Automated reduction the speckle noise of the panoramic ultrasound images of Muscles and Tendons

Shaima Ibraheem Jabbar,Charles Day,Edward Chadwick

doi:10.1088/1742-6596/1660/1/012085

Abstract

Removing or reducing speckle noise is one of the main goals to ensure high quality panoramic ultrasound images of muscles and tendons. The presence of noise in the ultrasound image adds a difficulty in the interpretation of the image by clinicians and researchers. In this work, non-liner filter (local adaptive median filter (LAMF1)) has been developed to do a precise detection for speckle noise pixels and reduce its impact on the ultrasound images. It has been applied on three different types of ultrasound images: Based on using set of assessment metrics: Speckle Suppression Index (SSI), Speckle Suppression Mean Preservation Index (SMPI), Enhanced Edge Index (EEI) and Mean Preservation Speckle Suppression Index (MPSSI)), the new local adaptive median filter (LAMF2) has been compared to LAMF1 and Anisotropic Diffusion Filter (ADF). The performance of developed filter (LAMF2) outperformed the performance of LAMF1 as follows: SSI (3%), SMPI (4.79%), EEI (3.7%) and MPSSI (40%). Besides that, ADF has a high level of SSI, SMPI and MPSSI compared to new filter (LAMF2). However, ADF reported better numerical evaluations (EEI) than LAMF2. It is possible to obtain further performance improvements by combining characteristics of both filters (LAMF2 and ADF).

Highlights

Musculoskeletal Image quality of ultrasound images is affected by the presence of noises
The edge preservation reported high scores in all three-different samples using LAMF2; see figure (4), figure (8) and figure (11). These results indicate that performance of Anisotropic Diffusion Filter (ADF) is better in speckle noise reduction, while LAMF2 outperforms ADF in edge preservation
The results obtained show superior performance of LAMF2 compared with LAMF1

Summary

Introduction

Musculoskeletal Image quality of ultrasound images is affected by the presence of noises. Image noises ruin the fine details of the ultrasound image and effects on the image quality significantly. There are different kinds of noises in the medical images such as impulse noises, Gaussian noises and speckle noises [1]. Impulse and Gaussian noises are additive noises, while speckle noise is a multiplicative noise. The main difference between additive and multiplicative noises is that additive noises do not change the intensity of the main signal, ; it is possible to address it using linear filtering techniques. Multiplicative noises cannot be dealt with using linear filtering techniques because of their complex structure which is due to the interference between grey level intensities of the noises and the original image [2]

Methods

Results

Discussion

Conclusion