung CT images can reduce its death rate. However manual evaluation of lung CT images requires expensive human labors, and can easily cause the misdiagnose of lung cancer. To this end, we focus on automatically diagnosing lung cancer by using the deep learning technique and developing an auxiliary lung cancer diagnostic system.
Deep learning technology uses massive tagged data to train a neural network model. This model can later be used to perform specific tasks. We plan to use tagged datasets, such as LIDC-IDRI, to train our model, and use NSLT datasets (mainly spiral CT datasets) to verify training results. Our team has the first-hand exiperience in developing neural network model to identify different types of lupus erythematosus and diagnose lung cancer. We believe we can develop a system that can help radiologists to identify lung nodules.
Aims
Build a diagnostic system than can:
(1) Identify lung nodules from CT scan;
(2) Give preliminary malignancy assessment;
With our system, doctors and hospitals can:
(1) Improve the efficiency of lung cancer diagnosis;
(2) Reduce the diagnosis error of lung cancer;
(3) Advance the detection time of lung cancer.
Collaborators
Dr. Jinlu