Lung abnormality detection from chest radiography
Principal Investigator
Name
Benedikt Graf
Degrees
Ph.D.
Institution
IBM Watson Health
Position Title
Senior Scientist, Imaging Analytics
Email
About this CDAS Project
Study
PLCO
(Learn more about this study)
Project ID
PLCO-359
Initial CDAS Request Approval
Apr 6, 2018
Title
Lung abnormality detection from chest radiography
Summary
Chest radiography is one of the most commonly used medical imaging modalities. The relatively low cost and low radiation exposure risk associated with chest radiography has made it often the first imaging test to help with diagnosis of suspected conditions. A large variety of lung abnormalities and diseases can be evaluated from chest radiographs, such as pulmonary nodules, pleural effusion, and emphysema.
The primary goal of this project is to perform image-based lung abnormality detection from chest radiographs using modern machine learning techniques. The target abnormalities include pulmonary nodules and Chronic Obstructive Pulmonary Diseases (COPDs). A secondary goal is to predict the detected diseases' severities, for example, the size of a nodule and the stage of the COPD. The detection and prediction outcomes will be evaluated against the clinical outcomes.
The primary goal of this project is to perform image-based lung abnormality detection from chest radiographs using modern machine learning techniques. The target abnormalities include pulmonary nodules and Chronic Obstructive Pulmonary Diseases (COPDs). A secondary goal is to predict the detected diseases' severities, for example, the size of a nodule and the stage of the COPD. The detection and prediction outcomes will be evaluated against the clinical outcomes.
Aims
1. Develop machine learning algorithms to detect lung abnormalities from chest radiographs;
2. Develop machine learning algorithms to predict the severities of detected lung abnormalities;
3. Validate the algorithms on clinical findings included in the database.
Collaborators
To be determined