Segmentation Enhanced Elastic Image Registration for 2D Speckle Tracking Echocardiography-Performance Study In Silico

Area of Science:

• Cardiology
• Medical Imaging
• Biomedical Engineering

Background:

• Two-dimensional Speckle Tracking Echocardiography (2D-STE) is vital in cardiology but needs improved repeatability and reliability.
• Limited research integrates left ventricle segmentation into displacement and strain estimation for better 2D-STE performance.

Purpose of the Study:

• To enhance displacement and strain estimation in echocardiography by incorporating left ventricle segmentation mask information.
• To evaluate the impact of segmentation quality on the accuracy of these estimations.

Main Methods:

• Proposed using segmentation mask information as a penalty in elastic image registration for displacement estimation.
• Utilized synthetic echocardiographic data (short-axis view) segmented with an active contour method.
• Assessed the influence of varying segmentation mask quality on displacement and strain estimation.

Main Results:

• Displacement and circumferential strain estimations demonstrated a dependency on mask quality.
• The proposed method showed that accuracy gains from good segmentation outweigh potential losses from poor segmentation.
• The approach offers improved accuracy and reliability in echocardiographic analysis.

Conclusions:

• Integrating segmentation mask information into elastic image registration is a promising method for enhancing echocardiographic analysis.
• The technique mitigates accuracy loss associated with imperfect segmentation, offering a more robust approach to cardiac strain estimation.
• This advancement holds potential for improving the diagnostic capabilities of 2D-STE in clinical cardiology.