Semi-automatic algorithm for construction of the left ventricular area variation curve over a complete cardiac cycle

Area of Science:

• Cardiology
• Medical Imaging
• Image Processing

Background:

• Two-dimensional echocardiography (2D-echo) is crucial for assessing cardiac structures and left ventricular function.
• Manual segmentation of the left ventricle is time-consuming and prone to operator variability.
• Automatic segmentation is challenging due to factors like mitral valve opening during diastole.

Purpose of the Study:

• To develop a semi-automatic method for segmenting the left ventricle in dynamic 2D-echo 4-chamber long-axis images.
• To enable accurate analysis of the left ventricle throughout the complete cardiac cycle.
• To facilitate the calculation of clinical parameters for assessing global systolic function.

Main Methods:

• A novel algorithm combining time-averaging, wavelet denoising, edge enhancement, morphological operations, homotopy modification, and watershed segmentation.
• Semi-automatic approach requiring user input for mitral valve position in the initial frame.
• Segmentation applied to dynamic 2D-echo images from two consecutive cardiac cycles.

Main Results:

• The method was evaluated on twelve healthy volunteers, comparing results against manual segmentation and four alternative methods.
• Quantitative evaluation used four distinct metrics to assess accuracy.
• Intra- and inter-operator variabilities of the method were also assessed.

Conclusions:

• The developed method automatically generates the left ventricle area variation curve over a complete cardiac cycle.
• This facilitates the identification of key clinical parameters, such as the area variation fraction.
• The area variation fraction shows potential for evaluating global left ventricular systolic function.