Beam hardening refers to a shift in the X-ray spectrum to higher energy photons caused by absorption of lower energy photons within very dense structures such as metallic material. It may lead to a virtual loss of CT density in surrounding soft tissue (i.e., it may look “darker” than it should and black streaks may occur). This is because higher energy photons are less sensitive to soft tissue and iodine attenuation. Figure 13.2 shows beam hardening with corresponding tissue or filling defects caused by high density contrast agent or metal. Dedicated image reconstruction algorithms (see below) may correct or even overcorrect for this effect and may thus eventually even exaggerate soft tissue densities.

So-called blooming of stent struts is an apparent increase in strut size in x-, y-, and z-direction and is mainly caused by partial volume effects. Its magnitude is influenced by the reconstruction algorithm used for image computation. Whenever a voxel includes two different densities (e.g., parts of stent struts as well as lumen), the average density of the two will be displayed. As CT density values of metal are extremely high, those average values will always be way above those of body tissue and – in the window-level settings used in CT angiography – will rather resemble those of the stent struts. Thus the better spatial resolution gets, which means the smaller the voxels are, the fewer partial volume effects will be seen. However, even state-of-the-art high-end CT scanners do not provide a spatial resolution equivalent to current stent strut size (0.07–0.15 mm). One of the most bothersome effects of blooming is artificial lumen narrowing. The in-stent lumen is systematically underestimated in CT: artificial narrowing ranges from 20 to 100% depending on stent material. When using current 64-row CT with smaller slice thickness and dedicated reconstruction kernels, artificial narrowing is reduced (Fig. 13.3) but still considerable at about 30–40%.

Considering the general challenges of cardiac CT (motion) and the special challenges of imaging coronary artery stents (metal artifacts enhanced by motion), individualized contrast timing and premedication (e.g., beta blocker and nitroglycerin) as well as a fast gantry rotation speed and the use of the scanner’s best possible temporal resolution are important. Please be reminded that a scanner’s temporal resolution is best at the center of the scan field since the influence of the fan angle increases in off-center positions. So try to ensure optimal positioning of the heart by adjusting table height and patient position on the table accordingly (Chap. 9).

The scanner’s detector size (e.g., 0.5 mm) determines the thinnest possible reconstructed slice thickness and thus spatial resolution in Z-direction. Coronary artery stent imaging demands submillimeter collimation. Yet, smaller detector elements will result in images with higher noise. Unfortunately, increased image noise degrades the depiction of fine details. Thus, the X-ray input needs to be increased when the smallest possible detector collimation is to be used, especially in larger individuals.

The reconstruction field of view determines pixel size within the axial image. For coronary stent evaluation, the field of view should be just large enough to contain the entire heart, but small enough to take advantage of its 512 × 512 matrix (Chap. 9). Of course, a second reconstruction with a larger field of view to cover possible noncoronary findings is also needed.

The reconstruction algorithms used for image calculation from raw data can manipulate the original data in several ways. They can smoothen the overall image impression by stretching the transition between different tissues/densities over several voxels or emphasize tissue differences by sharpening the delineation of neighboring CT attenuation values. Unless confounded by additional filtering or iteration loops, smoothing algorithms will decrease and edge-enhancing filters will enhance background image noise. The different image impressions resulting from use of a soft vs. intermediate reconstruction kernel with identical window-level settings are shown in Fig. 13.4. The recommended reconstruction kernels for the evaluation of coronary artery stents are frequently identical to those used in the presence of heavy coronary calcifications. Table 13.1 gives an overview of the reconstruction algorithms recommended for coronary artery stent imaging.

In addition, reconstruction algorithms may contain corrections for local artifacts such as beam hardening or weakening in the vicinity of high-density material. An increase in Hounsfield unit (HU) values within the stent lumen compared to lumen attenuation proximal and distal to the stented segment may result from both limited spatial resolution and overcorrection of the algorithm used. False-high in-stent densities can be reduced by using dedicated sharp reconstruction algorithms for coronary stent visualization (Fig. 13.5).

With increasing computational power of recent CT systems, so-called iterative reconstruction algorithms have become available in clinical practice. Unlike so-called weighted filtered back projection (WFBP) used in standard reconstruction algorithms, iterative reconstruction algorithms perform repeated calculations from raw data to image data and back (= iteration loops), generating retransferred raw data sets, which are compared with the originally scanned data in order to identify differences caused by noise. This can be repeated several times, as predefined by the user (two or more iteration steps) and finally be used to reduce noise without losing spatial resolution. Resulting images appear smoothened, yet detailed. However, as anything in life, iterative reconstruction can be overdone as well: as the number of iteration loops is increased (5 or more), the more the visual impression of our images deviates from what we are used to and deteriorates again (Fig. 13.6

Stay updated, free articles. Join our Telegram channel

Join