Sequential shape similarity for active contour based left ventricle segmentation in cardiac cine MR image

Area of Science:

• Medical Imaging
• Cardiovascular Imaging
• Image Segmentation

Background:

• Accurate Left Ventricle (LV) boundary delineation in cardiac Magnetic Resonance Images (MRI) is crucial for cardiac diagnostics.
• Existing segmentation methods may struggle with the temporal dynamics and shape variations in cardiac MRI sequences.

Purpose of the Study:

• To propose a novel approach for simultaneous segmentation of LV in cardiac MRI sequences.
• To introduce and validate the Sequential Shape Similarity (SSS) constraint for improved LV boundary extraction.

Main Methods:

• The proposed method utilizes an Active Contour Model (ACM) as the base segmentation model.
• Incorporates Gradient Vector Convolution (GVC) for external force calculation.
• Introduces the Sequential Shape Similarity (SSS) constraint, assuming shape similarity between adjacent frames in the MRI sequence.

Main Results:

• The SSS constraint enables accurate delineation of LV boundaries by leveraging temporal information.
• Evaluation on two cardiac MRI datasets using Mean Absolute Distance (MAD) and Hausdorff Distance (HD) metrics showed good performance.
• The method demonstrates effective segmentation of LV boundaries in cardiac MRI sequences.

Conclusions:

• The proposed segmentation method with the SSS constraint offers a robust solution for LV boundary delineation in cardiac MRI.
• This approach enhances the diagnostic power of cardiac MRI by providing accurate and consistent LV segmentation across image sequences.