METHODOLOGY

METHODOLOGY

Reconstruction Techniques

Early approaches to 3D echocardiography were

based on the principle that a 3D data set could be

reconstructed from a series of 2D images. In this

method, serial 2D images are obtained using either

freehand scanning or a mechanically driven transducer

that sequentially recorded images at predefined

intervals.4-8 With freehand scanning, a series

of images is obtained by manually tilting the transducer

along a fixed plane, and a spatial locator

attached to the transducer translates the 3D spatial

location onto a Cartesian coordinate system.

This approach has several practical limitations,

including the relative bulk of the acoustic spatial

locators, which makes transducer manipulation

difficult, and the need for a clear and direct path

between the acoustic locators and the transmitter.

For electromagnetic spatial locator systems, an

additional problem is the potential for interference

of the electromagnetic field by ferromagnetic

material in close proximity to the transducer

(eg, material in hospital beds and medical equipment).

9

An alternative to freehand scanning is the use of a

mechanized transducer to obtain serial images at set

intervals in a parallel fashion or by pivoting around

a fixed axis in a rotational, fanlike manner. Because

the intervals and angles between the 2D images are

defined, a 3D coordinate system can be derived from

the 2D images in which the volume is more uniformly

sampled than with the freehand scanning

approach.

More recently, the use of a transesophageal or

transthoracic multiplane probe has emerged as a

readily available method to obtain rotational images

at defined interval angles around a fixed axis.10-15

Typically, images are collected over a 180-degree

rotation at set intervals. To minimize reconstruction

artifacts, sequential images are gated to both electrocardiography

(ECG) and respiration. Acquisition

of a complete data set typically takes 1 to 5 minutes,

depending on respiratory and heart rates and the

predefined spatial intervals. During cardiac surgery,

respiration can be suspended during acquisition to

minimize the effects of respiratory motion.

The quality of 3D reconstructions from 2D

images depends on a number of factors, including

the intrinsic quality of the ultrasound images, the

number (or density) of the 2D images used to

reconstruct the 3D image, the ability to limit

motion artifact, and adequate ECG and respiratory

gating. In general, the greater the number of

images obtained (ie, the smaller the space intervals

between images), the better the 3D reconstruction.

However, increasing the number of

images also lengthens the acquisition time, which

can potentially introduce motion artifact. Consequently,

the optimal number of images necessary

for 3D reconstruction depends on the cardiac

structure being examined and the resolution required.

For example, 4 to 6 serial images are

usually adequate for volume reconstructions of

the left ventricle (LV), whereas more images are

often needed to visualize more complex, rapidly

moving structures, such as mitral and aortic

valves.

Once the 2D images have been obtained, they

are processed offline with customized or commercially

available software. The cardiac structures

are manually or semiautomatically traced to the

3D spatial coordinates to reconstruct a 3D image.