The Ultrasound Visualization Pipeline - Scientific Visualization: Uncertainty, Multifield, Biomedical, and Scalable Visualization - page 282

maximizing breast preservation [ 65 ] by projecting a segmented tumor onto a video
feed. The tumor is segmented using a minimal intensity projection based selection
of the volume of interest. In the final stage, the tumor is surface rendered and super-
imposed on the video image.
Stetten et al. show how tomographic reflection can provide a superimposed image
onto the body without any tracking systems [ 73 ]. The ultrasound probe carries a half-
silvered mirror. The mirror reflects the ultrasound image which is shown on a flat
panel monitor mounted on the probe. This technique was extended in the Sonic
Flashlight [ 68 ]. The tomographic reflection was shown to increase the localization
perception compared to conventional ultrasound [ 81 ].
Augmented reality shows a great potential benefit in medical ultrasound imaging.
Yet, there is a lag from technology development to the actual integration into every
day usage. Sielhorst et al. published a detailed review for advanced medical displays
in 2008 [ 69 ]. This chapter discuss the potential benefit and the increasing use for
augmented reality in medical imaging in general. They state that improvements in
both technologies are needed to be able to create a seamless integration into the
workflow of physicians and surgeons.
24.8 Summary and Discussion
In this chapter, we have categorized several of the most important works in what con-
stitute the ultrasound visualization pipeline. The pipeline is defined as the five major
categories in data processing and rendering. The five categories are pre-processing,
segmentation, registration, rendering and augmented reality.
Medical ultrasound data is very different compared to other medical imaging
modalities. Techniques for the individual steps in the visualization pipeline are tai-
lored to suit the special nature of the data. For instance, techniques meant for in-vivo
use have strong performance requirements to handle the high frame rate of ultrasound
images. Segmentation and registration becomes very challenging, due to inconsis-
tent data values for similar tissue. Still, ultrasound remains as one of the most used
imaging modalities in medicine. Research in advanced ultrasound visualization tech-
niques focuses greatly on 3D ultrasound, but the trend in diagnostics is mostly 2D
due to higher frame-rates, high resolution and a minimal requirement for interaction.
The temporal and spatial resolution for ultrasound is approaching the physical limits
of the speed of sound. It is very important to explore what strengths and weaknesses
the different modalities possess and combine the strengths into the natural work flow
of medical personnel.
Acknowledgments This work has been carried out within the IllustraSound research project
(#193170), which is funded by the VERDIKT program of the Norwegian Research Council with
support of the MedViz network in Bergen, Norway (PK1760-5897-Project 11). We would also like
to thank Helwig Hauser for invaluable help and fruitful discussions.