Using deep learning to enhance cancer diagnosis and classification
Principal Investigator
Name
Wang Qin
Degrees
master
Institution
AccuRad
Position Title
Algorithm Engineer
Email
About this CDAS Project
Study
NLST
(Learn more about this study)
Project ID
NLST-145
Initial CDAS Request Approval
Jul 30, 2015
Title
Using deep learning to enhance cancer diagnosis and classification
Summary
Using machine learning to facilitate and enhance medical analysis and diagnosis is a promising and important area. In our project, we study that how unsupervised feature learning from CT images can be used for nodule detection, cancer detection, and cancer type analysis. The main advantage of the proposed method over previous cancer detection approaches is the possibility of applying data from various types of cancer to automatically form features which help to enhance the detection and diagnosis of a specific one.
We will attempt to implement a convolutional deep learning network for nodule classification and detection:
1) First, extracting deep features from an autoencoder, which uses a linear or non-linear transformation to ”encode” the data.
2) Second, using nodule features extracted as input to a trained classifier.
3) Last, comparing the output of the classifier with the ground truth to test its accuracy and effectiveness.
We will attempt to implement a convolutional deep learning network for nodule classification and detection:
1) First, extracting deep features from an autoencoder, which uses a linear or non-linear transformation to ”encode” the data.
2) Second, using nodule features extracted as input to a trained classifier.
3) Last, comparing the output of the classifier with the ground truth to test its accuracy and effectiveness.
Aims
1) Use NLST CT images to do unsupervised feature learning on lung nodules.
2) Ultimately, to provide a reference to the doctor about lung cancer detection.