A Level Set Segmentation Method of the Four Heart Cavities in Pediatric Ultrasound Images

Echocardiography is the most used medical imaging in pediatric cardiology. It is a fundamental tool to analyze the major heart disease and abnormalities since it is non invasive and simple to use for physicians even when the children are wiggle. Ultrasound images are very noisy, making the segmentation a difficult, not accurate and time consuming task. In this work we propose an automatic segmentation method to extract the four heart cavity boundaries using a new pre-processing algorithm, based on phase symmetry. Experimental results using real echocardiographic images of children show good performance of the proposed method, providing a reliable tool to segment the heart walls that can be helpful for clinical practice.