Evaluation of linear filter kernel developed from ultrasound phantom image

Abstract

The resolution reduction in ultrasound images is due to the nature of ultrasound propagation and interaction with tissues, in which speckle noise is the dominant noise source. A tissue equivalent phantom, with ultrasound properties similar to human tissues, was imaged and a background kernel was extracted from the image to function as a linear filter. The phantom image has been studied to evaluate the kernel size and location that could be selected. The objective of using the ultrasound phantom image is to simulate homogenous noise distribution related to scanning parameters of the ultrasound system. Implementations of the proposed method in pre-diagnosed images were presented. Standard filters were implemented to assess the performance of the proposed algorithm. The proposed method presents better results than a number of common linear filters.