An Approach for Despeckling a Sequence of Ultrasound Images Based on Statistical Analysis

Abstract

In this paper, a denoising approach taking into account the statistical properties of the ultrasound image is proposed. In the proposed method, the time correlation between the different frames of a video sequence are taken into consideration to find the maximum a posteriori probability estimate of a noisy pixel. The proposed algorithm is tested on both synthetic and real ultrasound images to verify the effectiveness of the system. The paper provides a model for time correlation of the noise between different frames of the ultrasound video data. To simulate the speckle noise in a natural image, an attempt is made to study the probability density function of speckle noise in an ultrasound video sequence. From our analysis, it is concluded that the speckle noise is multiplicative Gaussian in nature. Certain statistical tests are also conducted to verify the effectiveness of our proposed method. Experimental results show that the proposed despeckling method shows superior performance than its counterparts.