Left ventricle segmentation in fetal echocardiography using a multi-texture active appearance model based on the steered Hermite transform

Highlights • A novel segmentation method applied to the left ventricle in fetal echocardiography is proposed. • The method consists of a multi-texture AAM scheme based on the steered Hermite transform. • The steered coefficients of the Hermite transform are used for coding the texture patterns of the multi-texture AAM. • An automatic initialization framework is presented to detect the left ventricle position. • The developed method constitutes a valuable tool for assessment of the fetal cardiac function using ultrasound images.