Research Overview

Geometric and surface-based shape analysis for neuroimaging

Core Area: Medical/Computer Vision · Computer Graphics · Geometric Learning

Research Topics: Non-Euclidean data (graph/sphere) processing and analysis, including but not limited to:

Spherical representations and geometric learning
Class imbalance and data scarcity in medical datasets
Interactive segmentation and annotation tools
Non-rigid shape synthesis for training and validation

For the broader research motivation and long-term direction of the team, see Research Statement.

Application Areas

Shape Matching

The establishment of a shape correspondence is mandatory for localized shape analysis. Our team has developed advanced surface registration techniques for complex shapes.

Shape Segmentation

Dividing 3D objects into meaningful parts is crucial for understanding shape patterns. Our team has developed automatic surface annotation techniques using deep neural networks.

Shape Feature Extraction

Surface feature extraction represents complex shapes using compact landmarks. Our team has developed automatic landmark extraction for structural variability analysis.

Shape Quantification

Quantifying 3D shapes provides quantitative explanations of shape patterns. Our team has developed novel techniques providing shape markers for brain disorder analysis.