A New Morphological Image Segmentation with Application in 3D Echographic Images

This paper reviews ultrasound segmentation methods, in a broad sense, focusing on techniques developed for medical ultrasound images. Segmenting abnormal from normal myocardium using highfrequency intracardiac echocardiography (ICE) the 3D real-time images present new challenges for image processing in LabVIEW. Gray-level intensity and texture features of ICE images of myocardium with the same structural/perfusion properties differ. There are significant limitation conflicts with the existing segmentation techniques. The novelty of this paper consists of a new seeded region growing method to overcome the limitations of the existing segmentation techniques. The segmentation techniques are implemented using graphical programming LabVIEW and Vision. We use three criteria for region growing control: First, each voxel is merged into the globally closest region in the multifeature space. Second structural similarity is introduced to overcome the problem that myocardial tissue, despite having the same property, may be segmented into several different regions using existing segmentation methods. Third equal opportunity competence criterion is employed making results independent of processing order. This novel watershed segmentation method is applied to in vivo intracardiac ultrasound images using pathology as the reference method for the ground truth. The corresponding results demonstrate that this method is reliable and effective. Key-Words: ultrasound, echocardiography, graphical programming, watershed transform, watershed segmentation, image processing, LabVIEW, Vision.