~ ~ ~~ - 7

Technical Note

IMP-RAGE Subtraction Venography: A New Technique1

Jeffrey Stevenson, BS Edmond A. Knopp, MD Andrew W. Litt, MD

Preliminary evaluation of a new magnetic resonance (MR) venography technique WM performed with data sets from five patients un- dergoing MR imaging of the brain before and after intra- venous administration of gadopentetate dimeglu- mine. Before contrast agent injection. the patients were imaged with MP-RAGE (magnetization-prepared rapid gradient-echo) and

quences. After contrast agent injection, the MP- RAGE sequence was re- peated. Images were post-

rithm that calculates, on a pixel-by-pixel basis. the ab- solute value of signal inten- sity of each postcontrast MP-RAGE partition minus that of each precontrast MP-RAGE partition. These subtracted partitions were then subjected to a stan- dard madmum-intensity- projection algorithm to ob- tain the renogram. In all cases, the new method af- forded a high-resolution venogrnm with clear depic- tion of venous sinus anat- omy. Cortical venous anat- omy was also clearly de- picted.

add turbo T2-weighted SC-

PTOCeSSed with *o-

MAGNETIC RESONANCE (MR) venogra- phy has been used to depict relatively large venous structures, namely the dural sinuses and major central deep cerebral veins (1-5). To our knowledge. no MR technique to date allows ad- equate visualization of the smaller ve- nous structures of the brain. The im- portance of visualizing small venous structures, including the cortical veins, lies in the resurgence of stereotaxic neurosurgery. The neurosurgeon must know the location of major cortical veins preoperatively to plan his surgical approach (Kelly PJ, personal communi- cation, 1994). Stereotaxic targeting and trajectory are altered by overlying corti- cal veins (Kelly PJ. personal communi- cation. 1994). At present. this requires intraarterial digital subtraction angiog- raphy. which has its own risks and limitations. We have developed a new MR technique designed specifically for this purpose.

venography with time-of-flight tech- niques is optimized for visualization of slower venous flow: however, that rate of flow is still greater than that in the smaller cortical veins. The technique of MP-RAGE (magnetization-prepared rapid gradient-echo) subtraction venog- raphy does not rely exclusively on time- of-flight effects to depict flow. Also. the TI -reducing effects of gadolinium and background subtraction are used. thereby improving depiction of the slower flow in cortical veins.

Conventional two-dimensional MR

0 MATERIALS A N D METHODS MP-RAGE subtraction venography

was evaluated in five patients undergo- ing MR imaging of the brain before and

Index terms: Arterlovenous malformations. cerebral. 10.75 Cerebral blood vessels. MR. 17.12142. 17.12143 Contrast enhancement Cadollnlum Image processlng Slnuses. dural Vascular slud- les Velns. cerebral. 176.12142. 176.12143. 177.12142. 177.12143

JHRI 1- 5239-241

I From the Department of Neuroradlology. New York Unlverslty Medlcal Center. New York. NY. From the nrst 1994 meetlng of the SMR. Recelved June 24. 1994: revisLon requested August 1 0 revlslon re- celved and accepted August 18. Mdreu reprht nqoeatm to E.A.K.. MRI Department. Basement COOP Care Bldg. NYlJ Medlcal Center. 530 First Ave. New York. NY 10016.

OSMR. 1995

after gadolinium enhancement, for a va- riety of reasons including preoperative tumor evaluation (n = 21. metastatic workup (n = 1). and evaluation of an ar- teriovenous malformation (n = 2). The age range of the patients was 3-76 years. The patients were imaged with a 1 .O-T system (Magnetom Impact: Sie- mens Medical Systems, Iselin. N J ) and a protocol consisting of pregadolinium. axial turbo proton-density and T2- weighted images (TR msec/TE msec = 3.500/17. 119; 250 x 256 matrix; 210- mm field of view; acquisition time of 3 minutes 3 seconds) and the first MP- RAGE sequence (12/4. 300-msec inver- sion time, 15" flip angle, 180 x 256 ma- trix, 230-mm field of view, and an ac- quisition time of 5 minutes 32 seconds). This was followed by the intravenous bolus administration of 0.1 mmol/kg gadopentetate dimeglumine. The identi- cal MP-RAGE sequence was then re- peated with no change in frequency, gradient, or transmitter adjustments, to hold unenhancing-pixel values con- stant. The start of the second MP-RAGE acquisition occurred 1 minute after completion of the bolus injection of con- trast agent. The patients were asked to hold still during data acquisition, with no attempt made to fix the head in posi- tion.

The MP-RAGE sequence uses a cycli- cal approach a s follows: (a) a nonselec- tive inversion pulse and a variable delay time in the contrast preparation period, (bl acquisition of three-dimensional par- titions with incremental in-plane phase encoding, and (c) a recovery period. The preparatory and recovery periods allow contrast control. Data acquisition is performed with a short TR (Turbo- FLASH [fast low-angle shot]) gradient- echo sequence (6.7). The advantages of the sequence are heavily T1 -weighted contrast, a relatively high signal-to- noise ratio, postprocessing capabilities, thin contiguous images, short acquisi- tion time, and the presence of flow-re- lated enhancement. The MP-RAGE se- quence was optimized, with the recov- ery time between three-dimensional partition iterations and preparation set to zero and the addition of a phase-en- coding rewinder pulse after data acqui-

239

Figure 1. Anteroposterior pro- jection (a) and lateral projection @) MP-RAGE subtraction veno- grams (1214, 300-msec inversion time, 15' flip angle) obtained in a patient with a known slow-flow arteriovenous malformation. The slow-flow nidus can be clearly seen, along with the deep venous drainage. More important. the cortical venous drainage is clearly depicted.

a.

Figure 2. Anteroposterior pro- jection (a) and lateral projection @I MP-RAGE subtraction veno- grams (1214. 300-msec inversion time, 15' flip angle) demonstrate the hyperdynamic appearance of the venous system in a young patient with a partially treated vein of Galen malformation. Both the deep and cortical venous anatomy are clearly seen.

a.

sition. The MP-RAGE data set con- sisted of 128 1.25-mm-thick sagittal partitions.

The two MP-RAGE data sets were postprocessed in two separate stages. The flrst stage, a subtraction algo- rithm, calculates a difference image for each individual partition. on a plxel- by-pixel basis. This new data set of 128 1.25-mm-thick partitions consists of images that show gadolinium en- hancement (including that related to T1 reduction and slow flow) with back- ground suppression.

The subtracted data set was then processed with the standard maxl- mum-intensity-projection algorithm after the volume of interest was tar- geted to remove the normally enhanc- ing nasal and oropharyngeal mucosa. This enhancement, although not ob- scuring anatomy, does detract from the appearance of the venogram. De- pending on the clinical needs, a vari- ety of projections can be obtained. Routine angiographic anteroposterior and lateral projections were routinely obtained. Also, in those patients un- dergoing stereotaxic neurosurgical therapy, a stereoscopic pair of the standard anteroposterior and lateral projections were calculated by obtain- ing an additional projection with a 6" offset. thereby allowing acquisition of a three-dimensional image of the

b.

b.

venogram. Finally. a pseudo hologram can be obtained by displaying the dy- namic output of 120 projections taken in a 360" arc every 3". This mode of viewing permits visualization of the position of the cortical veins in rela- tion to the enhancing intracranial neo- plasm ("tumor blush"). if present.

The MP-RAGE venograms were qualitatively assessed in terms of over- all quality and conspicuity of deep and cortical venous structures.

0 RESULTS With the techniques described

above, a high-resolution venogram is obtained. Lesions are readily shown because of the high anatomic resolu- tion. In a patient with a known arterio- venous malformation (Fig 11, the slow- flow nidus is clearly seen, along with both the deep and, more important, the cortical venous drainage. This case illustrates the utility of the technique in pathologic cases and, more irnpor- tant. its usefulness in the follow-up of arteriovenous malformations, in which assessment of cortical venous drain- age is of paramount importance. The latter assessment has heretofore been able to be made only with conven- tional intraarterial angiography.

Figure 2 illustrates another ex- ample in which the technique depicts venous anatomy with exquisite detail.

In this young patient with a partially treated vein of Galen malformation. we clearly see the hyperdynamic venous state as manifested by increased ve- nous flow in both the deep and the cortical venous systems.

0 DISCUSSION The MP-RAGE sequence, by virtue

of its very short TE gradient-echo com- ponent, depicts normally flowing arte- rial blood as bright and slower-flowing venous blood as dark (because of saturation effects). However, after the bolus Intravenous administration of gadolinium. the slower-flowing venous blood appears bright because of the T1 -reducing effects of the contrast agent. Arterial blood shows no sub- stantial change in signal intensity. These effects are seen on images ob- tained 1 minute after injection of the contrast agent. Obviously. any other structure (either normal or pathologic) will also be seen on both image sets. However, when both of these data sets are subjected to the subtraction algo- rithm. a n y structure with discernible signal intensity on the precontrast im- ages and no change in signal intensity on the postcontrast images will be re- moved from the postcontrast images. The result is an image with slow ve- nous flow as the dominant component. Naturally, any other enhancing struc-

240 JMRI March /April 1995

tures appear more conspicuous on the venogram because of the suppression of normal background tissue by the subtraction algorithm.

The venogram obtained with the MP-RAGE subtraction technique de- picts the venous sinuses normally seen on conventional MR venograms (1-5); however, we can also clearly see the deep venous structures: the infe- rior sagittal sinus. the internal cere- bral veins. and so forth. These struc- tures are not normally visualized with conventional MR techniques. The cor- tical veins are also seen. something not accomplished with routine MR im- aging. A qualitative comparison be- tween MP-RAGE venography and con- ventional two-dimensional time-of- flight oblique parasagittal venography is currently under way in a larger se- ries of patients.

In conclusion, MP-RAGE subtrac- tion venography is an innovative means of obtaining high-resolution

MR venograms with a combination of imaging techniques and postproces- sing. The technique allows visualiza- tion of venous anatomy unseen with conventional two-dimensional time-of- flight techniques. Its most important utility is its ability to demonstrate the cortical venous anatomy. This is of paramount importance In the preop- erative evaluation of patients undergo- ing resection of intracranial neo- plasms. In thls group of patients, there is no imaging time penalty. since most routine protocols call for imaging both before and after gadolinium ad- ministration. The high-resolution MP- RAGE venogram is therefore retrospec- tively "acquired" after the patient has left the department, and is precisely tailored to address specific clinical questions on a patient-by-patient basis.

References 1. Mattle H. Wentz K. Edelman R. et al.

Cerebral venography with MR. Radlology 1991: 178:453-458.

2. Edelman R. Wentz K. Mattle H. et al. lntracerebral arterlovenous malforma- tions: evaluation with selectlve MR anel-

Y

ography and venography. Radlology

3. Edelman R. Ahn S. Chlen D. et al. Im- 1989: 173:831-837.

proved time-of-fllght MR anglography of the braln wlth magnetlzallon transfer contrast. Radiology 1992: 184:395-399.

resonance imaging of thrombosed dural slnuses. Stroke 1994: 25:29-34.

5. Yanaka K. Shlral S . Shlbata Y. et al. Gadollnlum-enhanced magnetic reso- nance anglography In the plannlng of supratentorlal glloma surgery. Neurol Med Chlr (Tokyo) 1993: 33~439443 .

6. Mugler J. Brookernan J. Rapid three- dimensional T1 -weighted MR Imaging with the MP-RAGE sequence. JMRl 1991: 1:561-567.

7. Brant-Zawadzkl M. Gillan G. Nltz W. MP RAGE: a three-dlmenslonal. T1- weighted. gradlent-echo sequence-lni- tlal experlence In the braln. Radlology 1992: 182: 769-775.

4. lsensee C. Reul J. Thron A. Magnetlc

Volume 5 Number 2 JMRl 241