In Vivo Intravascular MR Imaging: TransvenousTechnique for Arterial Wall ImagingLawrence V. Hofmann, MD, Robert P. Liddell, MD, Aravind Arepally, MD, Brian Montague, MD,

Xiaoming Yang, MD, PhD, and David A. Bluemke, MD, PhD

PURPOSE: To determine, in vivo, the potential for transvenous magnetic resonance (MR) imaging of the arterial walland to assess appropriate MR pulse sequences for this method.

MATERIALS AND METHODS: MR imaging was performed on 19 vessels (right renal artery, N � 9; left renal arteryN � 2; external iliac artery, N � 4; abdominal aorta, N � 4) in nine swine. The animals were either low-densitylipoprotein receptor knockout (N � 5) or Yucatan mini-pigs fed an atherogenic diet for 6 to 11 weeks (N � 4). Theintravascular MR coil/guide wire (IVMRG) (Surgi-Vision, Gaithersburg, MD) was introduced via the external iliacvein into the inferior vena cava (IVC). The following electrocardiograph-gated MR pulse sequences were obtained:T1-weighted precontrast with and without fat saturation and T1-weighted postcontrast with fat saturation. Twoobservers scored wall signal and conspicuity and classified the vessel as normal, abnormal, or stented. Images werecompared with histopathologic findings.

RESULTS: The T1-weighted precontrast without fat saturation, T1-weighted precontrast with fat saturation, andT1-weighted postcontrast images correlated with histopathologic findings in 12 of 15 vessels, eight of 10 vessels, and14 of 16 vessels, respectively. Abnormal histopathologic findings included: arterial wall thickening (N � 3), arterialdissection (N � 2), focal fibrous plaque (N � 2), adherent thrombus (N � 1). The T1-weighted postcontrast images werenot compromised by artifacts and had the highest score for vessel wall signal and conspicuity. T1-weighted precon-trast images were compromised by chemical shift artifact and poor blood suppression. Negligible artifacts werecreated by the platinum stent.

CONCLUSION: The T1-weighted fat saturated postcontrast pulse sequence was superior to other sequences fortransvenous MR imaging of the arterial wall.

J Vasc Interv Radiol 2003; 14:1317–1327

Abbreviations: IVMRG � intravascular magnetic resonance guide wire, IVC � inferior vena cava

IN 1995, the American Heart Associa-tion Committee on Vascular Lesions ofthe Council on Arteriosclerosis createda histological classification for athero-sclerotic lesions (1). This classification

is based on the histological composi-tion and ultrastructure of atheroscle-rotic lesions, which is thought to beresponsible for certain clinical syn-dromes in the coronary and peripheralvascular circulation. Most atheroscle-rotic thromboembolic events are be-lieved to be secondary to fibrous caprupture with resultant exposure of thethrombogenic subendothelial matrix(2).

Intense research has focused onprospectively identifying these “atrisk” plaques. To that end, investiga-tors have explored different imagingmodalities that would allow them tovisualize the ultrastructure of a lesion,specifically the fibrous cap and itsthickness. Magnetic resonance (MR)

imaging, because of its superior softtissue resolution, has become a prom-ising modality for this task. Most re-search has focused on the carotid ar-teries because of their superficiallocation and the ability to obtain non-invasive, high-resolution images withuse of surface coils; and researchershave been able to discriminate the fi-brous cap from the lipid core andquantify the thickness of the fibrouscap (3–6).

MR imaging of the arterial wall andatherosclerotic plaque that involvesvessels within the abdomen and pelvisis difficult. The relatively large dis-tance between the artery of interestand the surface coil, as well as in-creased motion artifacts, make imag-

From the Russell H. Morgan Department of Radiol-ogy and Radiological Science, The Johns HopkinsMedical Institutions, Blalock 545, 600 North WolfeStreet, Baltimore, Maryland 21287. Received May 12,2003; revision received July 10; accepted July 13.Presented at the 2002 SIR Annual Meeting. Addresscorrespondence to L.V.H.; E-mail: lhofmann@jhmi.edu

This study was supported by a research grant fromSurgi-Vision (Gaithersburg, MD). None of the au-thors have identified a potential conflict of interest.

© SIR, 2003

DOI: 10.1097/01.RVI.0000092904.31640.BE

1317

ing within these regions exceedinglydifficult. In an attempt to overcomethese limitations, investigators haveexplored the use of intra-arterial MRreceiver coils (7–10). Although thesestudies provided high-resolution im-ages, the large device size (5–8 F) andthe need to place the device within theartery carries the risks of dissection,embolism, and possible occlusion,which are substantially reduced withvenous access.

In this study, a small (0.030-in di-ameter) intravascular MR coil/guidewire (IVMRG; Surgi-Vision, Gaithers-burg, MD) was placed in a vein adja-cent to the target artery. This tech-nique has been termed “transvenousMR imaging.” The authors hypothe-sized that in certain areas of the body,the proximity of the vein to the arterywould be sufficient to obtain high-res-olution images of the arterial wall.This technique should substantiallyreduce the risk of complications asso-

ciated with intra-arterial placement ofthe IVMRG. To that end, the authorssought to determine, in vivo, the fea-sibility of this approach and the opti-mal parameters for transvenous MRimaging of the arterial wall.

MATERIALS AND METHODS

IVMRG

The IVMRG (Surgi-Vision, Gaith-ersburg, MD) used in this study is a“receive only” coil and has been pre-viously described (9). Briefly, the de-vice is a 75-cm-long, 0.030-inch-diam-eter, loopless antenna consisting of asoft conducting wire that has an innerconductor from a 50-ohm, 0.6-mm, co-axial cable with a polyester jacket. Theproximal end of the coaxial cable wasconnected through a matching tuning-decoupling circuit to the MR scanner.The IVMRG can function as a conven-tional angiography guide wire with a

performance profile similar to that of a0.035-inch diameter nitinol guide wire.

IVMRG Placement

The Animal Care and Use Commit-tee of the institution approved all an-imal experiments. A total of 19 vesselsin nine swine were imaged. Five ge-netically engineered swine (low-densi-ty lipoprotein receptor knockout[LDL�], Atlanta Cardiovascular Re-search Institute, Norcross, GA), werefed standard hog chow for 9 months(Hog Grower Chow OTC 50, Purina-Mills, St. Louis, MO) before imaging.The remaining four swine were Yuca-tan mini-pigs (Charles River Laborato-ries, Wilmington, MA) that underwentballoon injury (11,12) of the right renalartery and the external iliac artery. Themini-pigs were fed an atherogenic dietcontaining 20% lard, 6% cholesterol,and 2% sodium cholate (ModifiedMini-Pig Grower Diet 5081, Purina-

Imaging and Histopathologic Results

AnimalNo. Artery

T1 T1 Fat Saturation

S* C† Image Findings S* C† Image Findings

1 RRA �1 1 Normal NILRA �1 1 Normal NIIliac NI NI

2 RRA �1 1 Normal; stent visualized with negligible artifact NI

Iliac NI NI3 RRA �1 1 Normal; chemical shift artifact 1 1 Normal

LRA �1 1 Normal; stent visualized with negligible artifact NI4 RRA 1 3 Thickened wall with dissection visualized 1 3 Thickened wall with dissection visualized

5 RRA 1 3 Thickened wall, dissection, perivascularhematoma

1 3 Thickened wall, dissection, perivascularhematoma

6 RRA �1 1 Normal; chemical shift artifact 0 2 NormalAorta 0 2 Normal; poor blood suppression �1 1 Normal; poor blood suppressionIliac NI NI

7 RRA �1 1 Normal 1 1 NormalAorta �1 2 Normal 0 2 Normal; poor blood suppression

8 RRA �1 1 Normal; chemical shift artifact 1 3 NormalAorta �1 1 Normal 1 3 Normal

9 RRA �1 1 Normal 0 1 NormalAorta �1 2 Normal; poor blood suppression NI

Iliac NI NI

* S � vessel wall signal compared to adjacent muscle [�1 (hypointense), 0 (isointense), 1 (hyperintense)]; † C � vessel wallconspicuity [1 (poor), 2 (fair), 3 (good), 4 (excellent)]; �NI � vessel was not imaged with that sequence; RRA � right renalartery; LRA � left renal artery; T1 � T1-weighted non–fat saturated images; T1 fat sat � T1-weighted fat saturated images; T1postcontrast fat sat � T1-weighted postcontrast fat saturated images.

1318 • In Vivo Intravascular MR Imaging October 2003 JVIR

Mills, Richmond, IN) for 6 to 11 weeks(average 8.8 � 2.4 weeks) beforeimaging.

In the fluoroscopy suite, the ani-mals were sedated with an intramus-cular injection of ketamine (22 mg/kg), acepromazine (1.1 mg/kg), andatropine (0.05 mg/kg). Intravenouspentobarbital (20 mg/kg) was also ad-ministered. The animal was intubatedand ventilated with 1.5% isoflurane.With ultrasound guidance, an 8-sheath was placed in the external iliacartery and a 6-F sheath was placed inthe external iliac vein. Abdominal andpelvic x-ray angiography was per-formed. In two renal arteries, the back-end of a guide wire was placedthrough a reversed curve catheter(Sos3; Angiodynamics, Queensbor-ough, NY) to create a renal artery dis-section. In four vessels, a platinumstent (Ominflex; Angiodynamics,Queensborough, NY) was placed (nor-mal renal artery [N � 2], dissected

renal artery [N � 1] and external iliacartery [N � 1]).

The IVMRG was introduced withfluoroscopic guidance, with the aid ofa 5-F catheter, into the inferior venacava (IVC) or right renal vein. The an-imal was then transported to the MRsuite.

MR Imaging

MR images were obtained on a1.5-T MR system (CV/I; General Elec-tric Medical Systems, Waukesha, WI).With use of the body coil, an initial fastmultiplanar spoiled gradient echo se-quence with a 40-cm field of viewserved as a scout image. Then, axialand coronal gradient echo sequenceswere obtained with three external sur-face coils (two posterior, one anterior)to localize the IVMRG and to deter-mine the course of the artery ofinterest.

Double-oblique, electrocardiograph-

gated, double inversion recovery, fastspin echo, black blood images, with-out breath hold or respiratory gating,were obtained perpendicular to thevessel lumen with use of the IVMRG.For high resolution imaging (ie, smallfield of view images), the surface coilswere turned to avoid “wrap around”artifact. T1-weighted images with andwithout fat suppression were initiallyobtained (TR � 1-R-R interval [aver-age TR, 650 msec]; TE, 12 msec; TI, 350msec; ETL, 8–16; field of view, 8 cm; 6signal averages; 256 � 256 resolutioninterpolated to 512 matrix; 3-mm slicethickness; 32.5-kHz bandwidth; aver-age acquisition time, 62 seconds perimage). Because the overall signalfrom the IVMRG was small, the fieldof view, matrix, bandwidth, and slicethickness were varied in the first ani-mal to derive adequate signal whilekeeping the imaging time at about 1minute per image. A T2-weightedpulse sequence was then performed

T1 Postcontrast Fat Saturation

PathologyDiameter

(mm)S* C† Image Findings

NI Normal 5NI Normal 5

1 4 Normal; stent visualized with negligibleartifact

Focal 40–120 �m fibrous plaque; stent 9

1 4 Normal; stent visualized with negligibleartifact

Normal; stent 5

1 4 Normal Focal 120–250 �m fibrous 51 4 Normal Normal 5

NI Normal; stent 51 4 Thickened wall with dissection visualized

with negligible artifactDissection with intramural and perivascular

hematoma; stent10

1 4 Thickened wall, dissection, perivascularhematoma

Dissection with intramural and perivascularhematoma

8

1 4 Minimal circumferential wall thickening Circumferential 100 �m intimal thickening 51 4 Minimal asymmetric wall thickening Asymmetric intimal thickening (500 �m) 7.51 4 Normal Normal 51 4 Normal No path 41 4 Normal No path 91 4 Normal No path 41 3 Normal No path 91 3 Normal Normal 61 3 Thickened wall plus 3-mm mass lesion

adherent to the wall3-mm area of adherent subacute thrombus

with focal intimal thickening9

1 3 40% vessel narrowing by markedasymmetric wall thickening

Marked asymmetric media thickeningbecause of hemorrhage, hylanization,neovascularity

5

Hofmann et al • 1319Volume 14 Number 10

(TR � 2R-R interval [average TR, 1300msec]; effective TE, 60 msec; ETL, 24;field of view, 8 cm; 6 NEX, 256 � 256resolution, 3-mm slice thickness, in-version recovery blood suppression,

32.5-kHz bandwidth, 83 seconds perimage). T2-weighted images were ob-tained in the first three animals andthen the sequence was discontinuedbecause of long imaging times and

non-visualization of the vessel wallsecondary to poor signal-to-noiseratio.

Gadodiamide (Omniscan; Amer-sham, Princeton, NJ) was then intrave-

1320 • In Vivo Intravascular MR Imaging October 2003 JVIR

nously administered (0.2 �mol/kg).After a 5-minute delay, double inver-sion recovery, fast spin echo, T1-weighted, fat-saturated imaging wasrepeated. Inversion times (median TI,150 msec; range, 125–200 msec) wereadjusted visually to suppress signalfrom the vessel lumen. In four ani-mals, after imaging the abdominal

vessels, the IVMRG was repositionedin the external iliac vein to obtain post-contrast images of the external iliacartery.

Vessel Analysis

After MR imaging, the animalswere killed with use of pentobarbital

(100 mg/kg) and a supersaturated so-lution of potassium chloride. The ves-sels were exposed by blunt and sharpdissection. The arterial tree, includingbilateral renal arteries, aorta, and iliacarteries to the level of sheath insertion,were harvested en bloc. The vesselswere placed in formalin solution.

The vessels containing stents wereembedded in methyl methacrylate,cut, ground, and polished with use ofthe EXAKT System (Werheim, Germa-ny), and stained with hematoxylin andeosin. Vessels without stents were em-bedded in paraffin and stained withhematoxylin and eosin, Massontrichrome, and Verhoeff elastic stains.Histopathologic specimens were la-beled as distances from known ana-tomic landmarks, such as the renal ar-tery ostium or aortic bifurcation topermit registration with the MR im-ages. Histopathologic features wereanalyzed and classified by a vascularbiologist.

Image Analysis

MR images were transferred to apersonal computer (Dell, Austin, TX)with eFilm Workstation 1.5.3 (eFilmMedical, Toronto, Canada) software.Standard settings were set to “linear”and “auto.” Two observers (L.V.H.,D.A.B.), blinded to the histopathologicresults, reviewed by consensus 41 sep-arate images of 19 vessels. The observ-ers compared the signal of the vesselwall to that of the adjacent muscle ashypointense [-1], isointense [0], or hy-perintense [1]. In addition, the overallconspicuity of the vessel wall and im-age quality was scored as poor [1], fair[2], good [3], or excellent [4]. For eachimage, the observers classified the ar-terial wall as either normal or abnor-mal. If the vessel wall was classified asabnormal, the observers described theextent and nature of the abnormality.In addition, the presence of artifactsthat could compromise interpretationwas recorded. The presence or absenceof a stent was recorded. The distancefrom the center of the vein to the cen-ter of the artery was measured on theimages and recorded.

The image data were compared tohistopathologic specimens for 15 ves-sels. The images were matched to thehistopathologic specimens by compar-ison of distances from the known an-atomic landmarks on the MR images

Figure 1. Right renal artery (RRA) dissection. The IVMRG is in the IVC and is the onlyreceiver coil used in the transvenous images. (a) Transvenous MR T1-weighted doubleinversion recovery fast spin echo non-fat saturated image shows an axial image of theright renal artery, posterior to the IVC, with significant thickening of the cephalad portionof the arterial wall (short arrow) and a larger signal void that corresponds with the lumen(long arrow) of the vessel. The IVMRG is not seen because it is in a different plane;however, the high signal from the IVMRG can be seen in the adjacent IVC. (b) Trans-venous MR T1-weighted fat saturated double inversion recovery fast spin echo sequence,at the same location as figure part a, demonstrates an increase in wall signal comparedwith figure part a. The high wall signal because of periadventitial hemorrhage is appre-ciated now that the perivascular fat signal is suppressed. A small lumen (arrow) is nowseen in the cephalad portion of the vessel. (c) Transvenous MR T1-weighted fat saturateddouble inversion recovery fast spin echo postcontrast image, at the same location as figurepart a, shows improved wall signal and conspicuity, particularly the thin wall (arrow)between the two lumens. (d) Summation image of all three surface coil images with useof the same imaging parameters and acquired at the same time as figure part c. Thearterial wall is faintly seen (arrow). (e) Masson trichrome stain of the right renal arterydemonstrates the true (TL) and false lumen (FL) of an arterial dissection with significantwall thickening and a thin wall between the true and false lumen. Postmortem thrombosisof the false lumen compresses the true lumen. Bar � 735 �m.

Hofmann et al • 1321Volume 14 Number 10

and gross specimens. Pathologic find-ings were not available in four ves-sels because of difficulty with im-age/gross specimen registration. Inthese vessels, all three pulse se-quences were concordant and usedas a surrogate for histopathology.

RESULTS

Forty-one images (T1-weightedprecontrast without fat saturation, N� 15); T1-weighted precontrast withfat saturation, N � 10; T1-weightedpostcontrast with fat saturation, N �16) of 19 arteries were reviewed. TheTable lists the imaging and his-topathologic results. Vessels ranged insize from 4 to 10 mm (average size, 6.6� 2.01 mm). The average distancefrom the center of the artery to thecenter of the vein was 11.6 � 6.0 mm.

T1-weighted Precontrast without FatSaturation

The T1-weighted precontrast im-ages without fat saturation (normalvessel, N � 8; stented vessel, N � 2;abnormal vessel, N � 5) exhibited thepoorest wall signal and conspicuity.The observers correctly identified allimages of normal and stented vessels.The stent struts were faintly seen assmall focal areas of signal void.

On the T1-weighted precontrast im-ages without fat saturation, the ob-servers were able to correctly identify2 of 5 abnormal arteries. These twovessels had the highest wall signal andconspicuity scores of all vessels im-aged with this sequence, due primar-ily to periadventitial hemorrhage thatproduced a thick wall with high sig-nal; two lumens were identified, cor-responding to the true and false lumen(Fig 1). However, in one of the dis-sected vessels, a stent in the false lu-men could not be visualized with thispulse sequence (Fig 2).

The three arteries that were incor-rectly classified as normal on T1-weighted precontrast images withoutfat saturation were compromised bypoor blood suppression (N � 2) orchemical shift artifact (N � 1). Poorblood suppression prevented the ob-servers from identifying focal areas ofwall thickening in one artery and ad-herent subacute thrombus in the otherartery. Chemical shift artifact compro-mised interpretation of a third abnor-

mal artery with 100 �m circumferen-tial intimal thickening.

The chemical shift between fat andwater results in misregistration of fatand water signal in the frequency en-coding direction. For the parametersused in this study, the chemical shiftbetween fat and water was 1.1 pixels.The chemical shift obscured the arte-rial wall in the frequency encoding di-rection in three of 15 vessels on thenon-fat suppressed pulsed sequence.The appearance was that of wall thick-ening on one side of the vessel wallwith diminished wall thickness on theopposite side. This effect was mostprominent for the smaller vessels (eg,renal arteries) and disappeared on thefat suppression sequences, confirmingour hypothesis that this was chemicalshift artifact.

T1-weighted Precontrast with FatSaturation

Ten arteries were imaged with aT1-weighted precontrast fat saturationpulse sequence. This sequence demon-strated an improved wall signal scorecompared to the T1-weighted precon-trast non-fat saturation pulse se-quences (P � .02; �2 analysis). As wasseen in the T1-weighted precontrastnon–fat saturation images, the dis-sected arterial walls demonstratedhigh wall signal and conspicuity. Thestent in the dissected vessel was notseen on this pulse sequence.

Two abnormal vessels were incor-rectly classified as normal. One vesselwith asymmetric intimal thickening(500-�m thick) was incorrectly catego-rized as normal because of poor bloodsuppression. Circumferential intimalthickening (100-�m thick) was not ap-preciated on the image of the secondvessel.

T1-weighted Postcontrast with FatSaturation

T1-weighted postcontrast imageswere obtained of 16 vessels. This wasthe only sequence in which all vesselsreceived a vessel wall signal score of 1(vessel wall signal greater than adja-cent muscle signal). In addition, allwall conspicuity scores were good orexcellent. All postcontrast imageswere superior in conspicuity to non–contrast-enhanced pulse sequences.Fourteen of 16 images demonstrated

changes that correlated with his-topathologic findings. The platinumstent was faintly visible in all imagesand was not thought to compromiseimage interpretation. Two vessels hadfocal fibrous plaques less than 250 �min thickness that were not appreciatedon the IVMRG images.

Arterial wall abnormalities andstents were best detected on the post-contrast images. For example, in thevessel with an arterial dissection and astent, the stent struts were perceptibleonly after contrast administration (Fig2). In another vessel, a 3-mm adherentsubacute thrombus was not seen onthe non–contrast-enhanced images,partly because of poor blood suppres-sion, but was readily seen after theadministration of contrast material(Fig 3). A third vessel wall was incor-rectly characterized as normal on theprecontrast images, but properly clas-sified as thickened on the postcontrastimages.

DISCUSSION

Recently, investigators have usedelectron-beam CT, multi-detector CT,transcutaneous ultrasound, intravas-cular ultrasound, and MR imaging toevaluate the arterial wall. Electron-

Figure 2. Right renal artery dissectionwith a platinum stent in false lumen. Theintravascular MR guide wire (bright sig-nal) is in the IVC. (a) Transvenous MRT1-weighted double inversion recoveryfast spin echo non-fat saturated imageshows an axial image of the right renalartery with two lumens (arrows) and sig-nificant wall thickening. (b) TransvenousMR T1-weighted fat saturated double in-version recovery fast spin echo sequence,at the same location as figure part a, dem-onstrates an increase in wall signal com-pared with figure part a because of thehigh signal from the perivadventitial hem-orrhage. Two lumens are again appreci-ated (arrows). (c) Transvenous MR T1-weighted fat saturated double inversionrecovery fast spin echo postcontrast image,at the same location as figure part a, dem-onstrates improved wall signal and conspi-cuity compared with the non–contrast-en-hanced images. Faint signal voids from thestent struts are seen (arrows). (d) Hematox-ylin and eosin stain of the right renal arterydemonstrates both the true lumen (TL) anda stent within the false lumen (FL). Bar �850 �m.™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™3

1322 • In Vivo Intravascular MR Imaging October 2003 JVIR

Hofmann et al • 1323Volume 14 Number 10

Figure 3. Subacute adherent aortic thrombus. (a) Transvenous MR T1-weighted non-fat saturated double inversion recovery fast spinecho image with the intravascular MR guide wire (long arrow) in the inferior vena cava, shows an axial image of the aorta (short arrow),compromised by poor blood suppression. (b) Transvenous MR T1-weighted fat saturated double inversion recovery fast spin echopostcontrast image, at the same location as figure part a, demonstrates a 3-mm high signal mass (arrow) adherent to the vessel wall, notseen in figure part a. (c) Gross specimen of the aorta at the same location as figure part a, shows a 3-mm mass adherent to the aorticwall. (d) Hematoxylin and eosin stain of figure part c shows a subacute thrombus adherent to a focal area of intimal thickening. Bar �1,250 �m.

1324 • In Vivo Intravascular MR Imaging October 2003 JVIR

beam CT and multi-detector CT,which have been used for calciumscoring of the coronary arteries, pri-marily identifies calcified plaques,(AHA type Vb) (1) but has a lowersensitivity in identifying earlier, lessadvanced lesions (13). Transcutaneousultrasound is able to not only quantifythe plaque burden (14,15), but also isable to evaluate the surface character-istics and echogenicity of plaque (16).However, this technique is limited be-cause only superficial vessels such asthe carotid and superficial femoral ar-tery can be examined. Intravascularultrasound is able to image deep ves-sels; however, it is often difficult tointerpret, poorly characterizes soft tis-sue, and is unable to image “behind” acalcified plaque (17). Intravascular MRimaging does not have these limita-tions (7). However, it is invasive andthe small field of view currently avail-able suggests that screening of largearterial segments is not practical. Im-aging times are also quite long (ap-proximately 1 minute per image). Be-cause this technique is in its infancy,further advances in MR technologywill hopefully diminish its currentdisadvantages.

Transvenous MR imaging, com-pared with intra-arterial intravascularMR imaging, significantly reduces therisk of a limb-threatening complica-tion during IVMRG placement andimaging. Martin et al (18) proposedimaging the artery by placing an 8-Flooped MR receiver coil in the adjacentvein in a swine model. However, withtheir large diameter receiver coil, theyfound arterial imaging from the ve-nous system was severely limited be-cause placement of their coil was time-consuming and ghosting artifactsoften compromised their images.

Arterial Wall Visualization

Despite respiratory motion and ar-terial pulsation, arterial wall visualiza-tion with the IVMRG was straightfor-ward with no significant limitations.Based on these features, transvenousMR visualization of the adjacent arte-rial wall was successful in not onlyidentifying normal vessels but also inproviding high-resolution imaging ofpathologic states in most vessels. Ofthe transvenous MR pulse sequencesstudied, this study found the T1-weighted postcontrast pulse sequence

provided the greatest wall signal andconspicuity scores.

Administration of intravenous gad-olinium-based contrast material pro-vided full circumferential enhance-ment of the arterial wall. The etiologyof this enhancement is unknown, butis believed to be associated with thedegree of development of the vasa va-sorum (19,20). The enhancement pro-vides the necessary signal for trans-venous MR imaging with a looplessantenna. In this coil design, signal de-creases at a rate proportional to 1/r,where r is the distance from the re-ceiver coil to the area of interest (21).Therefore, the increased signal fromgadolinium-related wall enhancementin part compensates for the signal lossbecause of the distance from theIVMRG to the arterial wall.

Our results demonstrate that a T1-weighted fat saturated pulse sequence,without contrast enhancement, wasnot sufficient to increase wall conspi-cuity compared to the T1-weightednon–fat saturated pulse sequence. Al-though there was a significant increasein the wall signal score compared withthe adjacent muscle, the overall de-creased signal of the entire image lim-ited wall conspicuity. In addition, theT2-weighted sequence was abandonedafter the first three animals because oflack of sufficient MR signal. Better sig-nal could have been obtained by pro-longing the imaging time, but we feltthat image acquisition times substan-tially longer than 1 minute wouldhave limited clinical application. T1images after gadolinium administra-tion and without fat suppression werenot assessed because of the high signalof both the arterial wall and surround-ing fat and thus low contrast to noiseratio.

Artifacts

Chemical shift misregistration of fatand water signal and poor blood sup-pression compromised interpretationof the non–contrast-enhanced images.On the T1-weighted non–fat satura-tion precontrast images, chemical shiftmisregistration of fat and water signalsimulated focal wall thickening inthree vessels. This artifactual wall“thickening” could be proved by ei-ther swapping the phase and fre-quency encoding directions or by re-duction of thickening on fat

suppressed images. This effect was no-ticed typically in smaller vessels (eg,renal arteries) when the normal wallthickness was similar to the degree ofchemical shift misregistration.

Poor blood suppression compro-mised the interpretation of four pre-contrast images, two without fat satu-ration and two with fat saturation. TheMR pulse sequence used incorporatesa 180° nonselective pulse that invertsthe spins of blood signal outside theimaging plane. After an appropriatedelay, related to the T1 time of bloodsignal, the blood flowing into the im-aging plane is “nulled.” However, thiswas not successful in all cases, likelybecause of slow flow along the vesselwall. Lack of blood suppression cansimulate vessel wall thickening. Addi-tional modification to the MR pulsesequence may reduce this effect.

Stents

Because stents have become indis-pensable in the treatment of vasculardisease, the authors were interested inthe intravascular MR appearance of aplatinum stent and the adjacent vesselwall. In this study, a balloon-expand-able platinum stent (AngiodynamicsInc., Queensbury, NY) was used be-cause platinum is weakly ferromag-netic and a previous study concludedthat stainless steel and most nitinolstents cause significant MR ferromag-netic artifacts (22).

For the cross-sectional images ofthe vessel lumen, black blood fast spinecho images were obtained. This pulsesequence minimizes susceptibility ar-tifacts compared with the gradientecho imaging technique. It is interest-ing to note that on the transvenousMR images it was somewhat difficultto visualize the struts of the platinumstent. Only a small focal area of de-creased signal similar in size to thestent strut was seen. There was no ob-vious degradation of image quality be-yond the expected boundary of thestent. Because only four vessels withstents were studied, further investiga-tion of the intravascular MR appear-ance of this platinum stent is required.

In animals, the IVMRG has beenused for real-time guidance and mon-itoring of angioplasty and stent de-ployment (23,24). However, in previ-ous studies the use of a nitinol stentprevented visualization of the arterial

Hofmann et al • 1325Volume 14 Number 10

wall. The results of this animal studydemonstrates the feasibility of trans-venous MR imaging of a platinumstent, a potential approach to fol-low-up of patients after renal arterystent placement without the risks ofconventional angiography or the ad-ministration of iodinated contrast ma-terial. More interestingly, by combin-ing MR contrast agents with drugsand/or genetic material, the deliveryof these agents into the arterial wallcan be monitored (25). This abilitywould ensure proper and accurate de-livery of therapeutic agents.

Limitations of the study design in-clude the lack of quantitative imageanalysis and lack of assessment of in-terobserver variability. The complexcomposition of human atheroscleroticplaque is difficult to duplicate in ani-mal models, therefore the goal of thisstudy was only to determine feasibil-ity of vessel wall imaging, before com-mencing human transvenous MR im-aging studies. The authors of thisstudy are currently developing soft-ware that would permit quantitativeanalysis of the arterial wall. Humanstudies will involve these importantmetrics of atherosclerotic plaque.

With the current IVMRG designand its inherent limitations, image ac-quisition is slow, requiring 50–60 sec-onds per image. Different techniques,such as coronal and sagittal views,would be necessary to screen long ves-sel segments. Further advances in coiltechnology would allow shorter imag-ing times. Although this study hasdemonstrated the feasibility of thetransvenous MR, the invasive natureof this technique must be justified bythe information gained.

The relative roles of transvenousand intra-arterial MR imaging has notbeen clearly defined. However, the au-thors believe that transvenous MR im-aging will be superior and used forreal-time monitoring of arterial inter-ventions. The IVMRG could be placedin the vein adjacent to the target arteryand used for high-resolution imaging.The main reason for advocating thetransvenous imaging approach duringan intervention is that the IVMRG isproduced in one standard length (100cm) with a 3.5-mm diameter distal hubto connect to the MR scanner. Thisshort length and distal hub makescatheter exchanges impossible, and

thus impractical as a working wire.Additionally, increasing the length ofthe IVMRG, to accommodate catheterexchanges, would degrade imagequality. For example, increasing itslength to 135 cm would decrease thesignal-to-noise ratio by 15% (Viohl I,personal communication, 2002).

Future studies could employ MRguidance instead of fluoroscopic guid-ance for the placement of the IVMRG.MR guidance was not evaluated inthis experiment because the primarygoal was arterial wall imaging.

In conclusion, the T1-weighted fatsaturated postcontrast pulse sequencewas the optimum sequence for trans-venous MR imaging of the arterialwall. Furthermore, arterial wall visu-alization was not compromised by thepresence of a platinum stent. In thefuture it is possible that this guidewire could be used to identify arterialpathology (unstable atheroscleroticplaque) and then direct conventionaltherapies (angioplasty/stent place-ment) and/or molecular therapies(gene/drug delivery).

References1. Stary HC, Chandler AB, Dinsmore RE,

et al. A definition of advanced typesof atherosclerotic lesions and a histo-logical classification of atherosclerosis.A report from the Committee on Vas-cular Lesions of the Council on Arte-riosclerosis, American Heart Associa-tion. Circulation 1995; 92:1355–1374.

2. Falk E. Coronary thrombosis: patho-genesis and clinical manifestations.Am J Cardiol 1991; 68:28B-35B.

3. Wasserman BA, Smith WI, Trout HH,III, Cannon RO, III, Balaban RS, AraiAE. Carotid artery atherosclerosis: invivo morphologic characterizationwith gadolinium-enhanced double-ob-lique MR imaging initial results. Radi-ology 2002; 223:566–573.

4. Hatsukami TS, Ross R, Polissar NL,Yuan C. Visualization of fibrous capthickness and rupture in human ath-erosclerotic carotid plaque in vivo withhigh-resolution magnetic resonanceimaging. Circulation 2000; 102:959–964.

5. Hayes CE, Mathis CM, Yuan C. Sur-face coil phased arrays for high-resolu-tion imaging of the carotid arteries. JMagn Reson Imaging 1996; 6:109–112.

6. von Ingersleben G, Schmiedl UP, Hat-sukami TS, et al. Characterization ofatherosclerotic plaques at the carotidbifurcation: correlation of high-resolu-tion MR imaging with histologic analy-sis–preliminary study. Radiographics1997; 17:1417–1423.

7. Rogers WJ, Prichard JW, Hu YL, et al.Characterization of signal properties inatherosclerotic plaque components byintravascular MRI. ArteriosclerThromb Vasc Biol 2000; 20:1824–1830.

8. Zimmermann GG, Erhart P, SchneiderJ, von Schulthess GK, Schmidt M, De-batin JF. Intravascular MR imaging ofatherosclerotic plaque: ex vivo analysisof human femoral arteries with histo-logic correlation. Radiology 1997; 204:769–774.

9. Ocali O, Atalar E. Intravascular mag-netic resonance imaging using a loop-less catheter antenna. Magn ResonMed 1997; 37:112–118.

10. Martin AJ, Henkelman RM. Intravas-cular MR imaging in a porcine animalmodel. Magn Reson Med 1994; 32:224–229.

11. Holvoet P, Theilmeier G, Shivalkar B,Flameng W, Collen D. LDL hyper-cholesterolemia is associated withaccumulation of oxidized LDL, athero-sclerotic plaque growth, and compen-satory vessel enlargement in coronaryarteries of miniature pigs. ArteriosclerThromb Vasc Biol 1998; 18:415–422.

12. Recchia D, Abendschein DR, Saffitz JE,Wickline SA. The biologic behaviorof balloon hyperinflation-induced arte-rial lesions in hypercholesterolemicpigs depends on the presence of foamcells. Arterioscler Thromb Vasc Biol1995; 15:924–929.

13. Baumgart D, Schmermund A, GoergeG, et al. Comparison of electron beamcomputed tomography with intracoro-nary ultrasound and coronary angiogra-phy for detection of coronary atheroscle-rosis. J Am Coll Cardiol 1997; 30:57–64.

14. Tang R, Mercuri M, Bond MG.B-mode ultrasound imaging for detect-ing and monitoring peripheral athero-sclerosis. Am J Card Imaging 1992;6:333–339.

15. Belcaro G, Laurora G, Cesarone MR,De Sanctis MT, Incandela L, Barsotti A.Progression of subclinical atherosclero-sis in 6 years. Ultrasound evaluation ofthe average, combined femoral and ca-rotid bifurcation intima-media thick-ness. Vasa 1995; 24:227–232.

16. Fitzgerald DE, O’Farrell CM. Prog-nostic value of ultrasound morphologyin carotid atherosclerosis. Int Angiol1993; 12:337–341.

17. Kimura BJ, Bhargava V, DeMaria AN.Value and limitations of intravascularultrasound imaging in characterizingcoronary atherosclerotic plaque. AmHeart J 1995; 130:386–396.

18. Martin AJ, McLoughlin RF, Chu KC,Barberi EA, Rutt BK. An expandableintravenous RF coil for arterial wall im-aging. J Magn Reson Imaging 1998;8:226–234.

19. Aoki S, Shirouzu I, Sasaki Y, et al. En-

1326 • In Vivo Intravascular MR Imaging October 2003 JVIR

hancement of the intracranial arterialwall at MR imaging: relationship to ce-rebral atherosclerosis. Radiology 1995;194:477–481.

20. Sharma S, Taneja K, Gupta AK, RajaniM. Morphologic mural changes inthe aorta revealed by CT in patientswith nonspecific aortoarteritis (Taka-yasu’s arteritis). AJR Am J Roentgenol1996; 167:1321–1325.

21. Atalar E, Bottomley PA, Ocali O, et al.

High resolution intravascular MRI andMRS by using a catheter receiver coil.Magn Reson Med 1996; 36:596–605.

22. Maintz D, Kugel H, Schellhammer F,Landwehr P. In vitro evaluation ofintravascular stent artifacts in three- di-mensional MR angiography. Invest Ra-diol 2001; 36:218–224.

23. Yang X, Bolster BD, Jr, Kraitchman DL,Atalar E. Intravascular MR-moni-tored balloon angioplasty: an in vivo

feasibility study. J Vasc Interv Radiol1998; 9:953–959.

24. Lardo AC. Real-time magnetic reso-nance imaging: diagnostic and inter-ventional applications. Pediatr Cardiol2000; 21:80–98.

25. Yang X, Atalar E, Li D, et al. Magneticresonance imaging permits in vivomonitoring of catheter-based vasculargene delivery. Circulation 2001; 104:1588–1590.

Hofmann et al • 1327Volume 14 Number 10