Statistical shape modeling of multi-organ anatomies with shared boundaries

Area of Science:

• Medical imaging and computational anatomy
• Statistical shape analysis
• Biomedical engineering

Background:

• Statistical shape modeling (SSM) is crucial for analyzing anatomical variations.
• Existing SSM methods struggle to explicitly model shared boundaries between organs.
• Failure to model shared boundaries limits the detection of subtle pathological changes.

Purpose of the Study:

• To develop a novel, data-driven approach for building statistical shape models of multi-organ anatomies with explicit shared boundary modeling.
• To enhance the capability of SSM in identifying pathological shape variations at contact surfaces.
• To provide a flexible framework for analyzing complex anatomical structures.

Main Methods:

• Utilized particle-based shape modeling (PSM), an advanced SSM technique.
• Developed a strategy to detect and extract shared boundary surfaces and contours.
• Formulated a correspondence-based optimization algorithm for multi-organ shape models.

Main Results:

• Successfully demonstrated the shared boundary modeling pipeline on toy and clinical datasets.
• Achieved consistent parameterization of the shared surface (interventricular septum) in biventricular heart models.
• Identified interventricular septum curvature as a key indicator of pathological shape differences.

Conclusions:

• The proposed method effectively models shared boundaries in multi-organ anatomies.
• Explicitly modeling contact surfaces enhances the identification of subtle, localized pathological changes.
• This approach offers improved quantitative analysis for complex anatomical structures and their variations.