Speckle removal in medical ultrasound images by compounding and filtering

• Main Author: Zhang, Lichen
• Other Authors: Wong Moon Chung, Eddie
• Format: Theses and Dissertations
• Published: 2008
• Subjects: DRNTU::Engineering::Electrical and electronic engineering::Control and instrumentation::Medical electronics
• Online Access: https://hdl.handle.net/10356/3983
• Institution: Nanyang Technological University

Ultrasound imaging is greatly affected by a kind of artifacts called as speckle. This research work focuses on developing efficient speckle reduction methods to increase the quality of ultrasound images. In this research, a spatial compounding using subapertures is discussed. The transmit-receive subaperture combination varies at each image construction turn, generating partially correlated sub-images for compounding. Simulation results show that the new compounding method suppress speckle effectively. Two newly developed approaches of filtering method are also presented in this thesis. Both of them first decompose speckle images into the Laplacian Pyramids (LP) and then perform filtering on each LP layer. The first approach employs the filters combining a sharpener and a smoother to process LP layers. The performances of the sharpener and smoother are adjusted according to the SNR lever at a specific LP layer. The second filtering scheme employs the adaptive Dutt filter, which makes it fit to smooth log-compressed ultrasound images. The filter’s performance is guided by a speckle statistical parameter, which is derived from the variance of the LP layer regarding to the operations involved in LP generation. Experiment results show the two approaches suppress speckle effectively without damaging important image features.