Recent TechnologicAdvances inMulti -Detector RowCardiac CT

Sandra Simon Halliburton, PhD

KEYWORDS� Computed tomography � Radiation dose� Temporal resolution � Spatial resolution� Scan time � Tissue differentiation

Ever since its inception, there has been a desire toapply x-ray CT to imaging of the cardiovascularsystem. Early CT, however, lacked many technicalcapabilities required for imaging of the heart.

Data acquisition with conventional CT tech-nology is accomplished by mechanically rotatingan x-ray tube and detector array around thepatient. The slow rotational speed of early equip-ment required 300 seconds for the acquisition ofenough data to reconstruct a single image.1 Withsignificant improvements between 1972 and1990,2 this time was reduced to 2 seconds butwas still too long for cardiac imaging. Additionally,data could only be acquired in the axial (ie,sequential or step-and-shoot) mode with limitedlongitudinal or z-coverage per rotation, resultingin very long scan times.

In 1983, electron beam CT technology wasintroduced.3 With this technology, an electronbeam, rather than an x-ray tube, is rotated aroundthe patient. X-rays are produced when the rotatingelectron beam strikes a tungsten target encirclingthe patient. Attenuated x-rays were measured bya stationary detector. This technology, then, elim-inated the requirement for mechanical motion ofan x-ray tube/detector system around the patientand reduced the acquisition time for each axialimage to 50 to 100 milliseconds. Although thetemporal resolution of electron beam CT was

S.S. Halliburton is the recipient of a research grant from Sand IT Division.Imaging Institute, Cardiovascular Imaging, Cleveland ClinUSAE-mail address: hallibs@ccf.org

Cardiol Clin 27 (2009) 655–664doi:10.1016/j.ccl.2009.06.0070733-8651/09/$ – see front matter ª 2009 Elsevier Inc. All

sufficient for cardiac imaging, long scan times,poor spatial resolution, and lack of x-ray powerlimited clinical application.

Parallel advances in conventional CT included,most notably, the introduction of helical (ie, spiral)scanning in 1989.4,5 Although significant improve-ments in volume coverage speed were realized,1 second was still needed to acquire enoughprojection data to reconstruct a single image, pro-hibiting diagnostic imaging of the heart.

The advent of systems with sub-second rotationtimes and ECG-synchronized scanning in 1994brought conventional CT into the domain ofcardiac imaging. Initial results demonstrated theclinical potential of conventional CT, but technicalrestrictions still limited cardiac applicationsprimarily to coronary artery calcium scoring.6,7

Conventional CT did not have a major impact oncardiac imaging until the introduction of multide-tector row CT (MDCT) scanners in 1998,8,9 permit-ting the simultaneous acquisition of four slices,rotation times as short as 500 milliseconds,images as thin as 1.25 mm, and scan times forcoverage of the entire heart equal to 35 to40 seconds. Further improvements were observedwith the introduction of 8-slice scanners in 2001,16-slice in 2002,10,11 64-slice in 2004,12 dual-source scanners in 2006,13 and 320-slice scannerin 2007.14 This evolution of MDCT scanners has

iemens Medical Solutions, Healthcare Sector Imaging

ic, 9500 Euclid Avenue/J1-222, Cleveland, OH 44195,

rights reserved. card

iolo

gy.th

ecli

nics

.com

Halliburton656

resulted in not only an increased number of slicesper rotation, but also rotation times as short as 270milliseconds, images as thin as 0.5 mm, and scantimes for the whole heart ranging from less than1 second to 10 seconds.13,14

Despite the significant technical advances thathave spurred the widespread application of CTto clinical cardiovascular imaging, efforts are stillbeing made to further reduce radiation dose,improve temporal and spatial resolution, decreasescan time, and improve tissue differentiation. Thisarticle discusses the recent technical advancesresulting from these efforts.

RADIATION DOSE

Amid reports of increased radiation exposure tothe public from medical imaging,15 namely CT,and growing concerns about the associated bio-logic risk,16,17 CT scanner manufacturers areaggressively pursuing radiation-exposure reduc-tion techniques. One new feature, adaptive prepa-tient z-collimation, is designed to reduce radiationexposure by preventing x-rays not contributingdata to image formation from reaching the patient,and is applied by default with the selection of thescan protocol. Other strategies modify the x-raytube current in real time according to user-definedvariables, thereby decreasing the number of x-rayphotons interacting with the patient. Still anotherstrategy, the use of noise-reducing reconstructionalgorithms, approaches dose reduction indirectlythrough improved use of the attenuation data.

Fig. 1. Side-view of patient imaged (A) without (B) and wx-rays pass outside the planned scan length and do not cWithout prepatient collimation (A), this results in patienprepatient collimation, only the desired anatomy is exNote: the actual extent of z-overscanning depends on thewidth.

Pre-patient Z-collimation

As the total collimated detector width in z increases(associated with an increased number of detectorrows), the number of x-rays passing outside theplanned scan length and not contributing to imageformation increases.18 Adaptive prepatient z-colli-mators were recently introduced on some CTsystems to reduce x-ray exposure by restrictingthe divergent x-ray beam along the z-axis and pre-venting x-rays outside the planned scan lengthfrom reaching the patient (Fig. 1). The amount bywhich the scan length actually exposed to x-raysexceeds the planned scan length is termed ‘‘z-over-scanning,’’ or overranging. Z-overscanning canoccur with either helical or axial imaging.

In the helical scan mode, reconstruction algo-rithms require extra rotations of the x-ray sourceoutside the desired scan length to obtain sufficientdata for image reconstruction at the start andend.2 X-ray attenuation data generated from theseextra rotations do not provide images at tablepositions outside the planned scan length. Areduction in z-overscanning with helical imagingcan be achieved using a dynamic prepatientz-collimator.19 Opposing collimator blades auto-matically open at the start of helical data acquisi-tion and close at the end of acquisition to blockradiation not contributing to image formation.

In the axial scan mode, z-overscanning occurswhen the desired scan length is not an integermultiple of the total z-collimated detector width.Although differences between the planned scanlength and the exposed scan length are relatively

ith prepatient z-collimation. During scanning, someontribute data to image formation (z-overscanning).t exposure to unnecessary radiation (light red). Withposed (dark red) and z-overscanning is minimized.scan mode (helical or axial) and the total z-detector

Advances in Multi-Detector Row Cardiac CT 657

small, with a narrow total z-axis collimated-detector width, these differences can becomesignificant with increasing z-axis detectorcoverage and constitute a major source of unnec-essary x-ray exposure. One approach to reducingz-overscanning for axial imaging is to provide a setof automatically selectable z-detector collimationswith a range of total widths to more preciselymatch the total z-collimated detector width tomany different scan ranges.19

Fig. 2. Basic helical technique for cardiac imaging withECG-based prospective on-line modulation of thetube current. Data are acquired continuously duringmovement of the patient table. Tube current (dottedline) is at a maximum during imaging of the desiredcardiac phase or phases but is reduced outside thisregion. Some CT scanners permit adjustment of theduration of maximum tube current (double-headeddiamond line) as well as the magnitude of theminimum tube current (double-headed circle line).Data are retrospectively referenced to the ECG-signal(solid line) and reconstructed at the desired timepoints. Vertical rectangles indicate reconstructionperiods. Small horizontal rectangles indicate recon-structed images.

ECG-based X-ray Tube-current Modulation

Recent advances in the prospectively ECG-trig-gered axial technique have allowed application toimaging of the coronary arteries.20–23 With thistechnique, the x-ray tube current is switched ononly during the desired cardiac phase, significantlylimiting x-ray exposure. Appropriate timing of dataacquisition relies on accurate estimates of theexpected length of the upcoming RR interval.Therefore, the main disadvantage of prospectivelyECG-triggered axial data acquisition is vulnera-bility to cardiac motion artifacts in patients withirregular heart rates. Technical developmentsaimed at minimizing cardiac motion artifacts withaxial imaging include collection of additional databeyond the minimum required for image recon-struction (ie, padding) to permit minor retrospec-tive adjustments of the reconstruction window. Itis important to note, however, that this approachsignificantly increases x-ray exposure. In addition,many manufacturers have recently introducedautomated arrhythmia-rejection methods thatpostpone axial data acquisition, if an irregularityis detected, until the heart rate stabilizes.

Despite the significant advances in axialimaging, retrospective ECG-gated helical dataacquisition is still preferred for patients with highor irregular heart rates. With this technique, thex-ray tube current is switched on during the entirescan. The ECG signal is recorded during dataacquisition and used to retrospectively gate datareconstruction to one or more desired cardiacphases. Retrospective referencing of the ECGsignal now allows the deletion of extra systolicbeats, the insertion of missed beats, and the shift-ing of R-peak locations to adjust for arrhythmia onmost CT systems. Retrospectively, ECG-gatedhelical techniques are then less sensitive toheart-rate irregularities than axial techniques.

ECG-based tube-current modulation is em-ployed with helical data acquisition to significantlydecrease x-ray exposure (Fig. 2). The tube currentis at a maximum only during the cardiac phase ofinterest and is reduced substantially outside thisphase. Although data are available during the

entire cardiac cycle, image quality during periodsof low current is limited. User selection of theminimum tube current (eg, 20% versus 4% ofmaximum tube current) was recently introducedto permit additional dose savings in many clinicalsituations.24

ECG-based tube-current modulation, however,imposes limitations on helical imaging of patientswith irregular heart rates. Because tube-currentmodulation is prescribed before scanning,changes in heart rate could result in unintendedlowering of the tube current during a desired phaseof reconstruction for a given cardiac cycle. For thisreason, ECG-based tube-current modulation isoften not used with helical imaging for suchpatients, resulting in very high radiation doses.25

Several manufacturers have recently developednew strategies for ECG-based tube-current modu-lation with helical imaging, increasing the robust-ness of the technique for patients with irregularheart rates. One strategy is to allow user adjust-ment of the maximum tube-current durationbefore scanning; rather than restricting the userto two discrete choices—maximum tube currentduring one phase of the cardiac cycle or maximumtube current during the entire cardiac cycle—theuser is given the flexibility to apply the maximumtube current over a continuous range. A secondstrategy, aimed at improving the utility of ECG-based tube-current modulation for patients withsevere arrhythmia, is the temporary or permanent

Halliburton658

suspension of tube-current modulation if beat-to-beat variation exceeds a threshold value duringdata acquisition. In this case, the risk of improperlytimed downward modulation of the tube current isvirtually eliminated. Although radiation exposurewith respect to traditional approaches to ECG-based tube-current modulation is increased withboth strategies, exposure is still less than if ECG-based tube-current modulation is not used.24

Anatomic-based Tube-current Modulation

The x-ray tube current can also be modulatedautomatically according to patient anatomy toreduce radiation exposure. The tube current canbe modulated along the x-, y-, and z-directionsduring scanning based on local tissue thicknessdetermined from scout imaging without sacrificingimage noise.26 Tube current is reduced at pro-jection angles and table positions requiring lessx-ray penetration. This approach can yield dosesavings with axial imaging but interferes withECG-based modulation and, therefore, has limitedapplication to helical imaging. In practice, ECG-based tube-current modulation is given priorityduring helical imaging and anatomic-based modu-lation serves only to determine the optimal nominaltube current necessary to achieve the desirednoise based on patient attenuation in the scoutimage. Although anatomic-based tube-currentmodulation is well-established, cardiac applica-tions have not benefited from the techniquebecause of the dominance of helical imaging.The recent shift toward axial imaging, however,should mean increased use of anatomic-based

Fig. 3. Flow chart diagramming the basic steps of iterative

tube-current modulation and further reduction incardiac CT dose.

High Pitch

A low-beam pitch (eg, 0.2) is typically required forhelical acquisition of cardiac data because a pitchtoo high for the patient’s heart rate results in thetable moving too far between consecutive cardiaccycles and gaps in the acquired data. This is ofparticular concern for multisegment reconstruc-tion algorithms that use data from two or moreconsecutive cardiac cycles (rather than a singlecardiac cycle) to reconstruct each image.

The high temporal resolution afforded by exist-ing dual source CT (DSCT) technology for cardiacimaging permits single-segment reconstructionand heart rate-dependent pitch. Higher pitchvalues (eg, 0.2–0.5) can be used with higher heartrates to achieve significant dose savings.27

A new application of DSCT technology permitsECG-referenced helical scanning at very high pitchvalues. By interleaving data measured from twodetector systems separated by 90�, pitch can beincreased up to 3.2 within the limited scan field-of-view covered by both detectors.28 Helical scan-ning with such high pitch values reduces theamount of redundant data collected substantiallydecreasing radiation exposure.

Reconstruction Algorithms

One option for decreasing x-ray exposure is theuse of noise-reducing statistical iterative recon-struction algorithms (Fig. 3).29 Iterative recon-struction algorithms were used with the first CT

reconstruction.

Advances in Multi-Detector Row Cardiac CT 659

scanners but were soon rejected as the standardfor CT image reconstruction in favor of faster,less computationally intensive filtered back-projection techniques. Iterative reconstructionalgorithms assume initial attenuation coefficientsfor all voxels and use these coefficients to predictprojection data. Predicted projection data arecompared with actual, measured projection dataand voxel attenuations are modified until the errorbetween estimated and measured projection datais acceptable. Compared with standard analyticalreconstruction methods based on filtered back-projections, statistical iterative reconstructionproduces equivalent signal-to-noise ratios (SNRs)at lower radiation doses without a loss in spatialresolution.29,30 Recent advances in computer-pro-cessing hardware and the increased efficiency ofnew iterative reconstruction algorithms29,30 havesparked the reemergence of these algorithms asan option for CT image reconstruction.

TEMPORAL RESOLUTION

The term ‘‘temporal resolution’’ is used in CTimaging to describe the time needed to acquireenough data for reconstruction of a single image.The temporal resolution needs to be sufficientlyhigh for cardiac imaging to limit cardiac motionartifacts. Because approximately 180� of projec-tion data are required by cardiac algorithms toreconstruct an image for a sufficiently small,centered object, the temporal resolution isapproximated as one-half the gantry rotationtime for single x-ray source CT scanners andone-fourth the gantry rotation time for dual x-raysource scanners. Temporal resolution has signifi-cantly improved from 250 to 400 millisecondswith early MDCT systems, to 75 to 175 millisec-onds with state-of-the-art systems as a result offaster gantry rotation times and inclusion ofa second x-ray source/detector system.

Gantry Rotation

The gantry rotation time is the time required for thex-ray tube/detector system to rotate 360� aroundthe patient. Faster gantry rotation means fasteracquisition of the data needed to reconstructeach image and, subsequently, improvedtemporal resolution. Values for state-of-the-artscanners range from 270 to 350 milliseconds.

Even faster gantry rotation is desired but limitedin part by the mechanical forces (eg, centrifugalforces and frictional forces) acting on the system.One manufacturer recently overcame some ofthe mechanical limitations by developing an air-bearing rotator design that reduces friction andallows faster gantry rotation (270 milliseconds).

Faster gantry rotation is also limited by require-ments on x-ray tube power. As rotation timesshorten it is necessary to increase the x-ray tubecurrent to maintain x-ray flux and achieve thedesired SNR. Although faster gantry rotationshave been possible recently because ofaugmented x-ray power, increased demands ongantry rotation would stretch tube power beyondcurrent limits.31 Finally, a third obstacle to fastergantry rotation is the data transfer rate.

Number of X-ray Sources

An alternative approach for improving temporalresolution is the addition of a second measure-ment system within the gantry. A commercialDSCT system was recently introduced employingtwo x-ray sources mounted opposite two detectorarrays and separated by 90�. The two x-raysource/detector systems rotate together onceevery 330 milliseconds and collect the dataneeded to reconstruct an image in 83 millisec-onds. This approach has obvious advantages forcardiac imaging, given the significant gains intemporal resolution. A challenge facing dual-source compared with single-source CT with thesame z coverage is additional cross-scatteredx-ray radiation, resulting in increased HounsfieldUnit magnitude of common streak and cuppingartifacts and increased noise. Phantom work hasshown that artifacts can be reduced by applyingscatter-correction algorithms and noise can berestored with additional x-ray doses.32 Clinicalstudies have demonstrated, however, that dose-saving features of commercially available DSCTsystems counter the dose cost resulting in imagesof the coronary arteries, with improved imagequality and decreased noise without an increasein radiation dose, compared with single-sourceCT.24 Further improvements in temporal resolutioncan be achieved with additional sources (ie,greater than two) but technical realization is notforthcoming.

SPATIAL RESOLUTION

Improved spatial resolution remains a priority forcardiac CT. The limits of spatial resolution aredictated by scanner geometry, x-ray focus size,aperture and movement during measurement,and detector element size and spacing.2 No signif-icant improvement in in-plane (x–y plane) spatialresolution has been realized with newer comparedwith older MDCT scanners. However, consider-able progress in through-plane (z-axis) resolutionhas been achieved during the evolution ofMDCT, primarily because of the availability ofthinner collimated detector widths. Early MDCT

Halliburton660

scanners provided approximately 1-mm colli-mated detector widths for routine cardiacscanning, but state-of-the-art systems provide0.5-mm to 0.625-mm widths. Further improve-ment in through-plane spatial resolution hasbeen achieved at these collimated detector widthson several newer MDCT systems with the imple-mentation of z-flying x-ray focal-spot techniques.

With the recent enhancement in z resolution,isotropic spatial resolution can be achieved forcardiac CT, typically of 0.5 � 0.5 � 0.5 mm2.Because improvement in spatial resolution isassociated with an increase in image noise (fewerx-ray photons reach thinner detectors), furtherimprovement requires an increase in x-ray expo-sure beyond acceptable levels, an increase indetector efficiency through the development ofnew detector materials or better detector elec-tronics, or implementation of noise-reducingreconstruction algorithms.

Z-flying X-ray Focal Spot

Deflection of the focal spot, or the idealized pointon the surface of the x-ray tube anode from whichx-rays emerge in the x-y plane, has long been usedto double the number of samples acquired acrossthe field-of-view and improve in-plane spatialresolution.2 More recently, several manufacturershave implemented x-ray focal spot deflectionalong the z-axis, such that two consecutivemeasurements are shifted by an amount equal tohalf of the collimated detector slice width. Twointerleaved readings are then acquired with twicethe number of measured values in the z-directionand at half the sampling distance (Fig. 4),

Fig. 4. Deflection of x-ray focal spot along z-axis. Twoconsecutive measurements are shifted by an amountequal to half the collimated detector slice width.Two interleaved readings are acquired with twicethe number of measured values in the z-directionand at half the sampling distance compared withacquisition of a single reading.

compared with acquisition of a single reading perrotation. Using this technique, the theoretical limitof through-plane resolution for 0.6-mm collimateddetector widths can be improved to 0.4 mm.12

Detector Composition

CT systems employ solid-state detectors toconvert x-rays to digital signals. Incident x-raysare converted to visible light by a solid-state scin-tillator material. The visible light is detected bya photodiode and converted into an electricalsignal. The development of scintillator materialwith improved detection efficiency or reducedafterglow might permit better spatial resolutionwithout a noise penalty. Although a new detectormaterial (Lutetium terbium aluminium garnet) wasrecently introduced for commercial use, it offersno improvements in detector efficiency and after-glow compared with existing materials used byother manufacturers (eg, gadolinium-oxide),2 andthus, the material alone is unlikely to raise theindustry standard for spatial resolution withouta significant increase in radiation dose.

Reconstruction Algorithms

Noise-reducing reconstruction algorithms, suchas iterative reconstruction algorithms, offer thegreatest potential for significant improvements inspatial resolution for cardiac CT. Iterative recon-struction algorithms can be used to provideimages with equivalent SNR compared withfiltered back-projection at lower radiation doseswithout a loss in spatial resolution, as describedabove or, alternatively, to provide SNR-equivalentimages at the same radiation dose with improvedspatial resolution.29

SCAN TIME

The term ‘‘scan time’’ is used in CT to describe thetime necessary to acquire all images in thescanned volume. Short scan times permitbreath-holding during data acquisition and reducethe probability of respiratory motion artifacts.Extremely short scan times permit data acquisitionduring a single cardiac cycle and, additionally,reduce the likelihood of cardiac motion artifactsalong the z-axis, such as banding or stair-step arti-facts (incidence of cardiac motion artifacts in-plane is dictated by temporal resolution). Amongthe most dramatic changes in CT during the lastdecade, impacting cardiovascular imaging, is thedecrease in scan time. The scan time for coronaryimaging decreased from approximately 35 to 40seconds with early MDCT scanners to less than1 second with some state-of-the-art systems.14,28

Fig. 5. Helical technique for cardiac imaging with veryhigh pitch. Data are acquired continuously duringmovement of the patient table. Data acquisition istriggered by the ECG signal (solid line) and completedduring a single cardiac cycle. Vertical rectangle indi-cates acquisition period. Small horizontal rectanglesindicate reconstructed images.

Advances in Multi-Detector Row Cardiac CT 661

This leap was accomplished primarily byincreasing z-coverage per rotation through thedesign of detector arrays with increasing numbersof detector rows, and by enabling very high pitchdata acquisition through the simultaneous use oftwo x-ray sources.

Wide Multidetector Arrays

Increased z-coverage per rotation and, subse-quently, decreased scan time, is associated witha greater number of detector rows in a multidetec-tor array. It is important to note that the number ofdetector rows does not necessarily equal thenumber of slices that can be acquired per rotation;some scanners sample each detector twice perrotation (using the z-flying focal-spot techniquedescribed above), such that the number of slicesacquired per rotation is two times the number ofdetector rows used. Z-coverage per rotation atthe isocenter varies significantly across recentlyintroduced scanners; new 128-slice scannersfrom two manufacturers sample 64 detector rowswith 0.6-mm slice collimation twice, yielding3.8 cm of z-coverage per rotation,31 a new 256-slice model samples 128 detector rows with0.625-mm slice collimation twice, achieving 8 cmof z-coverage during one rotation,19 and a 320-detector row system with 0.5-mm slice collimationprovides 16 cm of z-coverage per rotation,sampling the detectors only once.14 Scan timeson these new systems for coronary imaging rangefrom less than 5 seconds22 to less than 1 second.14,31

The 320-detector row system requires only one rota-tion to image the heart for low heart rates, so totalscan time is approximately equal to the temporalresolution at the isocenter or 175 milliseconds.

One of the greatest challenges facing CTimaging with wide multidetector arrays is the tran-sition from fan-beam to cone-beam geometry andthe implications for image reconstruction. Newreconstruction algorithms were recently devel-oped, enabling accurate reconstruction of cone-beam data.33 In addition, like DSCT scanning,scanning with wide multidetector arrays faceschallenges stemming from scattered radiation.X-ray scatter for a DSCT system is comparableto a single-source system with twice thez-coverage.32,34 Early results indicate that thesebarriers to wide-multidetector array imaging havebeen largely overcome with the recently intro-duced clinical systems.14,19,35

High Pitch

As described above, new DSCT technologypermits gapless z-sampling during helical cardiacimaging at extremely high pitch values. The

technology offers 0.28-second gantry rotationtime and approximately 4-cm z-coverage per rota-tion, such that helical imaging with a pitch of 3.2results in 43 cm per second scan speed.31 There-fore, typical cardiac ranges can be scanned in lessthan 0.3 seconds, confining helical data acquisi-tion to a single heart beat (for patients with lowheart rates) (Fig. 5).

TISSUE DIFFERENTIATION

Although CT currently provides good differentia-tion of tissues, there is a desire to push the tech-nology further. Each pixel, or picture element, inthe CT image corresponds to a three-dimensionalvoxel, or volume element, within the patient. Eachimage pixel displays a tissue attenuation coeffi-cient describing the average attenuation ofx-rays within each corresponding patient voxel.Tissue-attenuation coefficients depend on thecomposition and density of the tissue within thevoxel, as well as the x-ray photon energy.

X-rays at energies within the diagnostic imagingrange primarily interact with matter via Comptonscattering and the photoelectric effect. Photoelec-tric emissions dominate at lower x-ray energylevels and depend heavily on the atomic numberof the material, while Compton scatter dominatesat higher energies and is more dependent on thedensity of the material.36 Therefore, different

Fig. 6. X-ray spectra generated by applying a peaktube voltage of 80 kVp across an x-ray tube (blue),applying a peak tube voltage of 140 kVp across anx-ray tube (red), and applying a peak tube voltageof 140 kVp across an x-ray tube and filtering out thelowest energy photons (green). Filtering of the high-energy spectrum yields greater separation of thelow- and high-energy spectra.

Halliburton662

attenuation values are observed for the samematerial at different x-ray energies.

CT imaging typically relies on one set of attenu-ation data obtained using a broad x-ray spectrum,with energies ranging from 20 keV to 150 keV,depending on the peak voltage applied acrossthe x-ray tube. Dual-energy imaging describesthe acquisition of two spectrally distinct attenua-tion data sets and has been studied since thelate 1970s36,37 as a potential tool for improvedtissue discrimination with CT. Significantresources have recently been devoted to providingdual-energy techniques on clinical CT systems.

One approach to dual-energy imaging is the useof two x-ray tubes to simultaneously apply twox-ray spectra, each generated using a differentpeak x-ray tube voltage (eg, 80 kVp and 140 kVp),and therefore spanning different energy ranges.Two data sets with differing attenuation character-istics are then obtained from the region of interest.Dual-energy techniques using this approach arecommercially available and have already beenapplied to imaging of the heart.38,39 Although prom-ising, the ability to discriminate certain tissues islimited by the overlap of attenuation data, resultingat least in part from overlap of the low- and high-energy spectra.40 A new, recently introducedDSCT system uses an x-ray filter to remove thelowest energy photons from the high energy (eg,140 kVp) spectrum. This yields greater separationof the low- and high-energy spectra and mayreduce the overlap in attenuation values observedwithout the energy selective filter (Fig. 6).

A second approach to dual-energy CT imaginguses a single x-ray tube but alternates betweentwo peak tube voltages during data acquisitionand acquires two complete sets of projectionsduring each gantry rotation. A third approach op-erates one x-ray tube at a single peak tube voltageto generate one distinct x-ray spectrum and useslayers of energy-sensitive detectors to captureboth a low-energy and high-energy set of attenu-ated x-ray photons; lower energy radiation isabsorbed by the top detector and higher energyradiation is absorbed by the bottom detector.Phantom work has demonstrated improved visual-ization of calcified coronary plaque41 and coronaryartery stent lumen42 using dual energy.

Furthermore, from commercial realization is theuse of photon-counting detection systems toacquire more than two spectrally distinct measure-ments. All of the approaches to dual-energyimaging described above are limited by theconventional CT detection systems used. Conven-tional systems integrate incoming x-ray photonswithout regard to spectral characteristics, inher-ently restricting energy resolution.43 Photon-

counting technology, however, permits discrimina-tion of incoming photons; each x-ray event isassigned to one of multiple energy bins. Imagesare reconstructed using data from a single bin ora combination of bins. Preclinical results from mul-tienergy CT have demonstrated the feasibility ofusing photon-counting detectors44,45 and thepotential for improvement in luminal depiction andplaque characterization.45 However, challengesincluding limited gantry rotation speed because ofslow count rates will delay clinical application,particularly to imaging of the heart.

SUMMARY

Recent technical innovations in MDCT have re-sulted in lower radiation dose, improved temporaland spatial resolution, decreased scan time, andimproved tissue differentiation.

The refinement of axial data acquisition for coro-nary imaging has enabled its use, rather thanhelical data acquisition, in coronary CT patientswith low, stable heart rates, resulting in significantdose savings. Some systems permit anatomic-based tube-current modulation with ECG-triggered axial scanning, which can result inadditional dose savings. For patients with contra-indications to prospectively ECG-triggered axialimaging, the expanded flexibility of ECG-basedtube-current modulation has resulted in lowerdoses with helical imaging.

Advances in Multi-Detector Row Cardiac CT 663

Other advances aimed at lowering radiationdose for both axial and helical acquisitions includethe use of adaptive prepatient z-collimators. SomeCT systems use prepatient z-collimation to reducex-ray exposure by preventing x-rays outside theplanned scan length, but not contributing diag-nostic information, from reaching the patient. Inaddition, implementation of iterative reconstruc-tion algorithms is being explored for CT imagingin hopes of achieving a reduction in radiationdose without a loss in image quality.

Recent gains in temporal resolution have beenachieved by some manufacturers by decreasingthe gantry rotation time or increasing the numberof x-ray sources from one to two. Recent improve-ments in spatial resolution have been lessdramatic, although application of the flying focalspot technique in the z-direction has served toenhance z-resolution. Iterative reconstructionalgorithms, however, offer the greatest potentialfor significant improvements in spatial resolutionfor cardiac CT.

Significant decreases in scan time have beenaccomplished with newer MDCT systems byincreasing z-coverage per rotation through thedesign of detector arrays with increasing numbersof detector rows. A new approach to decreasingscan time is to significantly increase pitch throughthe simultaneous use of two x-ray sources. X-rayexposure is decreased by decreasing the amountof redundant data collected. Although extremelypromising, this approach still requires clinicalvalidation.

Attempts to gain spectral information from CTand improve tissue discrimination have resultedin the implementation of multiple approaches todual-energy data acquisition. Early results demon-strate the added value of the additional attenuationinformation. The extraction of even more spectralinformation beyond that provided by dual-energyimaging may be possible through the use ofphoton-counting detectors.

REFERENCES

1. Hounsfield GN. The E.M.I. scanner. Proc R Soc Lond

B Biol Sci 1977;195:281–9.

2. Kalendar WA. Computed tomography: fundamen-

tals, system technology, image quality, and applica-

tions. Germany: Publicis MCD Verlag; 2006.

3. Boyd DP, Lipton MJ. Cardiac computed tomog-

raphy. Proc IEEE 1983;71:298–307.

4. Kalendar WA, Seissler W, Klotz E, et al. Spiral volu-

metric CT with single-breath-hold technique, contin-

uous transport, and continuous scanner rotation.

Radiology 1990;176:181–3.

5. Crawford CR, King KF. Computed tomography scan-

ning with simultaneous patient translation. Med Phys

1990;17(6):967–82.

6. Carr JJ, Crouse JR, Goff DC, et al. Evaluation of sub-

second gated helical CT for quantification of coro-

nary artery calcium and comparison with electron

beam CT. Am J Roentgenol 2000;174:915–21.

7. Becker CR, Jakobs TF, Aydemir S, et al. Helical and

single-slice conventional CT versus electron beam

CT for the quantification of coronary artery calcifica-

tion. Am J Roentgenol 2000;174:543–7.

8. Klingenbeck-Regn K, Schaller S, Flohr T, et al. Sub-

second multi-slice computed tomography: basics

and applications. Eur J Radiol 1999;31:110–24.

9. Hu H, He HD, Foley WD, et al. Four multidectector-

row helical CT: image quality and volume coverage

speed. Radiology 2000;215:55–62.

10. Flohr T, Stierstorfer K, Bruder H, et al. New technical

developments in multislice CT, part 1: Approaching

isotropic resolution with sub-millimeter 16-slice

scanning. Rofo: fortschritte auf dem Gebiet der

Rontgenstrahlen und der bildgebenden Verfahren

2002;174(7):839–45.

11. Flohr T, Bruder H, Stierstorfer K, et al. New technical

developments in multislice CT, part 2: Sub-millimeter

16-slice scanning and increased gantry rotation

speed for cardiac imaging. Rofo: fortschritte auf

dem Gebiet der Rontgenstrahlen und der bildge-

benden Verfahren 2002;174(8):1022–7.

12. Flohr T, Stierstorfer K, Raupach R, et al. Perfor-

mance evaluation of a 64-slice CT system with

z-flying focal spot. Rofo 2004;176(12):1803–10.

13. Flohr TG, McCollough CH, Bruder H, et al. First

performance evaluation of a dual-source (DSCT)

system. Eur Radiol 2006;16:256–68.

14. Rybicki FJ, Otero HJ, Steigner ML, et al. Initial eval-

uation of coronary images from 320-detector row

computed tomography. Int J Cardiovasc Imaging

2008;24:535–46.

15. Mettler FA, Huda W, Yoshizumi TT, et al. Effective

doses in radiology and diagnostic nuclear medicine:

a catalog. Radiology 2008;248:254–63.

16. Einstein AJ, Henzlova MJ, Rajagopalan S. Esti-

mating risk of cancer associated with radiation

exposure from 64-slice computed tomography coro-

nary angiography. JAMA 2007;298(3):317–23.

17. Brenner DJ, Hall EJ. Computed tomography—an

increasing source of radiation exposure. N Engl J

Med 2007;357:2277–84.

18. Tzedakis A, Damilakis J, Perisinakis K, et al. The

effect of z overscanning on patient effective dose

from multi-detector helical computed tomography

examinations. Med Phys 2005;32(6):1621–9.

19. Walker MJ, Olzsewki ME, Desai MY, et al. New radi-

ation dose saving technologies for 256-slice cardiac

computed tomography angiography. Int J Cardio-

vasc Imaging 2009;25:189–99.

Halliburton664

20. Earls JP, Berman EL, Urban BA, et al. Prospectively

gated transverse coronary CT angiography versus

retrospectively gated helical technique: improved

image quality and reduced radiation dose. Radi-

ology 2008;246:742–53.

21. Stolzmann P, Leschka S, Scheffel H, et al. Dual-

source CT in step-and shoot mode: noninvasive

coronary angiography with low radiation dose. Radi-

ology 2008;249:71–80.

22. Weigold WG, Olszewski ME, Walker MJ. Low-dose

prospectively gated 256-slice coronary computed

tomography angiography. Int J Cardiovasc Imaging

2009;25:217–30.

23. Kitagawa K, Lardo AC, George RT. Prospective

ECG-gated 320 row detector computed tomog-

raphy: implications for CT angiography and perfu-

sion imaging. Int J Cardiovasc Imaging, in press.

24. Halliburton SS, Sola S, Kuzmiak SA, et al. Impact of

dual source cardiac computed tomography on

patient radiation dose in a clinical setting: compar-

ison to single source imaging. J Cardiovasc Comput

Tomogr 2008;2(6):392–400.

25. Hausleiter J, Meyer T, Hermann F, et al. Estimated

radiation dose associated with cardiac CT angiog-

raphy. JAMA 2009;301:500–7.

26. Mulkens TH, Bellinck P, Baeyaert M, et al. Use of an

automatic exposure control mechanism for dose

optimization in multi-detector row CT examinations:

clinical evaluation. Radiology 2005;237:213–23.

27. McCollough CH, Primak AN, Saba O, et al. Dose

performance of a 64-channel dual-source CT

scanner. Radiology 2007;243:773–84.

28. Petersilka M, Bruder H, Krauss B, et al. Technical

principles of dual source CT. Eur J Radiol 2008;

68(3):362–8.

29. Thibault J, Sauer KD, Bouman CA, et al. A three-

dimensional statistical approach to improved image

quality for multislice helical CT. Med Phys 2007;

34(11):4526–44.

30. Wang G, Yu H, De Man B. An outlook on x-ray CT

research and development. Med Phys 2008;35(3):

1051–64.

31. Flohr T, Raupach R, Bruder H, et al. How much can

temporal resolution, spatial resolution and volume

coverage be improved? J Cardiovasc Comput

Tomogr 2009;3(3):143–52.

32. Engel KJ, Herrman C, Zeitler G. X-ray scattering in

single- and dual-source CT. Med Phys 2007;35(1):

318–32.

33. Mori S, Endo M, Komatsu S, et al. A combination-

weighted Feldkamp-based reconstruction algo-

rithm for cone-beam CT. Phys Med Biol 2006;51:

3953–65.

34. Endo M, Mori S, Tsunoo T. Magnitude and effects of

x-ray scatter in a 256-slice CT scanner. Med Phys

2006;33(9):3359–68.

35. Hameed TA, Teague SD, Vembar M, et al. Non-coro-

nary CT chest angiography utilizing low radiation

dose ECG gated imaging techniques of a 256 slice

multidetector CT scanner. Int J Cardiovasc 2009;25:

267–78.

36. McCullough EC. Photon attenuation in computed

tomography. Med Phys 1975;2:307–20.

37. Alverez RE, Macovski A. Energy-selective recon-

structions in x-ray computerized tomography. Phys

Med Biol 1976;21:733–44.

38. Ruzsics B, Lee H, Zwerner PL, et al. Dual-energy CT

of the heart for diagnosing coronary artery stenosis

and myocardial ischemia-initial experience. Eur

Radiol 2008;18(11):2414–24.

39. Schwarz F, Ruzsics B, Schoepf UJ, et al. Dual-

energy CT of the heart—principles and protocols.

Eur J Radiol 2008;68:423–33.

40. Barreto M, Schoenhagen P, Nair A. Potential of

dual energy computed tomography to characterize

atherosclerotic plaque: ex vivo assessment of

human coronary arteries in comparison to

histology. J Cardiovasc Comput Tomogr 2008;

2(4):234–42.

41. Boll DT, Merkle EM, Paulson EK. Coronary stent

patency: dual-energy multidetector CT assessment

in a pilot study with anthropomorphic phantom.

Radiology 2008;247(3):687–95.

42. Boll DT, Merkle EM, Paulson EK, et al. Calcified

vascular plaque specimens: assessment with

cardiac dual-energy multidetector CT in anthropo-

morphically moving heart phantom. Radiology

2008;249(1):119–26.

43. Alvarez RE, Seibert JA, Thompson SK. Comparison

of dual energy detector system performance. Med

Phys 2004;31(3):556–65.

44. Shikhaliev PM. Energy-resolved computed tomog-

raphy: first experimental results. Phys Med Biol

2008;53:5595–613.

45. Feuerlein S, Roessl E, Proksa R. Multienergy photon-

counting K-edge imaging: potential for improved

luminal depiction in vascular imaging. Radiology

2008;249(3):1010–6.