Aortic Valve Tract Segmentation From 3D-TEE Using Shape-Based B-Spline Explicit Active Surfaces.

Area of Science:

• Medical Imaging
• Cardiovascular Ultrasound
• Computational Anatomy

Background:

• Accurate segmentation of the aortic valve (AV) wall is crucial for diagnosing cardiovascular diseases.
• Existing segmentation methods for 3D transesophageal echocardiography (TEE) datasets often lack precision or efficiency.

Purpose of the Study:

• To develop and validate a novel semi-automatic algorithm for precise AV wall segmentation in 3D TEE.
• To improve the accuracy and reproducibility of quantitative measurements derived from AV segmentation.

Main Methods:

• A 3D cylindrical B-spline Explicit Active Surfaces (BEAS) framework was employed in a dual-stage energy evolution process.
• Combined intensity and shape-based features, incorporating a statistical shape model (SSM) with novel regularization terms.
• Algorithm validated on 20 3D-TEE datasets, assessing segmentation accuracy and computational efficiency.

Main Results:

• The algorithm achieved an average accuracy of 0.78 mm in segmenting the AV wall from the ascending aorta (AA) to the left ventricular outflow tract (LVOT).
• Segmentation was computationally efficient, completing in less than 1 second.
• Semi-automatic measurements showed excellent agreement and higher reproducibility compared to manual measures.

Conclusions:

• The proposed semi-automatic BEAS algorithm offers an accurate, robust, and efficient solution for AV wall segmentation in 3D TEE.
• This method enhances the reliability of quantitative assessments, potentially improving clinical decision-making in valvular heart disease.