TDM : de la rotation continue à l’acquisition volumique.
Hans Baartman
_______
TOSHIBA MEDICAL SYSTEMS EUROPE
Zilverstraat 1
2718 RP Zoetermeer
Pays-Bas
E-mail : hbaartman@tmse.nl

Since the introduction of CT scanning in the late seventies, image reconstruction is performed through filtered back projection of views that represents the absorbed X-ray of the human body in a plane for the particular direction (view angle) of the X-ray beam. With the introduction of the third generation scanners whereby the tube and detector rotate around the patient simultaneously, these views contain 360-degree of axial absorption in one plane. Since all views are acquired in the same plane, the reconstruction of these views form a pure cylindrical slice. The collimation of the beam determines the represented thickness of the image pixel. Addition of more consecutive slices enables the acquisition of a volume. In general the distance between the images is set to the same dimension as the slice width. Therefore reconstruction of secondary images, coronal & sagital slices in MPR, as well as 3D images shows step artefact. This can be avoided through the acquisition a larger number of thinner slices. The disadvantage of this technique is the higher dose that is required to obtain the same signal to noise ratio in the original slices.

In 1990 Helical scanning was commercially introduced. With this scan technique the patient is moved in the longitudinal direction (Z-axis) simultaneously with the rotation of the X-ray tube / detector. The patient displacement during the scan can be defined in two ways:

    • By stating the real couch top speed in mm/rotation.
    • By stating the pitch, which is defined by the relation between the slice width and the couch top speed.

A pitch of 1:1 means a table displacement that is equal to the slice witch. In Helical scanning the individual views that represents the X-ray absorption describes a helicoidal movement over the patient. This means that for image reconstruction only one view is in the same plane as the to be reconstructed plane. All the other views are positioned before and after the reconstruction plane. Therefore the views required for a pure cylindrical data set required for an appropriate image reconstruction are calculated through interpolation of views with the same view angle (direction). The weighting factor of the individual views within this interpolation is defined by distance of this view to the reconstruction plane and has a linear relation. This interpolation technique is called a 360-degree linear interpolation.

 In a conventional axial scan basically each opposing view contains the same information although it is mirrored. Using the same idea in the interpolation of the views for the reconstruction of a helical data set interpolation is carried out over 180-degree of data on either sides of the reconstruction plane. This data set is also called complementary data. Again the weighting factor is defined by distance of the view to the reconstruction plane and has a linear relation. This interpolation technique is called a 180-degree linear interpolation.

 Each technique has its advantage and disadvantage. Considering a scan protocol whereby the pitch is set to 1:1 the following phenomenon’s will appear.

    • Since the 360-degree interpolation uses two views with an identical direction the interpolated view has an improved signal to noise ratio, and therefore the image contains less noise. The disadvantage is that the reconstructed slice width is a factor 1.35 larger than the acquired slice.
    • The 180-degree interpolation does not suffer from this enlarged slice width, however the signal to noise ration will be worse. Because in general practice representation of the correct reconstructed slice is preferred, the 180-degree interpolation is most commonly used.

Once a volume is acquired, this data set can be used to reconstruct intermediate slices at arbitrary intervals. Although the interpolated views represent the same slice thickness the pixel information of these intermediate slices is different because of the different weighting of the acquired views.

Addition of all these images into one volume a very smooth transient between the consecutive slices can be achieved. Therefore reconstruction of secondary images, coronal & sagital slices in MPR, as well as 3D images does not show the significant step artefacts shown in volume reconstruction based on conventional axial scanning.

In 1998 the next step towards single slice volume CT is set by the introduction of a multi row detector, enabling the acquisition and reconstruction of multiple slices out of one rotation data set. In a multi row detector more elements, at this moment 8, 16 or 34 (dependent on manufacturer), are placed in the Z-axis of the detector. This enables the acquisition of different data sets simultaneously.

In the development of the multi row detector there are two major design differences:

    • The so called adaptive array detector, whereby the size of the element increases when the element is positioned more to the edge of the detector. With this technique the slice thickness’ are determined by the on the centre of rotation projected sizes of the elements.
    • The real array detector, whereby the sizes of all elements are identical. With this technique any slice thickness that is a multiple of the element size is possible.

In general the pitch for the data acquisition will be set to 4:1. This is a consecutive acquisition of four slices in one rotation. In case the same interpolation technique as used for single slice Helical scan is used the image quality is comparable with an image of single slice Helical scan with a pitch of 4:1. With such a pitch a lot of image information is lost.

Since the data acquisition is carried out with four simultaneous data sets, the information of one rotation can be cut into four section of 90 degree. In this way the interpolation of the data set for one reconstructed image is covered by the acquired data of all four slices.

In order to obtain the same narrow slice width the 180-degree interpolation is used. In a four slice acquisition and a pitch of 4:1 the complementary data of the first 90-degree is identical to the data of the third 90-degree and vice versa. This means that although complementary data is used for interpolation the views are composed as if they were interpolated by a 360-degree interpolation. To overcome this phenomenon the use of so called optimised sampling scan recommended. In this mode the pitch is not set to an full integer figure but to 3.5:1 or 4.5:1. Now the complementary data of the first and the third 90-degree data are not identical.

For the reconstruction of secondary images, coronal & sagital slices in MPR, as well as 3D images, the volume acquisition the in multi slice Helical scan does not differ from the single slice Helical scan, except of the decreased acquisition time,. This decreased acquisition time however enables us to acquire more functional images, cardio vascular images and pure single phase contrast enhanced vessel study’s.