University of Bergen | Faculty of Mathematics and Natural Sciences | Department of Informatics | Visualization Group
Visualization
You are here: Department of Informatics > Visualization Group > Publications > Solteszova12Lowest
 Visualization
 > about
 > team & contact info
 > research
 > publications
 > projects
 > teaching
 > seminars
 > resources
 > network
 > events
 > links

Lowest-Variance Streamlines for Filtering of 3D Ultrasound

Veronika Šoltészová, Linn Emilie Sævil Helljesen, Wolfgang Wein, Odd Helge Gilja, Ivan Viola

INPROCEEDINGS, Eurographics Workshop on Visual Computing for Biology and Medicine (VCBM 2012), Sep, 2012

Abstract

Ultrasound as an acoustic imaging modality suffers from various kinds of noise. The presence of noise especially hinders the 3D visualization of ultrasound data, both in terms of resolving the spatial occlusion of the signal by surrounding noise, and mental decoupling of the signal from noise. This paper presents a novel type of structurepreserving filter that has been specifically designed to eliminate the presence of speckle and random noise in 3D ultrasound datasets. This filter is based on a local distribution of variance for a given voxel. The lowest variance direction is assumed to be aligned with the direction of the structure. A streamline integration over the lowest-variance vector field defines the filtered output value. The new filter is compared to other popular filtering approaches and its superiority is documented on several use cases. A case study where a clinician was delineating vascular structures of the liver from 3D visualizations further demonstrates the benefits of our approach compared to the state of the art.

Published

Eurographics Workshop on Visual Computing for Biology and Medicine (VCBM 2012)

Media

  • paper
  • www
  • Click to view
  • Click to view

BibTeX

@inproceedings{Solteszova12Lowest,
 title = {Lowest-Variance Streamlines for Filtering of 3D Ultrasound},
 author = {Veronika \v{S}olt{\'e}szov{\'a} and Linn Emilie S{\ae}vil Helljesen and 
  Wolfgang Wein and Odd Helge Gilja and Ivan Viola},
 year = {2012},
 month = {Sep},
 booktitle = {Eurographics Workshop on Visual Computing for Biology and Medicine 
  (VCBM 2012)},
 pages = {41--48},
 location = {Norrk{\"o}ping, Sweden},
 abstract = {Ultrasound as an acoustic imaging modality suffers from various kinds of 
  noise. The presence of noise especially hinders the 3D visualization of ultrasound 
  data, both in terms of resolving the spatial occlusion of the signal by surrounding 
  noise, and mental decoupling of the signal from noise. This paper presents a novel 
  type of structurepreserving filter that has been specifically designed to eliminate 
  the presence of speckle and random noise in 3D ultrasound datasets. This filter is 
  based on a local distribution of variance for a given voxel. The lowest variance 
  direction is assumed to be aligned with the direction of the structure. A streamline 
  integration over the lowest-variance vector field defines the filtered output value. 
  The new filter is compared to other popular filtering approaches and its superiority 
  is documented on several use cases. A case study where a clinician was delineating
  vascular structures of the liver from 3D visualizations further demonstrates the 
  benefits of our approach compared to the state of the art.},
 url = {http://diglib.eg.org/EG/DL/WS/VCBM/VCBM12},
 DOI = {10.2312/VCBM/VCBM12/041-048},

}






 Last Modified: Jean-Paul Balabanian, 2013-11-18