Evolution in Voxel Sizes Toward Isotropia

In making the transition from four to eight slices per gantry rotation, there is no change in the nominal voxel size. For example, using a field-of-view of 360 mm and a detector configuration of 4 or 8 x 1.25mm, the voxel size is 360/512 = 0.7mm in

3 mm thick at 1 mm intervals (67% overlap)

Figure 14 Overlapping reconstruction: example of overlapping reconstruction, where 3 mm thick slices are reconstructed at 1 mm intervals. It is noted that while there is a change in the reconstruction interval (e.g., 1 mm), there is no change in the slice thickness (e.g., 3 mm). An overlap of 60% or more is optimal for 3D or multiplanar reformations.

Figure 14 Overlapping reconstruction: example of overlapping reconstruction, where 3 mm thick slices are reconstructed at 1 mm intervals. It is noted that while there is a change in the reconstruction interval (e.g., 1 mm), there is no change in the slice thickness (e.g., 3 mm). An overlap of 60% or more is optimal for 3D or multiplanar reformations.

Figure 15 Examples of images acquired with a 16 x 0.625 mm detector configuration and reconstructed at various slice thicknesses. (A) An axial images reconstructed at 0.625 mm thick has increased noise due to fewer photons per voxel. There is, however, much less partial volume averaging in the z-axis (cranio-caudal direction). (B) The same axial image reconstructed at 5.00 mm thick has much less noise but more partial volume averaging. To make off-axis reformations, the axial images are first reconstructed 0.625 mm thick at 0.50 mm interval. This dataset was then rendered in either a straight or curved format. (C) In this coronal image, the slice thickness in the z-axis (antero-posterior direction) is 0.625 mm. (D) In this coronal image, the slice thickness in the z-axis is 5.00 mm, dramatically reducing the noise while increasing partial volume averaging.

Figure 15 Examples of images acquired with a 16 x 0.625 mm detector configuration and reconstructed at various slice thicknesses. (A) An axial images reconstructed at 0.625 mm thick has increased noise due to fewer photons per voxel. There is, however, much less partial volume averaging in the z-axis (cranio-caudal direction). (B) The same axial image reconstructed at 5.00 mm thick has much less noise but more partial volume averaging. To make off-axis reformations, the axial images are first reconstructed 0.625 mm thick at 0.50 mm interval. This dataset was then rendered in either a straight or curved format. (C) In this coronal image, the slice thickness in the z-axis (antero-posterior direction) is 0.625 mm. (D) In this coronal image, the slice thickness in the z-axis is 5.00 mm, dramatically reducing the noise while increasing partial volume averaging.

the x-axis, 0.7 mm in the y-axis, and 1.25 mm in the z-axis (plus nominal slice broadening as determined by the pitch). When making the transition from 8 to 16 slices per gantry rotation, the nominal voxel size can be smaller. For example, using a 360 mm field-of-view and a detector configuration of 16 x 0.625 mm, the voxel size is 360/ 512 = 0.7mm in the x-axis, 0.7mm in the y-axis and 0.625mm in the z-axis (plus nominal slice broadening as determined by the pitch). For the first time in the history of CT, the voxels can be isotropic, meaning that the dimensions of the voxel are

Figure 16 Coronal reformations of the entire abdomen and pelvis acquired with an 8 x 1.25mm detector configuration and pitch of 1.68. The axial images are reconstructed with a thickness of 1.25 mm at 0.50-mm intervals. A coronal image is then reformatted with a thickness of 5.00 mm in the z-axis (antero-posterior direction) to decrease noise. (A) There is a large metastasis in the right hepatic lobe. (B) No masses are noted on an image obtained more anteriorly, but there is excellent depiction of the stomach due to distention of the lumen with water and enhancement of the wall with intravenous contrast material.

Figure 16 Coronal reformations of the entire abdomen and pelvis acquired with an 8 x 1.25mm detector configuration and pitch of 1.68. The axial images are reconstructed with a thickness of 1.25 mm at 0.50-mm intervals. A coronal image is then reformatted with a thickness of 5.00 mm in the z-axis (antero-posterior direction) to decrease noise. (A) There is a large metastasis in the right hepatic lobe. (B) No masses are noted on an image obtained more anteriorly, but there is excellent depiction of the stomach due to distention of the lumen with water and enhancement of the wall with intravenous contrast material.

equal in the x-, y-, and z-axes. Although spatial resolution in all three axes is improved, the trade-off is fewer photons per voxel and higher image noise. Since kV and mA are set to near maximum on an eight-slice MDCT scanner to achieve adequate image quality with low noise, there are few options to increase the technique further on a 16-slice scanner. As a result, until, X-ray tubes with higher heat loading capacity are available, images acquired at 16 x 0.625 mm will be inherently noisy, which may limit some applications, such as CT angiography (Fig. 15).

0 0

Post a comment