Fast and Fully Automatic Left Ventricular Segmentation and Tracking in Echocardiography Using Shape-Based B-Spline...

Area of Science:

• Cardiology
• Medical Imaging
• Biomedical Engineering

Background:

• Cardiac volume and function assessment are crucial in cardiology.
• 3-D ultrasound is increasingly vital for these assessments.
• Accurate left ventricular segmentation in ultrasound is challenging due to image artifacts and low signal-to-noise ratio.

Purpose of the Study:

• To develop a fast and fully automatic framework for endocardial left ventricle segmentation across the cardiac cycle.
• To improve the accuracy and efficiency of cardiac function analysis using 3-D ultrasound.

Main Methods:

• Coupling B-spline explicit active surfaces with statistical shape models for segmentation.
• Utilizing localized anatomical affine optical flow to propagate segmentation throughout the cardiac cycle.
• Employing a purely image information approach enhanced by prior shape information.

Main Results:

• The proposed framework achieves high accuracy, outperforming state-of-the-art methods with mean average distances of 1.81±0.59 mm (end-diastole) and 1.98±0.66 mm (end-systole).
• The method is computationally efficient, averaging 11 seconds per 4-D image.
• The segmentation process is fully automatic, reducing manual intervention.

Conclusions:

• The developed framework offers a robust and efficient solution for fully automatic left ventricle segmentation in 3-D ultrasound.
• This approach enhances the accuracy of cardiac volume and function assessment, benefiting clinical cardiology.
• The combination of active surfaces and statistical shape models provides superior segmentation performance in challenging ultrasound data.