Mutual cineMR/RT3DUS cardiac segmentation.

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Computational Anatomy

Background:

• Cardiac segmentation is crucial for diagnosing heart conditions.
• Ultrasound (US) segmentation faces challenges with image quality and motion artifacts.
• Cardiac magnetic resonance imaging (CMR) offers high-resolution anatomical detail but is less accessible than US.

Purpose of the Study:

• To develop a novel method for segmenting cardiac RT3D ultrasound volumes.
• To enhance ultrasound segmentation accuracy by integrating patient-specific reference shapes from cardiac cine-MR.
• To achieve accurate registration between cine-MR short-axis slices and the ultrasound volume.

Main Methods:

• A level set framework was employed, incorporating region-based and shape-based energy terms.
• A reference shape, derived from segmented cardiac cine-MR short-axis slices, was iteratively registered to the ultrasound contour using an affine transform.
• The method integrates the registered segmentation of cardiac cine-MR into the segmentation of cardiac RT3D ultrasound volumes.

Main Results:

• The proposed method successfully segmented cardiac RT3D ultrasound volumes by leveraging registered cine-MR data.
• A close registration between cine-MR short-axis slices and the ultrasound volume was achieved as a byproduct.
• The method was validated using the MICCAI11 Motion Tracking Challenge database.

Conclusions:

• Integrating registered cardiac cine-MR segmentation significantly improves cardiac RT3D ultrasound segmentation.
• The developed method offers a robust approach for enhancing ultrasound-based cardiac analysis.
• This technique holds potential for improving the accuracy and reliability of cardiac imaging diagnostics.