Share Email Print

Proceedings Paper

Advanced noise reduction in placental ultrasound imaging using CPU and GPU: a comparative study
Author(s): G. Zombori; J. Ryan; F. McAuliffe; L. Rainford; M. Moran; P. Brennan
Format Member Price Non-Member Price
PDF $17.00 $21.00

Paper Abstract

This paper presents a comparison of different implementations of 3D anisotropic diffusion speckle noise reduction technique on ultrasound images. In this project we are developing a novel volumetric calcification assessment metric for the placenta, and providing a software tool for this purpose. The tool can also automatically segment and visualize (in 3D) ultrasound data. One of the first steps when developing such a tool is to find a fast and efficient way to eliminate speckle noise. Previous works on this topic by Duan, Q. [1] and Sun, Q. [2] have proven that the 3D noise reducing anisotropic diffusion (3D SRAD) method shows exceptional performance in enhancing ultrasound images for object segmentation. Therefore we have implemented this method in our software application and performed a comparative study on the different variants in terms of performance and computation time. To increase processing speed it was necessary to utilize the full potential of current state of the art Graphics Processing Units (GPUs). Our 3D datasets are represented in a spherical volume format. With the aim of 2D slice visualization and segmentation, a "scan conversion" or "slice-reconstruction" step is needed, which includes coordinate transformation from spherical to Cartesian, re-sampling of the volume and interpolation. Combining the noise filtering and slice reconstruction in one process on the GPU, we can achieve close to real-time operation on high quality data sets without the need for down-sampling or reducing image quality. For the GPU programming OpenCL language was used. Therefore the presented solution is fully portable.

Paper Details

Date Published: 12 March 2010
PDF: 9 pages
Proc. SPIE 7629, Medical Imaging 2010: Ultrasonic Imaging, Tomography, and Therapy, 762908 (12 March 2010); doi: 10.1117/12.844478
Show Author Affiliations
G. Zombori, Univ. College Dublin (Ireland)
J. Ryan, Univ. College Dublin (Ireland)
F. McAuliffe, National Maternity Hospital, Univ. College Dublin (Ireland)
L. Rainford, Univ. College Dublin (Ireland)
M. Moran, Univ. College Dublin (Ireland)
P. Brennan, The Univ. of Sydney (Australia)

Published in SPIE Proceedings Vol. 7629:
Medical Imaging 2010: Ultrasonic Imaging, Tomography, and Therapy
Jan D'hooge; Stephen A. McAleavey, Editor(s)

© SPIE. Terms of Use
Back to Top
Sign in to read the full article
Create a free SPIE account to get access to
premium articles and original research
Forgot your username?