CT Perfusion Imaging Principles

Farhood Saremi, MD

Erin Angel, PhD

▪ Introduction

In 1980, Leon Axel¹ introduced a method for assessing regional cerebral blood flow (CBF) using dynamic contrast-enhanced CT data. He applied the principles of indicator dilution analysis for nondiffusible indicators to the temporal changes of intravascular contrast concentration in order to measure blood volume, mean transit time (MTT) of the first passage of the bolus through the vessels, and the blood flow per unit of vascular volume. He described this method a few years after the first clinical introduction of CT for body imaging when clinical scanner’s operation was very slow to obtain rapid sequence perfusion imaging and motion artifacts and high radiation dose were problematic.² Because of these limitations, most tissue perfusion imaging (TPI) studies in the 1980s were limited to a few research centers using electron beam CT scans, the only commercially available fast scanners, that could acquire a pair of adjacent, cross-sectional images in 50 milliseconds with spatial resolution of 1.5 mm.³,⁴ and ⁵ The invention of multislice spiral scanners stimulated further interest and revolutionized perfusion imaging techniques.⁶,⁷ and ⁸ More recently, 640-slice volume CT scanners and dual-source CT imagers, in conjunction with radiation dose reduction strategies, have expanded the clinical applications of perfusion CT from the assessment of brain ischemia to oncology perfusion and functional analysis of different organs.⁹,¹⁰ Perfusion CT protocols are now readily incorporated into the existing CT workstations and postprocessing techniques for perfusion analysis and have been facilitated by the release of different commercial software packages.

In this chapter, a general overview of CT techniques is introduced and factors influencing TPI is discussed. Details of dose reduction strategies and tissue perfusion analysis models are presented in separate chapters.

▪ State-of-the-Art Multidetector CT (MDCT) Scanners

With the introduction of spiral (helical) CT in the early 1990s, a fundamental evolutionary step was made toward the development of angiographic applications and modification of most dynamic contrast-enhanced CT imaging techniques.¹¹,¹² One primary goal in dynamic CT imaging (including perfusion studies) of body organs is to obtain series of high-resolution images of a tissue volume over a short period of time. Although single-slice spiral CT scanners with 500-millisecond gantry rotation time were able to acquire images with temporal resolution of nearly 250 milliseconds, limitations exist to cover the entire volume of interest in a short period of time (i.e., a single breath-hold). Additionally, multiplanar reformations (MPRs) and three-dimensional (3D) reconstructions with the same spatial resolution of original axial images (isotropic resolution) were not technically feasible.¹²,¹³ To surpass these limitations and in order to achieve the primary goals for high-quality dynamic CT imaging, several CT manufacturers introduced MDCT scanners around the beginning of the year 2000, which provided considerable improvement in scanning speed and multiplanar resolution.¹⁴,¹⁵ and ¹⁶ These systems typically offered simultaneous acquisition of four sections at a gantry rotation time of 0.5 seconds. The increased performance of MDCT relative to older single-slice scanners allowed the optimization of a variety of clinical protocols especially in cardiac imaging.¹⁶,¹⁷ Simultaneous acquisition of multiple slices and increasing body coverage in a short period of time just by recruiting more detectors were strong motivators for all CT manufactures to try to upgrade their scanners with larger detector span. Production of new detector designs and development of more efficient reconstruction algorithms including cone-beam reconstructions allowed the release of 16, 32, and 64 MDCTs by different manufacturers, which made possible the routine acquisition of substantial anatomic coverage in a short time and with submillimeter isotropic spatial resolution. With the addition of electrocardiographic (ECG)-gating capability and faster rotation times (i.e., 275 ms) and dedicated image reconstruction approaches, the temporal resolution for motion-free volume acquisitions of the heart became feasible.¹⁸,¹⁹

320-Detector Row/640-Slice Volume CT

Initially, vendors introduced scanners with greater and greater z-axis coverage in order to shorten scan time and minimize patient motion. In 2007, Toshiba released the first volume CT scanner, which substantially increased the detector coverage to 16 cm in the z-direction (Fig. 2.1). The main driving factor for this innovation was functional imaging (such as perfusion CT). With 16 cm of coverage, the Aquilion ONE Dynamic Volume CT (Toshiba Medical Systems, Otawara, Japan) can image an entire organ in a single rotation with a rotation time as fast as 0.275 seconds. With this capability, all
iodinated contrast within a large organ (e.g., brain, heart) can be captured at the same instant in time; this is called isotemporal imaging. Isotemporal imaging is a key feature for accurate CT perfusion analysis in which contrast uptake and washout are imaged at multiple time points. As one could imagine, without isotemporal imaging, it is nearly impossible to separate the change in contrast due to the difference in acquisition time from the change in contrast due to physiologic processes. CT perfusion applications using volume CT are vast and range from neuro applications to tumor or wholeorgan perfusion in the body.²⁰,²¹,²²,²³,²⁴,²⁵,²⁶ and ²⁷ With isotemporal resolution, the contrast in the vessels and the surrounding tissues of an organ can be visualized at multiple time points to create a “movie” of the contrast uptake and wash out. This dynamic information can then be used to assess timing of contrast arrival and effect on surrounding tissues. Compared to the helical shuttle technique (see below), volume CT perfusion scanning is lower dose because there is no excess exposure from helical overranging. Helical overranging is the exposure at the beginning and end of a helical run.²⁸ Some of the overranging is necessary for helical reconstruction and some of it is excess. The excess exposure is often mediated with adaptive collimation, but the remaining overranging is unavoidable for helical reconstruction.

Figure 2-1. A: Tube-detector system configuration for volume scanner. B: A 16-cm detector panel. (Courtesy of Toshiba Medical System.)

Dual-Source MDCT

Although the idea of a multitube scanner was formed by Franke in 1979, clinical dual-source CT scanners were introduced years later in 2005 with the main goal of increasing temporal resolution to improve quality of cardiac imaging.²⁹ The first generation of dual-source CT was introduced with two x-ray tubes and two corresponding detectors mounted onto the rotating gantry with an angular offset of 90 degrees capable of acquiring 2 × 32 slice readings by an alternating (dual) z-flying focal spot technique in a gantry rotation speed of 330 milliseconds (Fig. 2.2A). In this gantry, the first tube-detector system covers 50-cm field of view (FOV) and the second tube-detector covers a smaller measurement field of 26 cm, both with a minimum reconstructable slice width of 0.6 mm. The second-generation
dual-source CT was modified to acquire 2 × 128 slices, the gantry rotation time was reduced to 280 milliseconds, and the second tube FOV was increased to 33 cm (Fig. 2.2B). The main goal of creating this scanner was to improve the quality of cardiac imaging by increasing the temporal resolution.³⁰ This construction led to a reduction of the rotation needed to acquire the required projection data for image reconstruction from half of a rotation in single-source CT to a quarter of a rotation in dual-source CT using a monosegment reconstruction mode that is consistent throughout various heart rates owing to individual adaptation of the table pitch.³¹ Another desirable approach in the new generation of the scanner has been widening of the scan field for the second tube/detector system to improve coverage for other applications, especially perfusion imaging and dual-energy scanning. This required decreasing gantry rotation speed in order to maintain the temporal resolution for cardiac scanning and a system angle of slightly larger than 90 degrees between the two tube-detector systems (Fig. 2.2B). With this new geometry, the two separate x-ray tube and detector array systems are arranged 94 degrees apart, which also allows helical data acquisition at a high pitch. In contrast to the single-source multidetector CT (MDCT), where the maximum pitch is roughly 1.5 in order to reconstruct gapless images, the second-generation dual-source CT can achieve gapless z-axis sampling even at a pitch value of up to 3.2.³² Therefore, it enables coverage of the entire heart in a single heartbeat. However, slow and regular heart rates are required for this single-heartbeat acquisition, which uses prospectively triggered ECG gating. For heart rates greater than 55 or 60 beats per minute, multiphase acquisition is used, which requires pitch ranges from 0.2 to 0.5 (depending on the heart rate), and a retrospective ECG gating will be used for data acquisition.³³,³⁴ Due to the high temporal resolution, the scanner has been commonly used for the assessment of coronary artery stenosis and is currently being assessed for myocardial perfusion.³⁵,³⁶ Another application of dualsource CT is tissue characterization when both detector systems are operated at different kilovolts, so-called “dual-energy CT.”³⁷ One limitation of this scanner particularly in perfusion studies has been small detector span to cover the entire anatomy of interest in one pass. This limitation has been partially solved using dynamic spiral scanning with variable pitch (“toggling table” technique).³⁸ One problem with the anatomic configuration of dual-source scanners is image degradation due to the presence of scatter radiation induction in the presence of two tubes. To solve this issue, special scattered radiation correction algorithms are used. The mounted x-ray tubes in dual-source scanners are also smaller and lighter than traditional x-ray tubes, which have larger anodes rotating in a vacuum. To improve heat capacity of the small tubes, the anode is placed in direct contact with cooling oil. When both tubes are working, the radiation dose will be doubled. To avoid this problem, lowering dose strategies have been implemented including, 3D adaptive noise reduction filter (while preserving edge detail), iterative reconstructions, heart rate-dependent pitch values, and ECG-based tube current modulation. A cardiac or body beam-shaping filter (bowtie or wedge filter) and small FOV can be used to attenuate or eliminate unnecessary radiation to the body periphery.

Figure 2-2. Dual-source CT system. A: First generation (Siemens SOMATOM Definition) with a 50-cm FOV for detector A and a smaller (26-cm) FOV for detector B, due to space limitations. B: Second generation (Siemens SOMATOM Definition Flash) increases the FOV for detector B to 33 cm by increasing the angle between the two tubes from 90 to 94 degrees.

▪ Dual-Energy and Spectral CT Techniques

Recent advances in MDT hardware and detector technology have not only enabled subsecond scanning and wider z-axis coverage but have also revived interest in dual-energy and spectral CT. Today investigators are examining how to use the additional information inherent in the full spectrum of an x-ray beam to add clinical value. Clinical work on the extraction of information from the full spectrum of an x-ray beam has opened new venues for the diagnostic capability of CT by combining x-ray beams acquired at different energy levels or separating of the x-ray beam into multiple energy windows at the detector level. The idea of dual-energy subtraction with using CT is not new and goes back to the 1970s.³⁹ Two separate scans were used in early investigations, but the results were not very promising due to technical limitations causing spatial misregistration, noisy images, and increased radiation exposure. In the 1980s, rapid kVp switching was introduced in which data at high- and low-voltage values were obtained during one scan.²⁹ A rapid surge in clinical application of dual-energy technique occurred after the introduction of dual-source CT in 2000s.³⁰

The dual-energy concept originated from the fact that tissue attenuation values represented by CT numbers are dependent not only on the material and electron density of the attenuating object but also on the energy spectrum of the x-ray beam. At the energy levels of CT, the attenuation mechanism is generally caused by a combination of photoelectric and Compton photon interactions. The photoelectric effect predominates at lower photon energies and has higher probability of occurrence for elements with high atomic numbers (Z), whereas the Compton scattering almost independently depends on the electron density of the material. It should be mentioned that for most tissues in the body, there is a higher probability of Compton interactions than photoelectric interactions in CT. Since Compton interaction probability is fairly constant at CT energy levels, these interactions do not contribute to the dual-energy signal. Nevertheless, when the net effects of all photon interactions are combined, there is a clear energy dependence for the mass attenuation coefficient of high Z materials. Taking advantage of this energy dependence, the dual-energy technique is able to distinguish structures with large differences in atomic numbers.⁴⁰,⁴¹ For example, the iodine signal of a 140-kVp beam (with an effective energy of approximately 60 keV) is approximately twice as strong as a 80-kVp beam (with an effective energy of 40 keV), while the bone signal does not fall as quickly (Fig. 2.3). CT attenuation value of iodine with Z = 53 is much higher for a low-energy beam (i.e., 40 kVp) compared to a high-energy x-ray (140 kVp). This difference is less apparent for calcium with Z = 20. Therefore, in a low-kVp image, the contrast between iodine and calcium is much higher than a highkVp CT image. Although traditional Hounsfield unit (HU) allows discrimination of many general categories (air, water, soft tissue, bone, etc.), there is considerable overlap in HU for many common anatomic materials. For example, HU values of a calcified structure (Z = 20) and a contrast-enhanced material can measure the same. With dualenergy technique, it is possible to characterize tissues by using different energy levels to separate the image into predefined basis material categories. Images from the two energy spectra can also be combined for other applications such as an iodine map, virtual noncontrast imaging, optimal contrast-to-noise image, blended image, etc. (Fig. 2.4). In addition to these image-based dual-energy CT techniques, systems can generate monochromatic images. In other words, the two polyenergetic spectra of the dual-energy raw data signal can be used to create an image that appears as if it was acquired at a single-energy level.³⁸ This requires matching of the projections between the two different kVp data sets and thus is performed in the projection or raw data domain. Raw data-based dual energy CT (DECT) can only be implemented for certain DECT system designs. Since this method calculates the attenuation of base materials prior to the reconstruction, it is, in principle, free of beam-hardening artifacts. The future of multienergy CT is in the spectral decomposition of the signal with photon-counting CT. Spectral CT enables tissues to be discriminated based on relative x-ray attenuation into “energy bands” of the spectrum instead of averaging across the entire polychromatic beam. For photon-counting systems, spectral discrimination usually occurs at detector level using energysensitive photon-counting detectors. Several photon-counting detector CT systems are currently under development.

DECT images are processed using various techniques that can be categorized as either raw data-based or image-based methods. For raw data-based methods, the dual-energy data processing is performed before images are reconstructed from high- and lowenergy sinograms (projection domain). Projection data from the
low- and high-energy spectra (e.g., 80 and 140 kVp) will be decomposed into two basis materials such as water and iodine to form the material density pair. Alternatively, DECT data can be decomposed into energy functions and can be used to generate monochromatic images through sophisticated reconstruction algorithms⁴² (Fig. 2.5). For image-based methods, the dual-energy data processing is performed after the reconstruction of high- and low-energy images (image-based domain). DECT images are created through various mathematical combinations of the low-energy and high-energy images such as subtraction and blending. Monochromatic spectral image processing is not possible with image-based dual-energy methods, only with raw data-based methods. Currently, CT systems from most vendors are capable of dual-energy material decomposition. Some of these systems are also capable of raw data-based
monoenergetic CT image reconstruction. There are three different CT designs to perform dual-energy decomposition (Table 2.1):

Single source with rapid kVp switching (Fig. 2.6A and B).
Dual-source scan in that high- and low-energy beams are produced during one scan (Fig. 2.7).
Single source using sandwich detectors or energy-discriminating photon-counting detectors (Fig. 2.6C). The latter method uses one kVp setting and detects radiation using innovative, spectral CT detectors that separate the x-ray beam into its spectral components (i.e., polychromatic x-ray detection).

Figure 2-3. Attenuation curves for iodine, calcium, and water plotted versus photon energy. The dotted lines represent a typical effective energy for 80- and 140-kVp beams. The difference in attenuation between the effective energies of the spectra at two different energies of 40 and 60 keV (vertical dashed lines) shows that iodine (I) demonstrates a greater decrease in attenuation than calcium (C) in the specified kVp range, whereas the attenuation of water remains almost constant.

Figure 2-4. A: Polychromatic x-ray energy spectrum. B: Dual-energy separation of high (140 kVp) and low energy (80 kVp). C: Photon-counting detectors image the spectrum of x-ray radiation in “energy bands” in order to form images based on the analysis of the spectral signature of tissues. Overlap between both spectra is quite broad, limiting spectral contrast (green spectrum).

Figure 2-5. An example of monochromatic images from 40 to 110 kV created from a dual-energy scan. Display window level and width are 40 and 400 HU, respectively. Optimal level of contrast and minimal noise are seen at 70 to 80 kV. (From Yu L, Christner JA, Leng S, et al. Virtual monochromatic imaging in dual-source dual-energy CT: radiation dose and image quality. Med Phys 2011;38(12):6371-6379, with permission.)

TABLE 2.1 Dual-Energy CT Techniques

Scanner Type	Tube Detector	Advantages	Disadvantages
Single tube with fast KVp switching	Single tube, single detector array	Good temporal and spatial registration Projection-based decomposition Great flexibility Easy to quantify iodine density Large 50-cm FOV	Limited spectral separation between high- and low-energy scans Higher image noise for lowervoltage image for some scanner designs (tube current may not be modulated to equalize dose and noise between both energy levels) Uncertain reliability of HUs on virtual unenhanced images
Dual-source CT	Two tubes, two detector arrays	Good spectral separation between high-and low-energy scans Easy to modulate each tube current to equalize dose and noise between both energy levels HUs can be measured on virtual unenhanced images	Limited temporal and spatial registration (i.e., cardiac imaging) due to time lag between the two acquisitions. Small FOV for dual-energy acquisition (33 cm) Cross scatter artifacts Low-energy image can be limited by higher noise levels Image domain dual-energy decomposition Increased radiation dose
Multilayered detector or photon counting	Single tube, dual-detector layers	Perfect temporal and spatial registration Reasonable radiation dose Large FOV Projection-based dual-energy decomposition	Limited energy separation with substantial spectral overlap. Longer scanning times with current detector technology

Figure 2-6. Illustration of different hardware approaches to dualenergy CT imaging for single-tube scanners. A, B: Show sequential kVp switching. A: The x-ray tube voltage is rapidly modulated to different kVp (140 and 70 kVp) levels during single tube rotation, producing spectra of lower and higher energies. For cardiac imaging, switching is done alternatively at 0.2-millisecond intervals during middiastole for every other heartbeat. B: kVp switching to 140 kVp is performed after finishing all projections at 70 kVp in the first rotation. The switching can be performed slice by slice or volume by volume. C: In energy-sensitive layer detectors technique, the top-layer detector absorbs lower-energy x-ray photons, whereas the bottom-layer detector detects higher-energy x-ray photons. Projections are acquired at 120 kVp only throughout the entire gantry rotation.

The basis material analysis of dual-energy CT images obtained with fast kVp switching can be processed in projection domain. The basis material decomposition for dual-source or dual-layer dual-energy CT scanners is performed only in the image domain.⁴³ Projection domain decomposition is preferred because it enables greater flexibility in material decomposition and permits the preprocessing correction of data to minimize beam-hardening artifacts.⁴³ Although beam-hardening correction techniques are usually applied before image reconstruction, it is not always perfect, and residual artifacts in low kVp especially for dense bones and iodine can be disturbing. The image domain decomposition allows for more readily derived HU measurements.

Rapid kVp Switching with a Single-Source Scanner

There are three types of rapid kVp switching implementations. The first, which will be referred to as “in-plane rapid kVp switching,” uses a conventional MDCT scanner (e.g., 64 detector helical scanner) with a unique x-ray tube and detector system that enables near-simultaneous acquisition of high- (140 kVp) and low-energy (70 kVp) projection data. The x-ray source undergoes rapid kV switching (0.5 milliseconds), which allows for a high degree of coregistration of the two energy acquisitions (Fig. 2.6A). This type of single-source dual-energy CT, introduced by GE (Discovery CT 750 High definition CT (HDCT), GE Healthcare, Milwaukee, WI), utilizes a garnet-based scintillator detector with low afterglow allowing fast sampling rate. The scan allows for a large FOV of 50 cm that covers the whole body. In dual-energy mode, a collimation of 64 × 0.625 mm can be used. The gantry rotation time is slower than single-energy mode (0.8 seconds). While previously limited to a single selection of 600 mA in dual-energy mode, new scanners provide for different selections (600, 375, and 260 mA), thus allowing for patient-optimized technique and dose reduction. This type of in-plane DECT fast kVp switching allows for good matching of the projection data thus permitting raw data-spaced dual-energy analysis. Within the axial plane, there is minimal time for patient motion
between kVps, which is optimal for reducing patient motion between kVp settings. No tube current modulation can be used (limiting dose reduction options), and the same mA must be used for both kVp levels. Since the mA is maintained at a constant value for both kVp levels, it is necessary to run the DECT mode at an mA level that is low enough to minimize excess exposure from the 140-kVp beam while ensuring it is high enough to obtain the necessary image quality from the 80 kVp. As a result, the 80-kVp images may be too noisy, or the 140-kVp images may be generated with unnecessarily high dose.

Figure 2-7. In dual-source CT approach for dual-energy technique separate, x-ray tubes are operated at different kVp levels, allowing simultaneous dual-energy data acquisition. A double-helix trajectory is formed by a helical scan in a dual-source geometry. Because the projection data from source A and source B are never coincident with each other during the helical scan, it is difficult to perform a dual-energy processing in projection data domain. This is one of the reasons why the monochromatic images are currently generated in image space for dual-source CT. (From Yu L, Christner JA, Leng S, et al. Virtual monochromatic imaging in dual-source dualenergy CT: radiation dose and image quality. Med Phys 2011;38(12):6371-6379, with permission.)

The second type of fast kVp switching is “volume rapid kVp switching.” In this implementation, the entire low-kVp image series is acquired in a single rotation, the high-kVp image series is then acquired with a subsequent single rotation. This DECT implementation requires a volume CT scanner such as the Toshiba Aquilion ONE scanner (Toshiba Medical Systems, Otawara, Japan). With 16 cm of detector coverage, volume CT enables image acquisition of whole organs in a single rotation (Fig. 2.6B). This implementation of DECT provides a nearly exact match of projection data with the only mismatch due to patient movement between the volume scans (subsecond interval). As a result, dual-energy mode is used only for noncardiac applications. The kVp switching takes less than 1 second, and scan starts are synchronized. Since projections are matched, raw data-based dual-energy analysis can be used. In-plane tube current modulation can be used for dose reduction. The mA levels for each kVp can be set independently, thus allowing for optimization of image quality for each level and avoiding excess dose. Two volume scans with different kVps are acquired, and mA can be set independently.

The third type of rapid kVp switching is “helical rapid kVp switching.” This implementation, found in Toshiba’s Aquilion PRIME and Aquilion ONE models, uses a different kVp with each rotation during a helical acquisition. When greater than 16 cm needs to be imaged, this scan mode is used in place of the volume rapid kVp switching mode. This implementation allows only for imagebased dual-energy analysis because the projections are not matched to allow for raw data-based analysis. There is a fourth category of kVp switching that does not fall under the umbrella of “rapid kVp switching” because it requires separate helical runs for the lowand high-energy acquisitions. Siemens Somatom Definition Edge and Somatom Definition AS models are single-source scanners and dual-energy mode consists of two successive spiral scans at different kVp and mA levels. This implementation can allow for projection matching and raw data dual-energy analysis, but the long time between scans warrants patient motion considerations.

The main benefit of kVp switching is a more accurate beam hardening correction when raw data-based DECT analysis is available, which becomes clinically apparent in brain, cardiac, and abdominal studies. The disadvantages of the fast kVp switching methods are that the energy separation is currently not as good as using two separate tubes with different filters, and automated tube current modulation techniques can be difficult to implement. The switching time delay precludes the direct combination of low- and high-energy images voxel by voxel to form a standard CT image. The time lag limits application to dynamic events like cardiac CT and suffers the same penalty of higher dose due to the use of an additional low-dose beam.

Pinho et al.⁴² assessed the image quality of virtual monochromatic spectral angiography images obtained with a rapid kVp switching scanner and showed less noise and higher contrast enhancement at a virtual 70-kV monochromatic beam compared with 120-kVp images of a single-source scanner. So et al.⁴⁴,⁴⁵ also used rapid kVp switching technique to assess myocardial perfusion. They were able to obtain high-quality images with less beam-hardening artifact and radiation dose using a prospective ECG-gated technique compared with the traditional singleenergy CT protocol.

Dual-Energy CT with Dual-Source Scanner

By operating both tubes at two different tube voltages, high- and low-energy data can be obtained.³⁷ The advantage of the dualsource design is that different tube currents for each tube can be selected and different filters can be used for the x-ray tubes to get the best energy separation. Increasing mAs for low-kVp tube may be necessary to lower quantum mottle noise levels (Fig. 2.7). This strategy will increase the dose when both tubes are working.

In dual-energy mode, the rotation speed should be reduced, and larger number of subvolumes (slabs) is required. Therefore, the scan time can extend to over 5 seconds, which could be problematic for cardiac perfusion imaging. Using a hybrid algorithm, Nance et al.⁴⁶ performed a cardiac perfusion study with dual-source scanner. They were able to decrease temporal resolution to 83 milliseconds using a hybrid algorithm reconstruction compared with standard algorithm with 165-millisecond temporal resolution while saving image quality.

Dual-source construction significantly improves the interscan time delay seen in the sequential approach. However, simultaneous data sampling and thus projection matching are still not possible since the interscan time delay due to the 90-degree tube offset is not completely eliminated. Another practical limitation is that the FOV is limited to 33 cm. Although this reduced FOV works fine for applications such as cardiac imaging, it may not be suitable for body applications. Since both x-ray tubes operate simultaneously, additional scatter generated artifacts may be created. These artifacts along with excess noise of low-energy tube could result in decreasing conspicuity of small objects. As mentioned earlier, the material separation in dual source is performed in image domain, because the raw data are not perfectly registered (they are acquired at different times, due to the angular offset of the tube/detector systems). As a result, beam-hardening artifacts may be more noticeable.

Single-Source, Dual-Layer Detector CT

In a single-source, dual-layer detector scanner configuration, one x-ray tube is used to expose a detector consisting of two layers (Fig. 2.6C). The first layer on the top absorbs most (approximately 50% of the beam) of the low-energy spectrum, and the layer on the bottom absorbs the remaining higher-energy photons.⁴⁰,⁴¹,⁴³ An advantage to this form of spectral energy separation system is the elimination of the time lag that exists with other dual-energy CT approaches, making it ideal for imaging moving tissue. The use of a single source also obviates the potential scatter limitation of dualsource techniques. Furthermore, if required, this approach allows a full 50-cm FOV without adding the radiation dose levels.

Photon counting represents the most advanced form of spectral CT. Photon-counting detectors count each individual incident x-ray, measuring the energy of each photon.⁴⁷,⁴⁸ Rather than separating the beam into merely high- and low-energy ranges, photon counting uses narrow subranges or bins of the spectrum that can be used to form images and classify spectral “k-edge” patterns of clinically relevant elements and molecules (Fig. 2.4C). Photon counting can potentially enhance the ability to detect these specifically localized molecules at much lower concentrations than the other methods.⁴¹ This technology requires very fast detectors. Because of current limitations in detector design, the CT rotation speed and x-ray tube current of existing prototypes are greatly reduced. The net effect is longer scanning times—several seconds per gantry rotation—and the goal of subsecond scanning is still elusive.

The Clinical Benefits of Dual-Energy CT

Clinical applications of DECT vary widely and many promising applications remain under investigation. DECT has shown promise in replacing true with synthesized unenhanced imaging to
save radiation, improving lesion conspicuity, iodine extraction, and improving tissue/material characterization (e.g., renal stone composition).⁴⁹,⁵⁰ and ⁵¹ In angiographic studies, higher intravascular attenuation on 50- and 70-keV monochromatic images can potentially enable high-quality CT angiography (CTA) studies at lower doses of contrast media.⁵²,⁵³ and ⁵⁴ The use of colored iodine overlay to improve the detection of endoleaks and the subtraction of heavily calcified plaques to improve visualization of vessel stenosis are additional clinical applications. While it is relatively easy to separate iodine in wellopacified vessels from dense calcium in cortical bone or large calcified atherosclerotic plaques, this may not be the case when the density of contrast and calcium is in the same range. It is also technically challenging to separate calcified plaques in small coronary arteries.

Iodine distribution maps obtained by dual-energy techniques allow for accurate assessment of cardiac “perfusion” defects compared with single photon emission CT.³⁷ Cardiac CT perfusion images can be obtained with minimal beam-hardening artifact. This could have important clinical application in order to measure perfusion parameters accurately. The diagnosis of pulmonary embolism can be improved by displaying the iodine distribution in the lung parenchyma.⁵⁵ Regional variation of lung perfusion can be mapped by dual-energy CT in patients suffering from emphysema.⁵⁶ In a recent animal study for liver tumors in rabbits, contrastenhanced dual-energy CT and perfusion CT were compared.⁵⁷ It was shown that viable tumor enhancement values in iodine maps correlated well with tumor perfusion CT parameters and that the radiation dose was less in dual-energy CT.

▪ Influencing Factors in CT Perfusion Imaging

The ideal CT scanner for perfusion imaging is the one with the best possible temporal and spatial resolution that is capable of reconstructing high-quality images of the entire anatomy of interest at multiple time points while delivering the smallest radiation dose to the patient. Current technology allows up to 16 cm of body coverage in less than 300 milliseconds with 0.35-mm isotropic spatial resolution. In addition, a CT perfusion study is not complete without availability of a robust postprocessing method to quantify perfusion parameters with good interoperator reproducibility and great interscan repeatability when follow-up comparisons are required.

Spatial Resolution

One of the greatest challenges in CT imaging is the detailed evaluation of small anatomic structures and visualization of small vessels that requires high-resolution imaging. To achieve this goal, thinner collimation, smaller detector size, and special reconstruction algorithms are required. Spatial resolution is a measure of the accuracy of an imaging system to discriminate between two adjacent highcontrast objects and is usually expressed as line pairs per millimeter.⁵⁸ In CT, voxel size is the major determinant of spatial resolution. A CT scanner with 0.4-mm³ isotropic voxels will depict a 2.2-mm contrast-enhanced vessel on approximately 8 voxels. Typical fluoroscopy tubes have a spatial resolution of approximately 0.16 mm, and the same vessel will be depicted on roughly 20 pixels. With large voxels volume, averaging artifacts occur due to thicker slice thickness. In CT, in-plane spatial resolution describes pixel dimensions in the x-y plane, which depends on the size of reconstruction matrix and on the reconstructed FOV and will be selected by the CT technologist at the time of image reconstruction. This is different from scan FOV, which is the total body area scanned at the time of image acquisition. Currently, CT images by convention have a 512 × 512 matrix size regardless of the reconstructed FOV. A quick way to think about inplane resolution and pixel size in a reconstructed 2D axial CT image is to divide the reconstructed FOV (in millimeters) by 512 for the matrix dimensions. For example, reconstructed FOV = 300 mm/512 provides pixel of roughly 0.6 × 0.6 mm, and reconstructed FOV = 250 mm/512 provides pixel of roughly 0.5 × 0.5 mm.³²

On the other hand, longitudinal (out-of-plane) spatial resolution describes the size of image voxel in z-axis and is related to x-ray collimation chosen and longitudinal dimension of the detectors. Since for most MDCT applications, the smallest collimation is essentially always chosen (i.e., 0.5 mm for cardiac imaging), the main limiting factor for the z-axis of a voxel is determined by the detector thickness. Longitudinal spatial resolution determines the minimum achievable slice thickness. In an ideal volumetric data reconstruction, true isotropic voxels can be reconstructed. That means that a voxel has the same size in all dimensions. Isotropic voxels are required to reconstruct images in all planes at the same high quality. Current CT systems provide an isotropic spatial resolution of up to 0.3 mm. The effect of detector size in the longitudinal spatial resolution is very significant and has become one of the driving forces in the advancement of MDCT technology. Further improvement of spatial resolution for detection of small structures may require a different detector design (i.e., flat panel) and a higher radiation exposure to maintain sufficient signalto-noise ratio (SNR).⁵⁹ Overall, increasing spatial resolution is not a simple task and necessities improvement in all three dimensions. For detailed recording of moving structures as in cardiac imaging, improving the fourth dimension of time or temporal resolution also is required (governed by gantry rotation time, number of x-ray sources, and segmented reconstruction techniques). With the continued evolution of detector technology and increasing z-axis resolution, now seen with some flat panel detectors in development, we are very near a time when maintaining isotropic voxels in reconstructed images may require a 1,024 × 1,024 matrix.

Contrast Resolution and Noise

Contrast resolution in CT is the ability of system to distinguish low-contrast anatomic or pathologic structures in an image.⁶⁰,⁶¹ and ⁶² For example, relative attenuation differences (HU) between gray and white matter in the brain or detectability of cirrhotic nodules relative to the normal liver tissue. Obviously, enhancement of one structure relative to another after intravenous contrast injection will increase soft tissue contrast resolution in post-contrast-enhanced images. The most important parameter influencing contrast resolution is noise. Image noise impairs the detection of lesions. Higher noise levels can be better tolerated in physiologic high-contrast area (e.g., bone-soft tissue interface) than in low-contrast region (e.g., liver). The SNR and contrast-to-noise ratio (CNR) are useful descriptors of CT image quality with respect to the noise. In general, noise in CT depends on the quantum noise (statistical fluctuation in the number of x-ray photons), patient size, quality of the detector system, and the reconstruction kernel (sharper algorithm gives noisier images).⁶⁰ Quantum noise is probably the most influential factor.⁶⁰ Accurate measurement of noise is difficult because noise on CT images is not mathematically uniform and does not follow a defined statistical distribution.⁶¹ Relative background noise in the image can be estimated in the air outside the image by placing a large area of interest (2 cm) in an artifact-free region. Thinner CT slice thickness decreases SNR unless exposure factors such as tube potential and current are increased. This results in an increased radiation dose to the patient. An increase in the SNR by a factor of 1.4 results in doubling of the radiation dose. Alternatively, less noisy images can be obtained by increasing the slice thickness. In perfusion studies of low-contrast regions (e.g., soft tissue tumors), it is suggested to use a slice thickness of not less than 5 mm, which guarantees a correct balance between optimal spatial resolution and acceptable SNR in perfusion scans.⁶³ It is important to realize that larger patients have a higher total x-ray CT attenuation and increasing exposure
factors may be necessary to compensate for photon starvation and decreased SNR. In MDCT, noise increases if pitch is greater than 1 (assuming technique factors are held constant). When pitch is greater than 1, photon-deficient zones can occur in the z-direction causing increasing noise. This phenomenon is less apparent for studies using pitch less than 1 (e.g., cardiac).⁶⁰ In most scanners, tube current-time product is automatically increased in proportion to increasing pitch to keep noise level low. The term “effective” mAs is used by Siemens and equals to mAs/pitch in noncardiac scan modes, and in cardiac mode, the term total mAs delivered per gantry rotation is used, which is defined as mAs/rotation.

High-quality images with clear depiction of low-contrast objects and minimal noise level are necessary for both qualitative and quantitative tissue perfusion analysis. However, radiation dose control in accordance with the ALARA (as low as reasonably achievable) principle demands careful adjustment of the perfusion acquisition protocols to limit the volume of scan to answer the specific clinical question. During image data reconstruction, a “soft” reconstruction kernel may be used to increase visual low-contrast lesion detectability. Thin-cut scanning to minimize partial volume averaging (PVA) effects, along with sliding-thin-slab averaging and adjustment of the window and level settings during image review may be applied to optimize the depiction of low-contrast lesions.⁶¹

CT perfusion studies have been performed at a range of x-ray energies and beam intensities, from 80 to 120 kVp and 80 to 210 mAs, respectively. In most of the recently published studies, in order to reduce the x-ray dose, it is recommended to use relatively low tube voltage (80 to 100 kVp) and current (120 to 200 mA) for perfusion scans.⁶³ In perfusion studies that demand higher image sampling frequency, the tube current may need to be reduced to lower radiation dose. Wintermark et al.⁶⁴ compared the use of 80 and 120 kVp at 200 mAs on quality of brain perfusion images. At 80 kVp, greater contrast between white and gray material was shown and the radiation dose was lower by a factor of 2.8 compared with 120 kVp. The difference in noise between 80- and 120-kVp images was not significant. The 80 kVp setting would also increase the conspicuity of intravenous contrast due to greater participation of the photoelectric effect for 80-kV photons, which are closer to the “k-edge” of iodine.⁶⁵

Only gold members can continue reading. Log In or Register to continue