Research

Teaching

Publications

Source Codes

My research on biomedical image analysis mainly concentrated on machine learning approaches to medical imaging tasks (in particular echocardiograph image processing).

Efficient detection in a high-dimensional space
	Probabilistic Boosting Network (PBN): An efficient approach to rigid structure detection. It breaks down the linear dependency in the number of orientations and scales.
	Probabilistic, Hierarchical, and Discriminant (PHD): A nicely-crafted framework of deformable structure detection with efficiency and accuracy.
Deformable shape segmentation
	Shape Regression Machine (SRM): A regression approach to deformable structure detection and shape segmentation. Image-Based Regression (IBR) Using Boosting: A multivariate multiple regression with feature selection.
	RankBoost: Example based non-rigid shape detection via ranking.
Deformable shape tracking
	Pairwise Active Appearance Models (PAAM): A unifying prior model of shape, appearance, and motion for online tracking of deformable contour.
	BoostMotion: Boosting a discriminant similarity function for motion estimation under varying appearances.
Object detection and categorization
	Image-Based Multiclass Boosting Cardiac view classification of echocardiogram