Multi-View Active Appearance Models: application to X-ray LV angiography and cardiac MRI

Area of Science:

• Medical image analysis
• Computer vision
• Biomedical engineering

Background:

• Accurate segmentation of cardiac views is crucial for diagnosing heart conditions.
• Existing methods often struggle with inter-view consistency and pose variations.

Purpose of the Study:

• To develop a Multi-View Active Appearance Model (AAM) for coherent segmentation of multiple cardiac views.
• To model and utilize the coherence in organ shape and intensities between different views.

Main Methods:

• Adapted Cootes' AAM framework to incorporate multi-view shape and intensity information.
• Developed a method to eliminate trivial pose differences between views.
• Integrated multi-view coherence into the image search process.

Main Results:

• Successfully validated the Multi-View AAM in two novel applications.
• Demonstrated coherent segmentation of left ventricular X-ray angiograms (end-diastolic and end-systolic phases).
• Achieved simultaneous segmentation of multiple cardiac MR views (four chamber, two chamber, short-axis).

Conclusions:

• The Multi-View AAM effectively models inter-view coherence for improved cardiac image segmentation.
• The method shows promise for complex cardiac imaging scenarios, including combined X-ray and MR data.