Spatial compounding of large numbers of multi-view 3D echocardiography images using feature consistency

Abstract

This paper presents a novel method for compounding large numbers of multi-view 3D echocardiography volumes based on feature consistency. Our proposed method directly addresses issues involved with reducing the effects of echocardiography artefacts in the final compounded volume. Quantitative validation experiments are carried out using an echocardiography heart phantom. Images are acquired through various intervening layers of soft-tissue and hard-tissue mimicking material. We use images acquired of the phantom with no intervening material as high-quality reference “gold-standard” images, and then investigate the effects of the introduced soft tissue and strongly reflecting boundaries images on image quality. Our compounding method is compared to the original, uncompounded, echocardiography images, and to images compounded using a published phase-based method. In addition we present qualitative results from a volunteer and a patient dataset. Results show the artefact has been detected and reduced, and a coherent compounded image is produced using large numbers of multi-view 3D volumes.