Speckle reduction in ultrasound images using temporal and spatial context

Abstract

Several different image restoration methods for speckle reduction in medical ultrasound images were explored. The methods were tested on ultrasound images of the left cardiac ventricle. The same methods were used for 1-D temporal filtering to improve the visual acuity of single image frames and for 2-D spatial filtering for edge enhancement. For temporal filtering, a novel statistical method based on a quantum-mechanical field model was better than geometric filter, median filtering, and adaptive noniterative linear filters. For spatial filtering, the quantum-mechanical field model and the geometric filter gave comparable results, clearly better than median filtering and adaptive noniterative linear filters. It is concluded that a substantial visual improvement and edge enhancement in medical ultrasound images are possible without loss of significant structures. >