Ultrasound Imaging in 3D

MGI group, UCL

This contains some descriptive text,....very much under development...

Introduction

In a nutshell: Medical Ultrasound (US) imaging usually records 2D sections through soft tissues, highlighting discontinuities in the acoustic impedance (it produces localised echoes). These discontinuities occur at organ boundaries, giving outlines; smaller structures such as vessels, giving anatomical detail; and from tissue infrastructure at a scale at or below the image resolution, giving rise to 'speckle'. Speckle is sometimes regarded useful (though subjective) information; but more often as noise. Attempts to free the user from this 'noise' in conventional 2D imaging (Speckle reduction) has been unsuccessful. Clinical users appear not to trust the images processed in this way, probably rightly, since any such processing will in general reduce the information content of the scan. In most current 3D US methods, a set of these 2D slices is recorded in such a way that their position in space (& thus of all contained echoes) is known. Combining these slices into some sort of 3D array (eg in computer memory) forms a 3D Image. If the set of 2D slices has been carefully acquired, this 3D image will contain a good representation of the 3D anatomical detail (fig 2). Since computer screens and the human retina is 2 dimensional, in order to visualise the information we usually need to provide lots of different views in order to allow perception of the 3D detail. The familiar shaded surface rendering (see figs below) gives one such view, in which a rough idea of surface can be obtained from simulated lighting/shading. Rotating sequences give a better idea of the surface, but in order to see what is behind, a variety of further interactions such as cutting; multiplanar reformatting; various translucent 'volume renderings' is needed.

If the target organ is moving, such as in the fetal heart, we do gated studies' and acquire a dynamic or 4D US study. This demands even more dynamic display methods.

History

We have been developing 3D US imaging techniques for over 8 years, using hardware and software developed here at UCL and UCL Hospitals. We have concentrated on important clinical problems, assessing and demonstrating the great potential for 3D (and 4D) Ultrasound, while continuing to develop relevant acquisition and display methods.
We aquire images from standard scanners as video frames, using a magnetic position sensor to record the probe position. After reconstruction we display the images on our MGI 3D Workstation. Much of our early success is due to the facilities for image surgery (editing) and the variety of display methods available on this powerful system.
The 3D data set is referred to as the 3D image; a variety of display methods including the familiar shaded surfaces and multiplanar reformatting are required to reveal all this information. Some examples from over 600 recorded scans are given below. Don't miss our most recent dynamic 3D (4D) fetal heart studies. >

For a more visual browse, until this is more complete, try Visual 3D Ultrasound

---

Disclaimer

Personnel

---

Back to