Three-Dimensional Model-Based Segmentation in Echocardiography Using High Temporal Tissue and Blood Flow Information.

Area of Science:

• Medical Imaging
• Cardiovascular Ultrasound
• Biomedical Engineering

Background:

• Accurate 3-D endocardial segmentation in echocardiography is difficult due to low image quality.
• Distinguishing between blood pool and myocardial tissue is crucial for cardiac analysis.

Purpose of the Study:

• To develop and validate a new method for 3-D endocardial segmentation using temporal decorrelation of radiofrequency (RF) signals.
• To assess the feasibility and accuracy of the proposed segmentation technique in phantoms and in vivo.

Main Methods:

• Utilized 3-D Doppler mode to acquire sequences of RF lines for temporal feature extraction.
• Leveraged the decorrelation of RF signals in blood flow versus the correlation in myocardial tissue.
• Implemented a 3-D segmentation model incorporating temporal information as an external constraint.

Main Results:

• The model successfully segmented an artificial blood lumen in a phantom, even at low flow velocities.
• Segmentation sensitivity to flow rate was demonstrated in phantom studies.
• High Dice similarity indices in healthy volunteers confirmed the feasibility of in vivo 3-D endocardial border segmentation.

Conclusions:

• Temporal decorrelation of RF signals provides a robust method for differentiating blood and tissue in 3-D echocardiography.
• The developed 3-D segmentation model is effective for accurate endocardial border delineation.
• This technique holds promise for improving 3-D cardiac imaging and analysis.