O

Di

ALSD

a

a

F

1d

Journal of Cardiovascular Computed Tomography (2008) 2, 172–180

riginal Research Article

ual-source coronary computed tomography angiographyn patients with atrial fibrillation: initial experience

rik Wolak, MDa, Ariel Gutstein, MDa, Victor Y. Cheng, MDa, Yasuyuki Suzuki, MDa,ouise E. Thomson, MBChBa, John Friedman, MDa,b, Damini Dey, PhDa,b,ean W. Hayes, MDa, Piotr J. Slomka, PhDa,b, Guido Germano, PhDa,b,aniel S. Berman, MDa,b,*

Departments of Imaging and Medicine, and CSMC Burns & Allen Research Institute, Cedars-Sinai Medical Center,

nd b Department of Medicine, University of California at Los Angeles, Los Angeles, CA, USA

BACKGROUND: Patients with atrial fibrillation (AF) are generally excluded from coronary CTangiography (CCTA) studies because of motion artifact resulting from irregular rhythm. The 83-millisecond temporal resolution of the dual-source CT (DSCT) may be sufficient to allow CCTA inpatients with AF.

OBJECTIVE: We examined the feasibility of DSCT in patients with AF referred for CCTA.METHODS: We compared results of CCTA with DSCT in 24 consecutive patients with AF with 119

control patients in sinus rhythm. Standard relative-delay phase reconstruction (40%– 80% of cardiaccycle) was used, with additional absolute delay reconstruction performed when indicated. Image qualitywas scored both subjectively and objectively.

RESULTS: Patients with AF were older (68.5 � 14.0 years versus 62.5 � 12.1 years; P � 0.03).Maximum heart rate during injection was 102.5 � 30.4 beats/min and 70.8 � 16.6 beats/min in the AFand control groups, respectively (P � 0.01). Mean (�SD) Agatston score was 321 � 366 (range,0 –1158) and 361 � 743 (range, 0 –3948) in the AF and control groups, respectively (P � 0.8). Nodifference was observed in the proportion of uninterpretable segments between the 2 groups, 7 (2%) inthe AF group and 12 (1%) in the control group (P � NS). Two (8%) of 24 studies in the AF group and12 (10%) of 119 studies in the control group were nondiagnostic (P � NS). Image quality was goodor excellent in 13 (54%) of 21 AF cases compared with 94 (79%) of 119 control cases (P � 0.01).Absolute delay reconstruction was needed in 9 (38%) of 24 AF cases.

CONCLUSIONS: These preliminary data show that interpretable CCTA data can be obtained inpatients with AF using DSCT. The need for absolute delay reconstruction is common.© 2008 Society of Cardiovascular Computed Tomography. All rights reserved.

KEYWORDS:Angiography;Atrial fibrillation;Coronary computedtomography

Conflict of interest: The authors report no conflicts of interest.Dr. Wolak and Dr. Gutstein contributed equally to this work.Dr. Wolak and Dr. Gutstein are Save A Heart Foundation, Inc, fellows at Cedars Sinai Medical Center, Los Angeles, CA, and American Physicians

ellowship, Boston, MA.This study was supported in part by grant 6318 from the Glazer Foundation, Los Angeles, CA, and in part by The Lincy Foundation, Los Angeles, CA.* Corresponding author.E-mail address: bermand@cshs.org

Submitted November 15, 2007. Accepted for publication March 19, 2008.

934-5925/$ -see front matter © 2008 Society of Cardiovascular Computed Tomography. All rights reserved.oi:10.1016/j.jcct.2008.03.003

I

pmtoApwdcdaaw

sarirctfadptfcpcn

siCr

M

P

pNrCdas2c

Ti

C

(mdsDd9i

6Cic�nasik(bflop0acf7pmAcmc

tovtg

I

nurp

173Wolak et al CCTA in patients with atrial fibrillation

ntroduction

Atrial fibrillation (AF) is a common arrhythmia, with arevalence of �5% in patients older than 69 years.1 Esti-ates are that 2.2 million Americans have atrial fibrilla-

ion.1 Because atherosclerotic cardiovascular disease is onef the most common causes of AF,2 patients presenting withF frequently undergo noninvasive imaging to evaluate theresence and extent of coronary artery disease (CAD). Itas reported that patients with AF with stress perfusionefects on myocardial perfusion single-photon emissionomputed tomography (SPECT) have a higher risk of car-iac death than do patients without AF.3 It is reasonable tossume that an accurate noninvasive test for excluding CADnd for detection and evaluation of CAD in patients with AFould be valuable in patient management.Although coronary CT angiography (CCTA) using 64-

lice multidetector CT has become a clinically useful ex-mination in patients with low heart rate and a regularhythm,4–7 this method has been considered contraindicatedn patients with AF because of its association with moreapid rates and characteristic beat-length irregularity. Inomparison, dual-source CT (DSCT) has an 83-millisecondemporal resolution, significantly reducing artifact resultingrom coronary artery motion during CCTA.8 In addition, thedjunct computer workstation provides manual electrocar-iogram (ECG) editing that allows selection of any desiredortion of the cardiac cycle to be used in data reconstruc-ion. On the basis of these characteristics, CCTA may beeasible with DSCT in patients with AF. Oncel et al9 re-ently described excellent performance of DSCT in com-arison to invasive angiography in patients with AF andoncluded that DSCT technology shows potential for diag-ostic noninvasive coronary angiography in such patients.

Given the improved temporal resolution of DSCT, weought to describe the technical features of performing CCTAn patients with AF and to determine the image quality ofCTA in AF in comparison to patients with normal sinus

hythm.

ethods

atient population

The study included CCTA data sets from 24 consecutiveatients with AF referred to our laboratory for CCTA fromovember 2006 to November 2007. Indications for CCTA

eferral included chest pain, shortness of breath, multipleAD risk factors, abnormal or equivocal SPECT myocar-ial perfusion study, and aortic or aortic valve disease. Andditional 119 consecutive patients in normal sinus rhythmtudied during 2 months within the study period (May–June007) comprised the control group. No patients were ex-

luded. Clinical characteristics of the patients are shown in m

able 1. The study was approved by the Cedars-Sinai Med-cal Center Institutional Review Board.

CTA acquisition

All CT scans were obtained by a 64-slice DSCT scannerDefinition; Siemens Medical Solutions, Forchheim, Ger-any) with gantry-rotation time of 330 milliseconds and stan-

ard detector collimation of 0.6 mm. By using 2 different focalpots (z-Sharp Technology; Siemens Medical Solutions),SCT acquires 64 overlapping 0.6-mm slices using two 32-etector rows.8 By using dual x-ray tube and detectors with0-degree separation, a temporal resolution of 83 millisecondss achieved.

Attempts were made to maintain the patient’s heart rate �5 beats/min with �-blockade before contrast infusion forCTA. �-Blockers were given orally (50–100 mg atenolol),

ntravenously (15–50 mg metoprolol), or by both routes unlessontraindicated or if the baseline heart rate before the scan was60 beats/min. After a single sublingual dose of 0.4 mg

itroglycerin for coronary vasodilation, all patients underwentn initial non–contrast CT acquisition for coronary calciumcoring, with prospective ECG triggering at 70% of the R-Rnterval, 350-mm field of view, 2.5-mm slice thickness, 120Vp, and 42 mAs. Then, 80–90 mL intravenous contrastOptiview 350) was administered at a rate of 5 mL/s, followedy another 12 mL of contrast at a rate of 1 mL/s during a salineush. With the use of the bolus tracking technique for the startf scanning, retrospective ECG-gated CCTA acquisition waserformed during a single breathhold, with a slice thickness of.6 mm, heart rate–dependent pitch, 120-kVp tube voltage,nd effective current of 350 mAs. For the 119 patients in theontrol group, tube current modulation was applied with theollowing settings: tube current was maximal during 45%–5% of the R-R interval and 4% of maximal for all remainingarts of the cardiac cycle. Prospective ECG-based tube currentodulation (ECG pulsing) was not used in the patients withF to provide maximal visualization throughout the cardiac

ycle. The effective radiation dose (in mSv) was calculated byultiplying recorded dose-length product by the conversion

oefficient 0.017 mSv · mGy�1 · cm�1.10

CCTA data were acquired from the bifurcation of the tracheao the level of the costophrenic angles. For patients with a historyf coronary bypass surgery or requiring aortic imaging, the field ofiew extended from the level clavicle to the diaphragms to includehe origin of the internal mammary arteries, proximal and distalraft anastomoses, and the aortic arch.

mage reconstruction

Raw data were reconstructed with 0.60-mm slice thick-ess, 0.3-mm slice increment, and 250-mm field-of-view,sing single-segment reconstruction and a medium smootheconstruction kernel (B26f). A standard relative delayhase (40%–80%) reconstruction algorithm at 5% incre-

ent was the default reconstruction algorithm for all stud-

iwiamtdrttMsra(msop�

I

ia

S

rwHa

scdftpon

�c(lmcsaottc

w(atglimb((

174 Journal of Cardiovascular Computed Tomography, Vol 2, No 3, May/June 2008

es. In all the cases in which image quality was suboptimal,e performed sequential absolute delay reconstruction us-

ng the following approach: data were initially reconstructedt �300 millisecond, then, if necessary, at �250 and �200illisecond. If prior reconstructions were still not satisfac-

ory, we performed manual reconstruction by redirectingata sampling to the end of the T wave (end-systole) or byemoving data sampling just before the premature beat withhe shortest coupling interval. Reconstructed data wereransferred to a separate workstation (Leonardo; Siemens

edical Solutions, Forchheim, Germany) for vessel analy-is. Left ventricular analysis was performed with a sepa-ately reconstructed data set with 1.5-mm slice thicknessnd reduced matrix size of 256, using commercial softwareCirculation; Siemens Medical Solutions, Forchheim, Ger-any). The non–contrast CT series for coronary calcium

coring was transferred to another workstation (SciImage)n which the Agatston score (AS) was calculated.11 Foratients with coronary stents or coronary plaques with AS

100, a sharp reconstruction kernel (B46f) was also used.

mage analysis

CCTA images were evaluated by assessing standard ax-al, sagittal, and coronal views, multiplanar reformations,nd thin-slab maximum intensity projections.

ubjective image quality assessment

All data sets were evaluated by 2 physicians with expe-ience in cardiac CT, using consensus reading. Segmentsere assessed on the basis of the 15-segment Americaneart Association model.12 Each segment of the coronary

Table 1 Patient characteristics

Atrial fibrillation

Age, mean � SD (range), y 68.5 � 14.0 (40–Male sex, n/total (%) 17/24 (71)BMI, mean � SD (range), kg/m2 27.6 � 6.1 (18.6Indication, n (%)

Chest pain 14 (58)Shortness of breath 4 (17)Risk factors for CAD or knownCAD

9 (38)

Aortic or valvular disease 4 (17)History, n (%)

Diabetes 2 (8)Hypertension 18 (75)Dyslipidemia 16 (67)Smoking 7 (29)Prior myocardial infarction 3 (13)Prior stent implantation 2 (8)Prior CABG 2 (8)

BMI, body mass index; CABG, coronary artery bypass graft; CAD, cor

rtery tree was graded for the presence and type of plaque or d

tenosis and for the presence and type of artifacts. Signifi-ant CAD was defined as visual assessment of luminaliameter stenosis � 50%. Artifacts were classified into theollowing 3 categories: motion or cardiac phase misregis-ration, beam hardening or blooming because of calcifiedlaque, and other artifacts, eg, because of large patient sizer contrast timing error (unfavorable contrast or signal-to-oise ratio [SNR]).

Image quality was classified for each segment as being “4excellent” (absence of artifacts related to motion or coronary

alcification), “3 � good” (minor artifacts), “2 � moderate”considerable artifacts but maintained visualization of arterialumen), or “1 � poor” (uninterpretable because of severeotion artifacts, extensive wall calcification, or inadequate

ontrast-to-noise ratio (CNR) across all phases).13,14 Overalltudy quality was defined by the lowest segmental quality forny segment � 1.5 mm. A nondiagnostic study was defined asne with any coronary artery segment � 1.5 mm in diameterhat could not be evaluated. Stented segments were included inhe analysis. Bypass grafts were excluded, and only the nativeoronary system was analyzed.

Left ventricular analysis data (20 phases at 5% interval)ere retrospectively evaluated by additional 2 readers

A.W. and A.G.) to determine the optimal phase(s) forssessment of coronary arteries and verification of artifactype. For the purpose of this analysis, segments wererouped by vessel into 6 proximal/mid and distal groups:eft main, proximal and mid left anterior descending (LAD;ncluding the first diagonal branch, D1) artery; proximal andid left circumflex (LCX; including the first marginal

ranch, M1) artery; proximal and mid right coronary arteryRCA); distal LAD artery (including D2); distal LCX arteryincluding M2); and distal RCA (including the posterior

4) Sinus rhythm (n � 119) P value

62.5 � 12.1(28–90) 0.0383/119 (70) 0.9226.9 � 4.9 (18.2–48.7) 0.55

46 (39) 0.1110 (8) 0.2555 (46) 0.29

17 (14) 0.76

20 (17) 0.3766 (56) 0.1174 (63) 0.8211 (9) 0.018 (7) 0.3912 (10) 17 (6) 0.65

rtery disease.

(n � 2

90)

–41.1)

onary a

escending artery and posterior-lateral branch). All 20

podtpia

O

aditntta

H

ashM

mhnpseais

S

UtedstePw

R

175Wolak et al CCTA in patients with atrial fibrillation

hases were reviewed in a systematic fashion to identify theptimal phase for evaluating the coronary segment groupsescribed above. “Optimal phase” was defined as the phasehat showed the fewest artifacts. Diastolic and systolichases were defined as 45%–0% and 5%–40% of the R-Rnterval, respectively. Two patients were excluded from thisnalysis because of missing data.

bjective image quality assessment

Objective image quality was assessed on 0.6-mm slices inll patients, as described previously.15,16 Image noise wasefined as the standard deviation (SD) of signal in the region ofnterest placed in the ascending aorta. SNR was calculated ashe ratio between signal in the coronary artery and imageoise.15,16 Vessel contrast was defined as the difference be-ween signal in the coronary artery and surrounding adiposeissue.15,16 CNR was calculated as the ratio of vessel contrastnd image noise.15,16

eart rate analysis

Heart rate on the ECG acquired before CCTA was defineds the initial heart rate. ECG recordings during CCTA acqui-ition were used to determine maximal heart rate, minimaleart rate, mean heart rate, and maximal heart rate variability.

Table 2 Finding on coronary computed tomography angiograp

Atrial fibril

Uninterpretable coronary segments, n/total (%)

7/360 (2)

Subjective study quality, n (%)Poor 2 (8)Moderate 9 (38)Good 9 (38)Excellent 4 (16)

Objective study quality, mean � SDSignal-to-noise ratio 14.2 � 4.1Contrast-to-noise ratio 17.1 � 4.8Noise 29.9 � 6.6

Agatston score, mean � SD (range) 321 � 366Significant CAD (�50% stenosis), n (%) 5 (21)Any CAD, n (%) 18 (75)Sublingual nitroglycerin, n (%) 22 (92)�-Blockers before CT scan, n (%) 13 (54)Initial heart rate, mean � SD (range),

beats/min54.2 � 13.

Maximal heart rate, mean � SD (range),beats/min

102.5 � 30

Mean heart rate mean � SD, beats/min 74.1 � 14.Maximal heart rate variation, mean � SD,

beats/min47.8 � 27.

Radiation dose, mean � SD, mSv 26.2 � 7.8

CAD, coronary artery disease; CT, computed tomography.

aximal heart rate was derived from the shortest R-R interval, c

inimal heart rate from the longest R-R interval, and meaneart rate from the ratio of all R-R intervals divided by theumber of R-R intervals on the ECG. Heart rate variation wasreviously described as the SD of all R-R intervals in thetudy.13 However, because this definition may not reflect theffect of short coupling intervals often seen in AF, we used anlternate definition, the difference between minimal and max-mal heart rates during acquisition, to account for the effect ofhort coupling intervals.

tatistical analysis

All continuous variables are expressed as mean � 1 SD.npaired t tests were used to compare differences in con-

inuous data, and variance was tested by the Levene test ofquality of variance. Chi-square tests were used to compareifferences in discrete data and in cases. The Pearson chi-quare test was used for comparisons whenever both groupshat contained �6 patients. In all other cases we used Fisherxact test. All statistical tests were 2-tailed, and a value of� 0.05 was considered significant. All data were analyzedith SPSS version 12 (SPSS Inc, Chicago, IL).

esults

Patients with AF were older and had higher rate of

(n � 24) Sinus rhythm (n � 119) P value

18/1785 (1) 0.76

0.0112 (10)13 (11)62 (52)32 (27)

12.2 � 4.9 0.0915.0 � 5.7 0.1235.1 � 12.7 0.02

58) 361 � 743 (0–3948) 0.824 (20) 184 (71) 0.271131 (95) 0.6281 (68) 0.19

88) 62.4 � 11.7 (39–90) �0.01

–173) 70.8 � 16.6 (45–153) �0.01

56.5 � 9.2 �0.018.4 � 10.7 �0.01

8.1 � 3.7 �0.01

hy

lation

(0–11

9 (36–

.4 (51

74

urrent smoking in comparison to the control group (Table

1am

agn(p(sg0agnmp�c

Phh

atmaciptpp(qSto

Fbgp ronary

176 Journal of Cardiovascular Computed Tomography, Vol 2, No 3, May/June 2008

). No differences were observed between patients in the AFnd control groups in terms of study indication and previousedical history.CCTA findings are shown in Table 2. Subjective quality

nalysis showed that 7 (2%) of 360 segments in the AFroup were uninterpretable (poor), resulting in a nondiag-ostic study in 2 (8%) of 24 patients. In comparison, 181%) of 1785 segments in the control group were uninter-retable, leading to 12 (10%) of 119 nondiagnostic studiesP � NS). Patients with AF had more poor- and fair-qualitytudies (46% compared with 21%; P � 0.01) and fewerood-to-excellent studies (54% compared with 79%; P �.01) in comparison to the control group. Objective qualitynalysis showed significantly more noise in the controlroup, a trend toward lower SNR in the control group, ando difference in the CNR. Effective radiation dose wasuch higher in patients with AF in comparison to control

atients (26.2 � 7.8 mSv compared with 8.1 � 3.7 mSv; P0.01). Patients in both groups had similar coronary cal-

igure 1 Distribution of optimal phase for analyzing the coronaetween the lowest and the highest best phases for the group of seray. Diastolic phases (defined as the phases 45%–0%) are showroximal and mid; D, distal; LXC, left circumflex; RCA, right co

ium scores and prevalence of any CAD or significant CAD. f

atients with AF had significantly lower initial heart rates,igher mean and maximal heart rates, and higher maximaleart rate variability.

Retrospective analysis to determine the prevalence ofrtifacts and optimal phase (available in 22 of the 24 pa-ients) in patients with AF showed that artifacts were com-on, being found in 19 (86%) of the 22 studies. Motion

rtifact was found in all 19 studies that showed any artifact;alcium-related artifacts were found in 9 (47%) of 19 stud-es, and other artifacts in 5 (26%) of 19 studies. In the 2atients with AF with non–diagnostic CCTA studies, all 3ypes of artifacts were present. Figure 1 shows the optimalhase for each coronary segment group on a patient-by-atient basis in the AF patients. Diastolic phases onlymarked in gray) were found to provide optimal imageuality on all segments in only 6 (27%) of 22 patients.ystolic phases only (in white) were found to provide op-

imal image quality in 4 (18%) of 22 patients. In 12 (55%)f the 22 patients, some coronary segment groups were

ent groups per patient. Phase window corresponds to the intervals. Systolic phases (defined as the phases 5%–40%) are shown inhite. LM indicates left main; LAD, left anterior descending; P/Martery; N, nondiagnostic,

ry segmgment

n in w

ound to be optimal in diastole and others in systole. The

pnAip

rf�t2rtm

D

trsopAth

tng

fmahCGfim

iwTAhdhpttiwcdsmAn

samtsmasetnpsm

DmpdrTvrfsaba

Falpprlm

177Wolak et al CCTA in patients with atrial fibrillation

hase range (range between the lowest and highest phaseseeded for interpretation for each study) for patients withF varied between 1 and 13 phases. In 9 (45%) of 20

nterpretable studies, the phase range was wider than 8hases.

Absolute delay reconstruction (millisecond or manualedirecting) was used in 9 (38%) of 24 patients. The mostrequently used millisecond reconstruction intervals were300 and �250 millisecond. Figure 2 and Figure 3 describe

he reconstruction method that was used in the study. Figurepresents a case in which simple automatic absolute delay

econstruction was performed with a subsequent satisfac-ory result. Figure 3 and Figure 4 show cases illustrating

ethods used in this study.

iscussion

Our study shows that interpretable CCTA with DSCTechnology in patients with AF is feasible. These promisingesults were achieved despite the unselected nature of thetudy group; ie, although there may have been a preselectionf the patients referred for testing, we did not excludeatients for any reason, and they often had features beyondF known to degrade image quality, including obesity,

achycardia, heart rate variability during acquisition, andigh AS.

In the current study, 2% of coronary segments and 8% ofhe studies in the AF group were deemed uninterpretable orondiagnostic in comparison to 1% and 10% in the control

igure 2 Case illustration of one-step reconstruction process for49-year-old female patient with AF who underwent CCTA. The

ower part of the image shows the ECG sampling, and the upperart shows the corresponding reconstructed image (curved multi-lanar reformatted view). The unsatisfactory results of 70% phaseeconstruction are shown (A). Successful results from �300-mil-isecond reconstruction are shown (B). In this case one-step auto-atic absolute delay reconstruction gave an acceptable result.

roup. This 2% rate of uninterpretable segments compares t

avorably with the 6% rate in the work of Oncel et al9 anday be due to the lower heart rates in our study. Objective

nalysis of image quality found that the noise level wasigher, and there was a trend to toward higher SNR andNR in the control group in comparison to the AF group.iven that there was no difference in body mass index, thisnding is probably related to the absence of tube currentodulation in the AF group.Although our results show that CCTA in AF with DSCT

s feasible, acquiring and interpreting AF CCTA obtainedith DSCT remains a complicated and challenging process.he mean initial heart rate was lower in our patients withF in comparison to the control group. However, the peakeart rate, mean heart rate, and maximal heart rate variationuring image acquisition were higher. This low initial meaneart rate is not surprising in a group of patients withersistent or permanent atrial fibrillation in whom rate con-rol is the standard of care. Interestingly, the heart rates ofhe patients in the control group decreased during contrastnfusion and image acquisition, as previously described,hereas the heart rates of the patients with AF in-

reased.17,18 This surge in heart rate in patients with AFuring contrast agent administration and acquisition, despiteimilar rate of �-blocker usage, suggests that �-blockadeay not be sufficient to control heart rate in patients withF undergoing CCTA and that further interventions may beecessary to reduce the chance for potential motion artifact.

The majority of CCTA studies (86%) in patients with AFhowed �1 artifact. All studies with artifact showed motionrtifact and most of the studies showed �2 types of artifact,ost commonly a combination of motion and calcium ar-

ifacts. The 2 nondiagnostic studies in patients with AFhowed all 3 types of artifacts. Because of the dominance ofotion artifact, the most important step in the process of

nalyzing AF studies is cardiac phase selection for recon-truction. Notably, systolic and diastolic phases are repeat-dly needed for optimal CCTA interpretation, and no dis-inct pattern predicts whether systole or diastole will beeeded for analysis. Thus, our study suggests that, whenerforming CCTA in patients with AF, full tube currenthould be used throughout the entire cardiac cycle (by closeodulation) to permit nonlimited reconstruction.Although when significant arrhythmia is present the

SCT scanner automatically deactivates ECG-based doseodulation, we did not attempt dose modulation for the

atients with AF in this study. The mean effective radiationose in our population was 26.2 � 7.8 mSv, higher than theadiation dose described by Oncel et al9 (13.8 � 1.4 mSv).he difference observed is probably due to a larger field ofiew and a longer acquisition time in our study. In thisegard, the field of view was extended in 6 patients (25%)or coronary artery bypass graft or aortic assessment, anduch patients were excluded in the study by Oncel et al.9 Inddition, the mean heart rate of our patients was lower (74eats/min compared with 83 beats/min), reflecting longercquisition times and higher dose-length products. Radia-

ion exposure is primarily a concern in younger patients

wdb

tamwssf

rwcsrs

arb

FumBp(

178 Journal of Cardiovascular Computed Tomography, Vol 2, No 3, May/June 2008

ith long life expectancy. Because our AF group was pre-ominantly composed of elderly men with multiple comor-idities, the high radiation exposure is less of a concern.

Absolute delay reconstruction was necessary in morehan one-third of the cases. This process can be performedutomatically by defining constant millisecond delay or byanually redirecting the data sampling to the end of the Tave. The end of the T wave was chosen because it corre-

ponds to relative stasis of the myocardium during end-ystole or isovolumic relaxation. Furthermore, we also

igure 3 Case illustration of multistep reconstruction process fonderwent CCTA. The lower part of the image shows the ECultiplanar reformatted view). The results of 70% phase, �300- aecause in all 3 cases the result was unsatisfactory, a manual recoerformed (D). The arrow (D) shows a significant residual misregiE).

ound empirically that for patients with AF, when the heart c

ate is high, skipping beats with short coupling intervalhile maintaining sampling at regular interval yielded suc-

essful results. This was done without losing any 3-dimen-ional data. However, it is time consuming and should beeserved for cases in which the previous steps are notuccessful.

Improved temporal resolution of DSCT is a significantdvantage for imaging patients with AF while temporalesolution can be improved with single source CT scannersy multisegment reconstruction; in AF beat-to-beat cardiac

year-old male patient with AF and rapid ventricular response whothe upper part the corresponding reconstructed image (curved50-millisecond reconstructions are shown (A to C), respectively.ion by redirecting the data sampling to the end of the T wave was

in the RCA that was successfully handled by skipping ECG data

r a 59-G andnd �2

nstructstration

ycle variation also poses a significant technical challenge

wrtoh

ptctisptaArcom

C

d

eTisocr

R

Fbdtsp studie

179Wolak et al CCTA in patients with atrial fibrillation

ith this approach. One should expect beat-to-beat heartate variation to seriously deteriorate image quality in mul-isegment reconstruction. Although 320-slice CT could the-retically acquire all the data from a single cardiac beat, thisas not yet been reported to the best of our knowledge.

Limitations of our study include the small number ofatients and the lack of angiographic correlation. In addi-ion, left ventricular analysis reconstruction with fixed per-entage phases was used to asses the optimal phase. Al-hough this kind of analysis provides a wide range of 5%nterval phases, it lacks the flexibility of manual or milli-econd reconstruction that may prove to be a superior ap-roach to reconstruction of CCTA in these patients. Al-hough we did not record and compare the time necessary tocquire, reconstruct, and read CCTA data from patients withF in comparison to patients with sinus rhythm, our expe-

ience is that interpretation of CCTA data in AF is timeonsuming because of multiple reconstructions, higher ratef poor image quality, and evaluation of segments in nu-erous phases.

onclusions

These preliminary data show that interpretable CCTA

igure 4 An 88-year-old patient with a history of aortic valve rreath and mild troponin elevation. The top row shows the proxescending (LM-LAD) artery, left circumflex (LCX) artery, and rihe same vessels. The diagnostic phase for each segment is given iegment groups that were used to define the optimal image phase.roximal RCA. The patient did not require any further additional

ata can be obtained in patients with AF with the DSCT,

ven in those with relatively high heart rate and high AF.he need for absolute delay reconstruction is common (used

n one-third of the cases) and appears to improve coronaryegmental assessment. Further assessment of the accuracyf CCTA through correlations between DSCT and invasiveoronary angiography in patients with AF appears war-anted.

eferences

1. Feinberg WM, Cornell ES, Nightingale SD, Pearce LA, Tracy RP,Hart RG, Bovill EG: Relationship between prothrombin activationfragment F1.2 and international normalized ratio in patients with atrialfibrillation. Stroke 1997;28(6):1101–6.

2. Fuster V, Ryden LE, Cannom DS, Crijns HJ, Curtis AB, EllebogenKA, Halperin JL, Le Heuzey JY, Kay GN, Lowe JE, Olsson SB,Prystowsky EN, Tamargo JL, Wann S, Smith SC Jr, Jacobs AK,Adams CD, Anderson JL, Antman EM, Halperin JL, Hunt SA, Nish-imura R, Ornato JP, Page RL, Riegel B, Priori SG, Blanc JJ, Budaj A,Camm AJ, Dean V, Deckers JW, Despres C, Dickstein K, Lekakis J,McGregor K, Metra M, Morais J, Osterspey A, Tamargo JL, Zamo-rano JL; American College of Cardiology/American Heart AssociationTask Force on Practice Guidelines; European Society of CardiologyCommittee for Practice Guidelines; European Heart Rhythm Associ-ation; Heart Rhythm Society: ACC/AHA/ESC 2006 Guidelines for theManagement of Patients with Atrial Fibrillation: a report of the Amer-

ment 2 years before the CCTA study presented with shortness ofid portions of the coronary tree for the left main left anterior

onary artery (RCA). The bottom row gives the distal portions forntheses. The study quality was excellent. The figure shows all thehe CCTA, �25% luminal stenosis was found in the mid LAD ands and was not hospitalized within 9 months of follow-up.

eplaceimal-m

ght corn pareFrom t

ican College of Cardiology/American Heart Association Task Force on

1

1

1

1

1

1

1

1

1

180 Journal of Cardiovascular Computed Tomography, Vol 2, No 3, May/June 2008

Practice Guidelines and the European Society of Cardiology Commit-tee for Practice Guidelines (Writing Committee to Revise the 2001Guidelines for the Management of Patients With Atrial Fibrillation):developed in collaboration with the European Heart Rhythm Associ-ation and the Heart Rhythm Society. Circulation. 2006;114(7):e257–354.

3. Abidov A, Hachamovitch R, Rozanski A, Hayes SW, Santos MM,Sciammarella MG, Cohen I, Gerlach J, Friedman JD, Germano G,Berman DS: Prognostic implications of atrial fibrillation in patientsundergoing myocardial perfusion single-photon emission computedtomography. J Am Coll Cardiol 2004;44(5):1062–70.

4. Raff GL, Gallagher MJ, O’Neill WW, Goldstein JA: Diagnostic ac-curacy of noninvasive coronary angiography using 64-slice spiralcomputed tomography. J Am Coll Cardiol 2005;46(3):552–7.

5. Leber AW, Knez A, von Ziegler F, Becker A, Nikolaou K, Paul S,Wintersperger B, Reiser M, Becker CR, Steinbeck G, Boekstegers P:Quantification of obstructive and nonobstructive coronary lesions by64-slice computed tomography: a comparative study with quantitativecoronary angiography and intravascular ultrasound. J Am Coll Cardiol2005;46(1):147–54.

6. Mollet NR, Cademartiri F, van Mieghem CA, Runza G, McFadden EP,Baks T, Serruys PW, Krestin GP, de Feyter PJ: High-resolution spiralcomputed tomography coronary angiography in patients referred for di-agnostic conventional coronary angiography. Circulation 2005;112(15):2318–23.

7. Ropers D, Rixe J, Anders K, Kuttner A, Baum U, Bautz W, DanielWG, Achenbach S: Usefulness of multidetector row spiral computedtomography with 64- � 0.6-mm collimation and 330-ms rotation forthe noninvasive detection of significant coronary artery stenoses. Am JCardiol 2006;97(3):343–8.

8. Flohr TG, McCollough CH, Bruder H, Petersilka M, Gruber K, Suss C,Grasruck M, Stierstorfer K, Krauss B, Raupach R, Primak AN, KuttnerA, Achenbach S, Becker C, Kopp A, Ohnesorge BM: First perfor-mance evaluation of a dual-source CT (DSCT) system. Eur Radiol2006;16(2):256–68.

9. Oncel D, Oncel G, Tastan A: Effectiveness of dual-source CT coro-nary angiography for the evaluation of coronary artery disease inpatients with atrial fibrillation: initial experience. Radiology 2007;

245(3):703–11.

0. Morin RL, Gerber TC, McCollough CH: Radiation dose in computedtomography of the heart. Circulation 2003;107(6):917–22.

1. Agatston AS, Janowitz WR, Hildner FJ, Zusmer NR, Viamonte M Jr,Detrano R: Quantification of coronary artery calcium using ultrafastcomputed tomography. J Am Coll Cardiol 1990;15(4):827–32.

2. Austen WG, Edwards JE, Frye RL, Gensini GG, Gott VL, Griffith LS,McGoon DC, Murphy ML, Roe BB: A reporting system on patientsevaluated for coronary artery disease. Report of the Ad Hoc Committeefor Grading of Coronary Artery Disease, Council on Cardiovascular Surgery,American Heart Association. Circulation 1975;51(4 Suppl):5–40.

3. Leschka S, Wildermuth S, Boehm T, Desbiolles L, Husmann L, PlassA, Koepfli P, Schepis T, Marincek B, Kaufmann PA, Alkadhi H:Noninvasive coronary angiography with 64-section CT: effect of av-erage heart rate and heart rate variability on image quality. Radiology2006;241(2):378–85.

4. Matt D, Scheffel H, Leschka S, Flohr TG, Marincek B, Kaufmann PA,Alkadhi H: Dual-source CT coronary angiography: image quality, mean heartrate, and heart rate variability. AJR Am J Roentgenol 2007;189(3):567–73.

5. Achenbach S, Giesler T, Ropers D, Ulzheimer S, Anders K, Wenkel E,Pohle K, Kachelriess M, Derlien H, Kalender WA, Daniel WG, BautzW, Baum U: Comparison of image quality in contrast-enhanced cor-onary-artery visualization by electron beam tomography and retrospec-tively electrocardiogram-gated multislice spiral computed tomogra-phy. Invest Radiol 2003;38(2):119–28.

6. Lembcke A, Wiese TH, Schnorr J, Wagner S, Mews J, Kroencke TJ,Enzweiler CN, Hamm B, Taupitz M: Image quality of noninvasivecoronary angiography using multislice spiral computed tomographyand electron-beam computed tomography: intraindividual comparisonin an animal model. Invest Radiol 2004;39(6):357–64.

7. Gutstein A, Wolak A, Lee C, Dey D, Ohba M, Suzuki Y, Cheng V,Gransar H, Suzuki S, Friedman J, Thomson LE, Hayes S, Pimentel R,Paz W, Slomka P, Berman DS: Predicting success of prospective andretrospective gating with dual-source coronary computed tomographyangiography: development of selection criteria and initial experience.J Cardiovasc Comput Tomogr 2008;2:81–90.

8. Nieman K, Rensing BJ, van Geuns RJ, Vos J, Pattynama PM, KrestinGP, Serruys PW, de Feyter PJ: Non-invasive coronary angiographywith multislice spiral computed tomography: impact of heart rate.

Heart 2002;88(5):470–4.