Peter J. Yim, PhD, John L. Nosher, MD, Anthony Burgos, MD, Ihab Haddadin, MD
Rationale and Objectives. The technique of subtraction computed tomographic angiography (sCTA) has been proposedfor the evaluation of atherosclerotic disease to address limitations in CTA in highly calcified arteries. However, sCTA hasnot gained acceptance in clinical practice, in part, due to image artifacts caused by patient motion that occur between theacquisition of the two component images. The purpose of this study was to evaluate the effectiveness of computationalimage co-registration to obtain sCTA.
Materials and Methods. The study was conducted using a semi-automated implementation of the mutual information(MI) registration algorithm. The results of sCTA were evaluated quantitatively in a phantom representing a calcified ar-tery. Technical success of sCTA was evaluated in 14 calcified arterial segments in two patients. An observer study wascarried out to determine interobserver agreement in the interpretation of sCTA. Qualitative observations were made be-tween sCTA and CTA.
Results. Computation time for performing the co-registration for each 2-cm calcification is less than 1 second. The neces-sary user interaction required minimal expertise. Measurements of the degree of stenosis in the calcified artery phantomagreed to within 8 � 4% of gold-standard measurements. Technical success was demonstrated in all calcifications. Stronginterobserver agreement was obtained for the detection of hemodynamically significant stenoses (� � 0.86). Several ap-parent pitfalls in the interpretation of CTA in calcified arteries were noted that could potentially be obviated by sCTA.
Conclusions. The study supports the use of a straight-forward implementation of the MI algorithm and provides prelimi-nary evidence validating the use of sCTA in the setting of atherosclerotic disease of the lower extremities.
The accuracy of the computed tomographic angiographic(CTA) examination may be degraded by the presence ofcalcified plaque. Detection and quantification of such cal-cification may have value in the diagnosis and therapyplanning of this disease (1–3); however, in relation to theprimary criteria of disease severity (the degree of steno-
Acad Radiol 2009; 16:257–265
1 From the Department of Radiology, UMDNJ-Robert Wood Johnson Medi-cal School, Medical Education Building 404, New Brunswick, NJ 08903.Received March 19, 2008; revised June 18, 2008; accepted July 10, 2008.Address correspondence to: P.J.Y. e-mail: [email protected]
sis), such calcifications are simply a nuisance. Calcifica-tions appear in CTA as small objects with a high com-puted tomographic density that can sometimes approachthat of bone. In the cross-sectional view, the lumenal re-gion of the calcified artery may appear distinctive as aregion with intermediate intensity surrounded either par-tially or completely by a high-intensity region correspond-ing to the calcification. Thus, common radiologic practicefor assessing calcified arteries in CTA is to review thecross-sectional images. However, even with this approach,recent studies suggest that detection of hemodynamicallysignificant stenoses in the presence of calcified plaque canbe inaccurate. For example, in the study of Willmann
YIM ET AL Academic Radiology, Vol 16, No 3, March 2009
et al (4), the presence of calcification tended to cause anoverestimation of the degree of stenosis. In their study of39 consecutive patients with 35 hemodynamically signifi-cant stenoses, overestimation of stenosis occurred in 26vessel segments. In 20 of the cases in which the stenosiswas overestimated, the primary cause of overestimation ofstenosis was the presence of calcification. Ouwendinket al (5) found that wall calcifications in CTA often lim-ited the diagnostic value of CTA and were a statisticallysignificant predictor of when a patient would need addi-tional imaging studies.
In the study presented here, we elaborate on a promis-ing new approach to computed tomographic angiographyinvolving, in essence, digital subtraction of the non-con-trast computed tomography from the computed tomogra-phy angiography. Our study addresses practical consider-ations that arise for implementation of this techniqueincluding, most importantly, the need to correct for pa-tient motion that occurs between the acquisition of thenoncontrast CT (ncCT) and the computed tomographicangiography. A computational image registration algo-rithm plays a large role in this respect. The study alsoinvolves robust validation of the methodology including astudy using a realistic phantom as well as observationsfrom a clinical study.
MATERIALS AND METHODS
PatientsThis study was based on imaging from two patients
who underwent a combined CTA and ncCT study for sus-picion of arterial occlusive disease of the lower extremi-ties. Analysis of the images was approved by our institu-tional review board. Acquisition of the combined CTAand ncCT was performed for clinical considerations inthese patients.
was simulated using a vascular phantom. The phantomwas constructed to include four relevant components ofcalcified arteries, the including blood pool, vessel wall,calcification, and extravascular tissue. Construction of thephantom was performed in-house. The vessel wall wasformed from a polypropylene straw with a wall thicknessof 0.05 mm. The phantom consisted of two stenosis mod-els. The stenoses were concentric in cross-sectional shapeand cross-sectional position. Calcification was formed
258
with plasticine clay that, as applied to the phantom, has amaximum computed tomographic density of 1230Hounsfield Units (HU) at 120 kVp, which is close to thatof highly calcified plaque. The clay was applied in a thinlayer to the outside of the vessel wall at the locations ofthe stenoses, with a maximal thickness of approximately2 mm. The shape of the calcifications was irregular, andthey were applied asymmetrically in relation to the axisof the stenosis. The area surrounding the vessel, corre-sponding to soft tissue was filled with water. The bloodpool was modeled alternately with water and with a sug-ar-based solution that produces a computed tomographicdensity of 300 HU at 120 kVp at the center of the lumen.The water and the sugar-based solution represent noncon-trast and contrast-filled conditions respectively. Flow con-nections were made to a syringe to allow for exchange ofthe two blood pool fluids during the imaging study. Thephantom is shown schematically in Figure 1.
Image Acquisition: Vascular PhantomAll computed tomographic images of the phantom
were acquired with a four-detector row Lightspeed PlusCT (GE Medical Systems, Waukesha, WI). Images wereacquired in one session in which the technique of sCTAwas simulated and in a second session in which goldstandard images were obtained. In the sCTA session, thevascular phantom was imaged in two phases. In the firstphase, the lumenal space was filled with water to simulateblood alone and in then second phase, the lumenal spacewas filled with a sugar solution. The phantom was shiftedand turned slightly between the two phases of the imageacquisition to simulate patient motion. The image acquisi-tion in both phases was identical using the “runoff ” pro-tocol. Parameters of the acquisition included a tube volt-age of 120 kVp, a tube current of 150 mA, a reconstruc-tion slice thickness of 1.25 mm, and a 35-cmreconstruction field-of-view.
In the second session, gold standard images were ac-quired. In this session the clay, simulating calcifiedplaque, was removed from the phantom. Also, the lume-nal space was filled with the sugar solution. Image acqui-sition was performed with the “routine head” protocol.Parameters of the acquisition included a tube voltage of140 kVp, a tube current of 255 mA, a reconstruction slicethickness of 1.25 mm, and a 17.5 cm field-of-view.
Image Acquisition: Clinical CasesImaging was performed on a 64-detector-row Light-
speed VCT. The images were acquired with the standard-
Academic Radiology, Vol 16, No 3, March 2009 SUBTRACTION COMPUTED TOMOGRAPHIC ANGIOGRAPHY
Figure 1. Schematic of the calcified artery phantom. The phantom represents the relevant four elements of the imaging field: lumen (L),calcification (C), vessel wall (VW), and extravascular space (EVS). The phantom was constructed with two stenoses of varying stenosislength (SL) and stenosis diameter (SD).
of-care clinical protocol that included a tube voltage of120 kVp and a tube current of 101 mA and were recon-structed at a slice thickness of 1.25 mm and an in-planeresolution of 0.75 mm. Images were acquired with a heli-cal acquisition with a pitch of 1.375. Motion of the sub-ject was not restricted during the computed tomographicexamination, and the subject was not requested to remainmotionless between acquisition of the ncCT and the CTA.
Image RegistrationSome patient motion is expected to occur between the
two phases of the subtraction CTA acquisition, even ifthe patient is instructed not to move. Thus, in general, acorrection must be applied to one of the images to com-pensate for that motion. A semi-automated co-registrationalgorithm was used in this study. The algorithm includesmanual registration, manual identification of calcificationsand computational registration. Details of the registrationprocess are given in the following.
The first step in the registration process is manual reg-istration of the noncontrast computed tomography and theCTA. Manual registration is expected to be relatively reli-able because interscan motion will usually be small scaleand primarily translational in nature. Thus, correspondingfeatures in the two images can be recognized without ex-tensive searching. Furthermore, the structures that are ofmost interest are the calcifications that have distinctivesmall-scale features that can be matched between the twoimages. Thus, the manual registration process for a givenarterial territory consists of identifying a matching pair ofpoints from the two images on a calcification. One of theimages is then shifted both in- and out-of-plane to pro-duce alignment of those matching points.
In the proposed registration algorithm, the entire im-ages are not registered to one another because the spatialtransformation or motion correction between the two im-ages may become complex for large volumes of tissue,including elastic deformations. Instead, the images are
registered to one another in segments with relatively lim-ited volumes. Within such volumes, assumptions of therigidity of the tissue are valid or nearly so. More specifi-cally, these registration subvolumes are defined to encom-pass calcifications and ideally are positioned such that thecalcifications are centered within the volumes. The accu-racy of the subsequent computational registration willdepend to some extent on whether there are an adequatenumber of points within the calcification within the sub-volume and on whether those points are sufficiently cen-tered within the subvolume. These subvolumes can beobtained in the following manner. First, the user places aseries of points along a calcified segment in the arterywith each point separated from the previous point by ap-proximately 2 cm. This series of points defines the pathof a calcified segment of an artery and also subdivides theartery in the axial direction. Each pair of consecutivepoints in this series can then be used to define a boundingbox that surrounds the vessel and calcifications. First, abox is defined by considering the two points as its dia-metrically opposed corners. Then, that box is extendedoutwards by 1 cm in all directions to form the subvolumebounding box. Thus, the bounding box has a minimumpossible dimension of 2 cm in any direction and a maxi-mum dimension of 4 cm. Computational registration isthen applied sequentially to each of the subvolumeswithin the respective bounding boxes. The process ofconstructing the subvolumes for registration is shownschematically in Figure 2. In our study, all interactiveaspects of the registration process were performed inMIPAV (National Institutes of Health, Bethesda, MD)including identifying points for manual registration, iden-tifying points along the calcified vessel axes, and crop-ping of the images.
Corresponding subvolumes of the noncontrast com-puted tomography and the CTA were then co-registeredusing the mutual-information algorithm (6). The softwarealgorithm was implemented in-house in “C” on the IRIS
259
YIM ET AL Academic Radiology, Vol 16, No 3, March 2009
Explorer prototyping platform (Numerical AlgorithmsGroup Ltd, Oxford, UK) on a Windows 2000 worksta-tion. Preprocessing of the subvolumes included resam-pling to increase the resolution by a factor of two using athird-order b-spline function as implemented in MIPAV(National Institutes of Health). Within each CTA subvol-ume, a set of points was defined that were expected to beeither lumen or calcification. These points were identifiedby thresholding of the CTA at a level of 100 HU and bymanual contouring to exclude possible regions of bone inthe subvolumes. Points within a 0.5-cm margin in theCTA were also excluded from this set to ensure completeoverlap with the noncontrast CT for a wide range oftranslations and rotations. Calculations of mutual informa-tion were then based entirely on these lumen/calcificationpoints. The histographic bin size for computing the mu-tual information cost function was set, such that there was
Figure 2. Image regions included in co-registration of noncon-trast computed tomography and the computed tomographic an-giography. The regions are based on a series of two or morepoints along the calcified artery identified by the user. A box isthen constructed, such as a pair of consecutive points (P1, P2),such that those points are located at the diametrically oppositecorners. The box is then extended outwards by 0.5 and 1.0 cm toform image subvolumes in computed tomographic angiography(SVCTA) and in noncontrast computed tomogaphy (SVNCCT).
260
an average of one image-lumen/calcification point per binin the joint histogram. Registration was based on a rigidmodel of motion with both translational and rotationalcomponents. Intraregistration resampling was performedusing trilinear interpolation. In the registration process,the CTA served as the reference image. whereas the non-contrast computed tomogram served as the floating image.Optimization was performed using the gradient-descentmethod with translational increments of 0.1 mm in eachdirection and rotational increments of 0.2° around eachaxis.
Based on the results of the mutual information regis-tration, the optimal motion correction was applied to theentire noncontrast computed tomographic subvolume andthen was arithmetically subtracted from the CTA subvol-ume. The resulting subtraction subvolume was then rein-serted into its original position in the CTA after croppinga margin of 0.5 cm, where subtraction may not have beeneffective due to nonoverlap of the CTA sub-volume andthe noncontrast computed tomographic subvolume.
Image Analysis: Vascular PhantomsCTA of the phantom was rendered with the maximum
intensity projection (MIP) at a series of views perpendicu-lar to the native slice plane. The views spanned 180° at30° increments. Image resampling for performing the ro-tations was done using trilinear interpolation. The or-thonormal perspective was used for creating the MIPs.The series of MIPs was then viewed by researcher(P.J.Y.) using MIPAV (National Institutes of Health) toselect a point of maximal stenosis in one of the views.Lines for forming image-intensity profiles were drawn atthe point of maximal stenosis and at a normal segmentof the model using the line tool. An edge enhancementfilter, the gradient magnitude, was then applied to theMIP using MIPAV. The convolution kernel of the filterwas set to a space constant of 1.0 � 1.0 pixels. Image-intensity profiles of the gradient-magnitude image at thestenotic and normal locations were then analyzed to deter-mine the respective lumenal diameters. Specifically, thediameter was considered to be the distance between thetwo highest maxima in the image intensity profiles corre-sponding to the two opposing sides of the lumen. A simi-lar process was used for measuring diameters in the goldstandard images although profiles were obtained from thecross-sectional imaging.
The effect of initialization error on the accuracy of theregistration was also evaluated. The initialization position ofthe floating image was varied by displacements of up to 5.0
Academic Radiology, Vol 16, No 3, March 2009 SUBTRACTION COMPUTED TOMOGRAPHIC ANGIOGRAPHY
Figure 3. Subtraction computed tomographic angiography of calcified artery phantom. The im-ages represent the results of simulation of the computed tomographic angiography (a,e), the non-contrast computed tomography (b,f), and the subtraction computed tomographic angiography(c,g). Reference images (d,h) were obtained from imaging of the phantom without calcification ma-terial present.
mm in each direction from the displacement obtained byregistration using a best guess initialization.
Observer Study: Clinical CasessCTA of clinical cases was reviewed by two radiology
residents (A.B., I.H.), for detection of hemodynamically sig-nificant stenoses. Such stenoses were defined as points with
greater than or equal to 50% narrowing that could be identi-fied with at least moderate confidence. Rendering of theclinical sCTA was performed in a similar manner to that ofthe phantom study. Rendering sCTA was performed afterremoval of bone using manual contouring.
Twelve calcified arterial segments were included in theanalysis from the superficial femoral artery (n � 3), pop-
261
YIM ET AL
liteal artery (n � 2), and popliteal trifurcation arteries(n � 7). The total length of calcified segments of arteriesincluded in the study was approximately 28 cm. The im-age co-registration and subtraction was performed forfourteen 2-cm calcification subvolumes.
RESULTS
A high degree of suppression of the calcification com-ponent was achieved in the phantom study as is shown inFigure 3. Residual artifact from the calcification was seenbut was significantly lower in intensity than the lumenand did not obstruct the view of the lumen in the MIP.Measurement of the degree of stenosis from the sCTAwas within 8 � 4% for the two models (Table 1). Theregistration process was found to be insensitive to theinitialization for error in the initialization position of upto 2.0 mm. Significant registration error was found to oc-cur for larger initialization errors as shown in Figure 4.
sCTA was considered to be technically successful inall 14 2-cm calcifications as judged (by P.J.Y.) to producean arterial segment with a realistic appearance. An exam-ple of sCTA is shown in Figure 5. A high degree of in-terobserver agreement was also obtained for the detectionof hemodynamically significant stenoses (� � 0.86). Abreakdown of the outcome combinations is shown inTable 2.
Figure 4. Effect of transformation error in the initialization of theimage registration on the accuracy of the image transformationfollowing registration. Only the translational component of the im-age transformation is represented in the error. The true imagetransformation was estimated as that obtained from a best guessinitialization followed by mutual information registration.
262
Practical aspects of the clinical use of sCTA includethe degree of user interaction and computation time. Theuse of this implementation of sCTA was found to requiremodest user interaction that primarily relates to the identi-fication of corresponding points in the ncCT and the CTAfor performing manual registration. Identification of suchpoints was not found to be problematic in the cases in-cluded in this study. The identification of one pointwithin a given vascular territory was found to be ade-quate. Manual registration had to be repeated in one ofthe three vascular territories when obvious misregistrationwas seen in the sCTA. The computation time for per-forming sCTA was approximately 9 seconds for each2-cm calcification, including 8 seconds for image interpo-lation for each 2-cm calcification and 1 second for mutualinformation (MI) image co-registration.
Possible errors in the interpretation of CTA were notedthat could potentially be resolved by using sCTA. Theseinclude: (1) the underestimation of the degree of stenosisdue to the presence of calcified plaque that is isodense withthe lumen, and (2) the false impression of vascular occlusionassociated with a concentric calcification. These effects areshown in Figure 6.
Table 1Results from the Phantom Study: Measurements fromSubtracted Computed Tomograpic Angiography (sCTA)Compared with Gold Standard (GS) Measurements
Table 2Breakdown of Outcomes from Interobserver Agreement Studyfor the Detection of Stenosis
Negative Positive Total
Negative 6 0 6Positive 1 5 6Total 7 5 12 segments
Negative, number of arterial segments in which stenoses wasnot detected; Positive, number of arterial segments in which ste-nosis was detected.
Academic Radiology, Vol 16, No 3, March 2009
Academic Radiology, Vol 16, No 3, March 2009 SUBTRACTION COMPUTED TOMOGRAPHIC ANGIOGRAPHY
Figure 5. Subtraction computed tomographic angiography in the territory of the superficial femo-ral artery. Subtraction computed tomographic angiography (c) obtained from computed tomo-graphic angiography (a) and noncontrast computed tomography (b).
Figure 6. Types of calcified arterial segments in which computed tomographic angiography islikely to be misinterpreted. One such type is where the calcification is highly inhomogenous asshown in the computed tomographic angiogram (b) and noncontrast computed tomography (c) andwhose position is indicated by the top line in (a). Another type is where the calcification concentri-cally surrounds the lumen as shown in the computed tomographic angiogram (d) and the noncon-trast computed tomography (e) and whose position is indicated by the bottom line in (a). In the firsttype of arterial segment, the degree of stenosis may be underestimated in computed tomographicangiography. In the second type of arterial segment, total occlusion may be suspected based onreading of the computed tomographic angiogram.
DISCUSSION
The present study strongly suggests the potentially ac-curate and robust application of sCTA for the suppressionof calcification. In particular, the use of mutual informa-
tion registration appears to be justified in this context.This study builds on earlier reports of the use of sCTA byproviding more robust validation as well as a more practi-cal framework for clinical implementation. Highlights ofour findings include the validation of the accuracy of
263
YIM ET AL Academic Radiology, Vol 16, No 3, March 2009
sCTA in a vascular phantom. Also, results from a clinicalstudy shows that sCTA produces images that have ahighly realistic appearance that can be interpreted in areproducible manner. Also, possible limitations of con-ventional CTA in visualization of the lumen were notedwhere sCTA may be advantageous.
Other important findings relate to the practicality ofsCTA. Our study shows that this technique can be imple-mented in a manner that is fully compatible with clinicalpractice. In particular, the study proposes a semi-auto-mated technique for performing sCTA that involves man-ual co-registration of the noncontrast computed tomogramand the CTA for each vascular territory and for the userto trace the path of each calcified arterial segment. Thesesteps are not time-consuming and do not require signifi-cant expertise. Finally, the computational requirements ofsCTA are relatively minor and thus the process can becarried out virtually in real time. The registration processhas been found to be essentially insensitive to error in themanual initialization of the registration of up to 2.0 mm.The preliminary studies on clinical cases suggest that thisdegree of accuracy in the manual registration is feasible.
There have been several previous studies addressing theprospects of sCTA. A major focus of all of these studieswas on co-registration of the noncontrast computed tomogra-phy and CTA. Poletti et al (7) found a high rate of true-positive findings with sCTA (95.9%) in comparison withdigital subtraction angiography. However, the techniqueof Poletti et al required that patient motion be minimizedby placing restraints on the legs of the patient. Even so,diagnostic use of the sCTA in this study was not feasiblein 20% of cases, presumably due to excessive patient mo-tion. Registration of those images was performed in aninteractive manner using in-house software. The softwareallowed for correction of translational motion. A singletranslation was applied to the entire image.
The use of a computational algorithm for co-registra-tion of the noncontrast computed tomography and theCTA offers the potential of improving the precision of theresult over what can be obtained by manual methods. Theappropriate approach to performing this registration isvery much up for debate. The core of the registration pro-cess, the cost function, that represents the criteria for de-fining correct alignment, has been defined variously.Kwon et al (8) and Loeckx et al (9) have used the mutualinformation cost function that has been proposed in wide-spread applications both within and outside of medicalimaging for image co-registration. This cost function hasbeen found to be particularly useful for the co-registration
264
of images obtained with different modalities, where theimages have different appearances and may evenhave reversals in the relative contrasts between objects inthe image. An example of this type of application is inthe registration of positron emission tomography andcomputed tomography (10). In the case of sCTA, obvi-ously the two images are acquired with the same modal-ity, but, in the vicinity of the arteries, the images have avery distinctive appearance due to the absence of lumenalcontrast in one and the presence in the other. Mutual in-formation between two images is maximized, in generalterms, when a given intensity in one image correspondswith a high probability to another, potentially different,intensity in the second image.
Other criteria that have been considered for image reg-istration in sCTA include the deterministic sign-change(DSC) criteria (11), a maximum cancellation (MC) costfunction (12) and a modified least-squares (MLS) criteria(13). Of these, MC and MLS have been applied to sCTAfor the purpose of suppression of calcification. It has notbeen shown as to whether these different registration cri-teria produce any substantive differences in the registra-tion accuracy. In this study, the use of MI was chosensimply because its underlying principle and applicabilityare somewhat more general than that of the other tech-niques. Also, it was noted in qualitative preliminary test-ing that the results using MI were at least equivalent tothose obtained by MC.
Study LimitationsAlthough the results of the present study are very
promising, the study is preliminary in nature, involvingonly a phantom study and a limited number of clinicalcases. Also, even in the clinical cases where the techniquewas evaluated, a clear standard-of-comparison, such asintra-arterial digital subtraction angiography was notavailable. Further, more extensive clinical studies of thistechnique will certainly be needed to clarify the its reli-ability and accuracy.
ConclusionsA very promising option for co-registration of the
component images for forming sCTA has been developedand validated. The method has only moderate computa-tional and user-interaction demands that are well withinthe usual constraints of clinical practice. The co-registra-tion process was designed to be as simplistic as possible,including the use of a piecewise-rigid model of patientmotion, gradient-descent optimization, and a generic ver-
Academic Radiology, Vol 16, No 3, March 2009 SUBTRACTION COMPUTED TOMOGRAPHIC ANGIOGRAPHY
sion of the MI cost function. This approach was seen tobe effective. However, some artifacts in the sCTA, partic-ularly in the phantom study, are likely due to residualmisregistration and may be subject to further improve-ment. Also, manual interaction, particularly for initializa-tion of the co-registration, plays an important role in ob-taining sCTA. We believe this is probably acceptable inclinical practice but ideally, the user interaction should befurther reduced or eliminated.
REFERENCES
1. Rutten A, Isgum I, Prokop M. Coronary calcification: effect of smallvariation of scan starting position on Agatston, volume, and massscores. Radiology 2008; 246:90–98..
2. Suzuki M, Ozaki Y, Komura S, Nakanishi A. Intracranial carotid calcifi-cation on CT images as an indicator of atheromatous plaque: analysisof high-resolution CTA images using a 64-multidetector scanner. RadiatMed 2007; 25:378–385.
3. Maldonado TS. What are current preprocedure imaging requirementsfor carotid artery stenting and carotid endarterectomy: have magneticresonance angiography and computed tomographic angiography madea difference? Semin Vasc Surg 2007; 20:205–215.
4. Willmann JK, Baumert B, Schertler T, et al. Aortoiliac and lower ex-tremity arteries assessed with 16-detector row CT angiography: pro-spective comparison with digital subtraction angiography. Radiology2005; 236:1083–1093.
5. Ouwendijk R, Kock MC, van Dijk LC, van Sambeek MR, Stijnen T,Hunink MG. Vessel wall calcifications at multi-detector row CT an-giography in patients with peripheral arterial disease: effect on clini-cal utility and clinical predictors. Radiology 2006; 241:603– 608.
6. Pluim JP, Maintz JB, Viergever MA. Mutual-information-based registra-tion of medical images: a survey. IEEE Trans Med Imaging 2003; 22:986–1004.
7. Poletti PA, Rosset A, Didier D, Bachmann P, Verdun FR, et al. Subtrac-tion CT angiography of the lower limbs: a new technique for the evalu-ation of acute arterial occlusion. ASR Am J Roentgenol 2004; 183:1445–1448.
8. Kwon SM, Kim YS, Kim T, Ra JB. Novel digital subtraction CT angiog-raphy based on 3D registration and refinement. SPIE Med Imaging2004; 5370:1156–1165.
9. Loeckx D, Drisis S, Maes F, Vandermeulen D, Marchal G, Suetens P.Removal of plaque and stent artifacts in subtraction CT angiographyusing nonrigid registration and a volume penalty. Conf Proc IEEE EngMed Biol Soc 2005; 4:4294–4297.
10. Vogel WV, Schinagl DA, Van Dalen JA, Kaanders JH, Oyen WJ. Vali-dated image fusion of dedicated PET and CT for external beam radia-tion therapy in the head and neck area. Q J Nucl Med Mol Imaging.2008; 52:74–83.
11. Bani-Hashemi AR, Krishnan A, Samaddar S. Warped matching for digi-tal subtraction of CT-angiography studies. Proceedings SPIE 1996;2710:428–437.
12. Yim PJ, Nosher JL. Subtraction computed tomographic angiography:use of pre-contrast images for calcification removal. Proceedings SPIE2006; 6143:805–811.
13. van Straten M, Venema HW, Streekstra GJ, Reekers JA, den HeetenGJ, Grimbergen CA. Removal of arterial wall calcifications in CT an-giography by local subtraction. Med Phys 2003; 30:761–770.
