Quality Of Ultrasound Images Mixed With Noise Filter Of Edge Detection

Ultrasound imaging functions by transmitting a beam of ultrasound into the body, collecting and analyzing the returning echo coding the information about internal tissue structures. Ultrasound images have in general low signal to noise ratio (SNR). On the one hand, the returning ultrasound signal from the receiving crystal is of the order of10-100μV, and then fed to the circuits of radio frequency (RF) amplifier, demodulator, filter and so on for further processing, which introduces additive noise into the image. Meanwhile, ultrasound scanners use pulsed signals that have short duration in time and therefore a broad spectrum in frequency domain. As a result, the signals can be affected by multiple noise sources. On the other hand, the spatial interference of the echo from randomly distributed structure scatters produces multiplicative speckle noise in the ultrasound images. Therefore, we propose an ultrasound image enhancement approach by using switching bilateral filter (SBF), which incorporates a texture and noise detector for removing the universal noise as well as preserving the edge and detail information in ultrasound images. Experiments are carried out on synthetic images to compare with other six filtering methods on the basis of signal to noise ratio (SNR), mean structure similarity (MSSIM) and figure of merit (FOM). Simulation results show that the SBF method achieves improved comparative indices and image quality by efficiently removing mixed additive multipl-noise and multiplicative speckle-noise under extremely corrupted condition. Application of the proposed method on real diagnostic ultrasound images also shows a clear improvement over other methods.