Pulmonary embolism is a common cardiovascular emergency
with approximately 600,000 incidents and 200,000 deaths occurring annually in
the United States. CT pulmonary angiography (CTPA) has become the reference
standard for pulmonary embolism diagnosis, but this technique has several issues
with interpretation of the intricate branching structure of the pulmonary
vessels, artifacts that may obscure or mimic embolisms, suboptimal contrast, and
inhomogeneities.
To overcome these shortcomings, researchers at Arizona State
University have developed a machine learning-based approach for automatically
detecting the pulmonary trunk. By using a cascaded Adaptive Boosting machine
learning algorithm with a large number of digital image object recognition
features, this method automatically identifies the pulmonary trunk by
sequentially scanning the CTPA images and classifying each encountered sub-image
with the trained classifier.
This approach outperforms existing anatomy-based approaches.
It requires no explicit representation of anatomical knowledge and achieves a
nearly 100% accuracy as tested on a large number of cases.
Potential Applications
- Diagnosis of pulmonary embolism
- Discrimination of pulmonary embolism from other
hyperbaric injuries
Benefits and Advantages
- Outperforms existing anatomy-based approaches
- Dynamically adapts to suboptimal image contrast
- Discriminates artifacts that may obscure or mimic
embolisms
- Capable of detecting central pulmonary embolisms
- Distinguishes the pulmonary artery from the vein to
remove false positives
- Requires no explicit representation of anatomical
knowledge
- Achieves nearly 100% accuracy as tested on a large number
of cases
Download Original PDF
For more information about the inventor(s) and their
research, please see
Dr. Liang's departmental webpage
