Glenn S. Slavin, PhDSteven D. Wolff, MD, PhDSandeep N. Gupta, PhDThomas K. F. Foo, PhD

Index terms:Heart, perfusion, 511.12144Magnetic resonance (MR), contrast

enhancement, 511.12143Myocardium, blood supply, 511.76,

511.771Myocardium, MR, 511.121412,

511.12143

Radiology 2001; 219:258–263

Abbreviation:TI 5 inversion time

1 From GE Medical Systems, Milwaukee,Wis (G.S.S., S.N.G., T.K.F.F.); and theCardiovascular Research Foundation andthe Lenox Hill Heart and Vascular Insti-tute, New York, NY (S.D.W.). ReceivedApril 3, 2000; revision requested June 5;revision received June 30; accepted July11. Address correspondence to G.S.S.,GE Medical Systems, Johns Hopkins Hos-pital, 600 N Wolfe St, Rm 110 MRI, Balti-more, MD 21287 (e-mail: glenn.slavin@med.ge.com).© RSNA, 2001

Author contributions:Guarantors of integrity of entire study,G.S.S., S.D.W., T.K.F.F.; study conceptsand design, G.S.S., S.D.W., T.K.F.F.; def-inition of intellectual content, G.S.S.,S.D.W., T.K.F.F.; literature research, G.S.S.,T.K.F.F.; clinical studies, S.D.W.; experi-mental studies, G.S.S.; data acquisition,S.D.W.; data analysis, S.N.G.; statisticalanalysis, G.S.S., S.N.G.; manuscript prep-aration, G.S.S., T.K.F.F.; manuscript edit-ing and review, all authors.

First-Pass MyocardialPerfusion MR Imagingwith InterleavedNotched Saturation:Feasibility Study1

The authors evaluated a magnetizationpreparation scheme with a “notched”section profile for T1-weighted first-passmyocardial perfusion magnetic reso-nance (MR) imaging at 1.5 T. Thepulse sequence consisted of a prepara-tion sequence followed by an inter-leaved gradient-echo echo-planar se-quence. Image contrast was evaluatedin a feasibility study in 12 adult pa-tients. The notched saturation pulseallowed long magnetization recoverytimes without sacrificing section cov-erage. Image contrast between nor-mal and ischemic myocardium was ex-cellent.

Diagnosis of coronary artery disease is amajor challenge for noninvasive medicalimaging and an active area of research inmagnetic resonance (MR) imaging. Al-though coronary MR angiography canhelp identify the presence of intracoro-nary lesions, their hemodynamic or func-tional significance cannot necessarily bedetermined (1). Direct visualization ormeasurement of blood flow to the myo-cardium would be desirable.

The feasibility of first-pass contrast ma-terial–enhanced perfusion MR imaginghas been demonstrated by several groups(2–6), but the difficulty in optimizing allaspects of the acquisition sequence hasresulted in compromised image qualityor reduced anatomic coverage. Successfulfirst-pass perfusion MR imaging requiresthat several issues be addressed, includ-ing the following: (a) coverage of theheart from apex to base (which typicallyrequires at least six short-axis views),

(b) temporal resolution (ie, time betweenrepeated acquisitions at the same sectionlocation) that is sufficient to allow ade-quate sampling of the first pass of thecontrast material bolus, (c) a relationshipbetween signal intensity and contrastmaterial dose (ie, input function) that isquantifiable (preferably linear), and (d) con-trast- and signal-to-noise ratios that aresufficiently high to allow discriminationbetween normal and ischemic regions ofthe myocardium.

To meet these goals, current perfusiontechniques make use of fast gradient-echo MR imaging sequences in combina-tion with various magnetization prepara-tion schemes for T1 weighting, includinginversion recovery (2–5), saturation recov-ery (7–10), or partial saturation (11,12).Shortcomings of these approaches includethe following: (a) the number of sectionsacquired is limited owing to long acquisi-tion time or magnetization recovery inver-sion time (TI), (b) image contrast and sig-nal-to-noise ratio are inadequate owing touse of low preparation flip angles or shortTIs, and (c) input functions are not quan-tifiable owing to sensitivity to arrhythmias.

Regardless of the type of preparationused, the magnetization recovery TI playsa major role in determining contrast, sig-nal-to-noise ratio, and the maximumnumber of sections that can be acquired.A short recovery time permits the acqui-sition of more sections per cardiac cyclebut sacrifices signal-to-noise ratio andcontrast owing to minimal relaxation. Along TI provides higher signal-to-noiseratio and better contrast, but the delayinserted between preparation and acqui-sition reduces the maximum number ofsections that can be acquired.

The purpose of this study was to de-velop and evaluate a technique for simul-

Technical Developments

258

taneously achieving long TIs and multi-section capability (ie, multiple sectionsper cardiac cycle) for T1-weighted first-pass myocardial perfusion MR imaging.This technique was developed withoutinserting explicit delays into the sequenceor sacrificing section coverage.

Materials and Methods

Pulse Sequence

Pulse sequence timing is shown in Fig-ure 1. The pulse sequence consists of apreparation sequence followed immedi-

ately by an interleaved gradient-echoecho-planar sequence. The use of echo-planar sequences with short echo trains(ie, data acquisition windows of approx-imately 10 msec) in the heart (10,13–15)results in good to excellent image qual-ity, while providing immunity to motionand off-resonance effects that can cor-rupt images obtained with sequenceswith long echo trains (16).

The basic sequence structure used inthis study differs from previously re-ported perfusion techniques in two ways.First, as depicted in Figure 1, there is noexplicit delay for relaxation prescribedbetween preparation and acquisition.Second, the consecutive preparation andacquisition pulses do not affect the samesection. To account for the absence of an

explicit delay, this method makes use of aradio-frequency saturation pulse with aband-stop or “notched” section profilerather than a conventional section-selec-tive or nonselective pulse. As shown inFigure 1b, saturation bands are createdon both sides of a notch, which is cen-tered such that spins in the subsequentlyimaged section are unaffected by thepreparation pulse. The presence of thenotch essentially decouples the preparation-acquisition combination, causing each prep-aration pulse to affect the next (as opposedto the immediately following) section. Asa result, a longer recovery time is achievedwithout sacrificing spatial coverage or in-troducing any delay between preparationand acquisition.

The effect of the interleaved notchedsaturation is demonstrated in Figure 2.The radio-frequency preparation pulsefor section n 1 1 (t3 in Fig 2), whichpreceded the acquisition of section n (t4),saturated all tissue in the selected volumeexcept that in section n. Spins in sectionn were unaffected by this pulse becausethey fell within the notch. These spins,however, were saturated by the prepara-tion pulse (t1) that immediately precededthe acquisition of section n 2 1 (t2).Therefore, the spins in section n recov-ered for a time spanning the acquisitionof section n 2 1 and the preparation ofsection n 1 1. This time is the effective TI,which equaled 185 msec for the imagingparameters used in this study.

As shown in Figure 3, a saturation re-covery preparation with a 90° pulse and a185-msec TI can provide considerableimprovement in signal intensity and

Figure 2. Schematic of magnetization prepa-ration with interleaved notched saturation.Time points tn are defined in Figure 1a. Thetime between the preparation and acquisitionof section n (t4 and t1, respectively) is the ef-fective TI. The sections are labeled to reflect theactual temporal order of the interleaved sec-tions (odd sections first, then even sections).EPI 5 echo-planar imaging.

Figure 3. Bloch simulations show signal in-tensity response versus T1 for magnetizationpreparation with notched saturation (90° prep-aration pulse, 185-msec TI, 25° excitation flipangle) and partial saturation (45° or 60° prep-aration pulse, 10-msec TI, 12° excitation flipangle). Notched saturation produces consider-ably better contrast between short T1 (enhanc-ing) and long T1 (nonenhancing) tissue. a.u. 5arbitrary units.

Figure 1. (a) Diagram shows pulse sequence for first-pass myocardial perfusion MR imagingwith interleaved notched saturation recovery. The preparation sequence (at t1 and t3) consists ofa radio-frequency pulse with a notched section profile that saturates tissue in the passband of thepreparation volume. The resultant transverse magnetization is dephased by means of a gradientcrusher pulse Gk, which is immediately followed by a gradient-echo echo-planar sequence (t2, t4),during which all the data for one section are acquired. No delay is inserted between thepreparation and acquisition sequences. The preparation-acquisition combination is repeated forall prescribed sections across two cardiac cycles, yielding one temporal phase for each section.Thirty phases are acquired during 60 heartbeats. Time points tn are presented in the text. ECG 5electrocardiography, Gss 5 section-select gradient, Nshots 5 total number of acquisitions requiredfor a complete image. (b) Actual section profile of the notched saturation pulse.

Volume 219 z Number 1 Interleaved Notched Saturation z 259

contrast over a partial saturation prepa-ration with a 45° or 60° pulse and a 10-msec TI. Multisection partial saturationhas been reported in two clinical studies(11,12); despite low signal-to-noise ratio,the relatively high contrast material doses(0.14–0.15 mmol per kilogram of bodyweight) permitted diagnostic accuraciesof between 70% and 85% in blinded as-sessment of the perfusion data. Quantita-tive analysis of the same data proved dif-ficult, however, owing to the low signal-to-noise ratio.

MR Imaging

MR imaging was performed with a1.5-T system (Signa CV/i; GE Medical Sys-tems, Milwaukee, Wis) with 40 mT/mpeak gradients and 150 T/m/sec maxi-mum slew rate. The acquisition was aninterleaved echo-planar sequence (10)with a short echo train, which was alsoused in the partial saturation protocol.The following parameters were used forall MR imaging studies: electrocardio-graphic triggering, repetition time msec/echo time msec/TI msec of 6.6/1.4/185,echo train length of four, 20° or 25° exci-tation flip angle, 36 3 27-cm field of view,10-mm section thickness with 0–2-mmspacing (adjusted for coverage from apexto base), 250-kHz receiver bandwidth,and 128 3 128 acquisition matrix (96 ky

lines). Seven interleaved short-axis sec-tions were collected after a single electro-cardiographic trigger, with data acquisi-tion spanning two cardiac cycles. Thisyielded a temporal resolution of one im-age every two R-R intervals. Eight studieswere performed with a contrast material(gadopentetate dimeglumine, Magnevist;Berlex Laboratories, Wayne, NJ) dose of

0.10 mmol/kg, and two studies each wereperformed with doses of 0.05 and 0.15mmol/kg.

The preparation sequence was a 15-msec radio-frequency pulse with a notchedfrequency response and 2,600-kHz band-width that was used in the presence of asection-select gradient. Because sevensections were acquired in an interleavedmanner (ie, sections 1, 3, 5, 7, 2, 4, 6), thepassband needed to be at least as wide asthe largest spacing between any two con-secutive sections. In this case, section 2was saturated when the notch was at sec-tion 7. Therefore, the width of each sat-uration band was designed to be fivetimes the width of the notch; the overallwidth of the preparation volume was 165

mm. To allow some motion of the sec-tion during the TI, the notch width was50% greater than the section thickness.The total duration of the preparation se-quence was 18 msec, including time forthe gradient dephasers. The shape of theradio-frequency pulse is shown in Figure1a, and an actual section profile is shownin Figure 1b.

Patient Study

A feasibility study to evaluate imagecontrast was conducted with 12 outpa-tients (eight men and four women; agerange, 54–76 years; mean age, 66.6 years 66.2 [SD]) with known or suspected coro-nary disease. All patients had undergone

Mean Peak Contrast Enhancement for Notched Saturationand Partial Saturation Protocols

Contrast MaterialDose (mmol/kg)

Mean Peak Enhancement

ImprovementFactor‡

NotchedSaturation*

PartialSaturation†

0.05 158% 6 23§ NA NA0.10 248% 6 49 93% 6 27 2.70.15 349% 6 2\ 90% 6 23 3.9

Note.—Data are the mean 6 SD. NA 5 not applicable.* Data from studies acquired with flip angles of 20° and 25°.† Differences were not statistically significant.‡ Data are the ratio of peak enhancement for notched saturation to that for partial saturation.§ Difference compared with dose of 0.10 mmol/kg, P , .05.\ Difference compared with dose of 0.10 mmol/kg, P , .001.

Figure 5. Stress perfusion MR images depict six sections at one time point in a patient with 80%stenosis of the right coronary artery, as determined at conventional angiography. The region ofperfusion deficit in the posterior septal wall (arrows) conforms to the vascular territory of theright coronary artery and does not appear in the rest images (not shown), indicating reversibleischemia.

Figure 4. Scatterplot shows mean (Avg.) peakcontrast enhancement for notched saturationand partial saturation protocols. Note the lin-ear relationship between peak enhancementand contrast material dose with notched satu-ration (r2 5 0.9989). Error bars indicate plus orminus 2% for notched saturation with 0.15mmol/kg.

260 z Radiology z April 2001 Slavin et al

cardiac catheterization as part of theirclinical evaluation. The only exclusion

criteria were contraindications to adeno-sine or contrast-enhanced MR imaging.

All studies were performed after signedinformed consent was received from eachpatient, in accordance with a protocolapproved by our institutional review board.

Patients underwent first-pass perfusionMR studies both at rest and with pharma-cologic stress. A cardiologist was presentin the MR imaging room during all stud-ies. For imaging with stress, adenosine(Adenoscan; Fujisawa Healthcare, Deer-field, Ill) was administered intravenouslyinto an antecubital vein at 140 mg/kg/min, starting 2 minutes before the start ofMR imaging. After 2 minutes of adeno-sine infusion, gadopentetate dimeglu-mine was administered intravenouslyinto the contralateral arm at 5 mL/secwith use of a power injector (Spectris;Medrad, Pittsburgh, Pa), simultaneouswith the start of MR imaging. Patientswere instructed to hold their breath andto continue the breath hold as long aspossible during imaging. Thirty images(phases) were acquired at each sectionlocation during 60 heartbeats. The aden-osine infusion was discontinued at thecompletion of MR imaging. After 15 min-utes, the rest study was performed with asecond dose of contrast material andidentical imaging parameters.

Regions of interest were drawn by oneof the authors (S.N.G.) in the posterior oranterior septal wall of a midventricularsection for each of the 30 phases. Regionsof interest were approximately 150 mm2.In each case, signal intensity as a func-tion of time was measured and averagedfor each contrast material dose. Peak en-hancement was calculated as the differ-ence between peak and baseline signalintensity in the regions of interest di-vided by the baseline signal intensity.These results were compared with thoseof the partial saturation study (45° prep-aration pulse, 10-msec TI, 12° excitationflip angle, contrast material doses of 0.10and 0.15 mmol/kg) (11). Ten cases witheach contrast material dose in the partialsaturation study were selected retrospec-tively with use of the criterion that nolesions were detected in the left anteriordescending coronary artery with conven-tional coronary angiography. Results wereevaluated with a two-sample t test. Differ-ences with a P value less than .05 wereconsidered statistically significant.

Results

Figure 4 shows the mean peak contrastenhancement at the three contrast mate-rial doses in the notched saturation pro-tocol. The mean peak enhancement for

Figure 6. Line graph presents the percentage contrast enhancementversus time for regions of normal and ischemic myocardium fromperfusion MR imaging studies with notched saturation (NS) and par-tial saturation (PS). In addition to greater contrast enhancement withnotched saturation, the actual baseline signal intensity is desirablylower than that with partial saturation, even with the use of a higherexcitation flip angle (25° vs 12°). The notched saturation data are fromsection 4 in Figure 5. Each time point corresponds to two cardiac cycles.

Figure 7. Stress perfusion MR images depict the time course of the first pass of the contrast agentthrough a midventricular section in a patient with severe three-vessel disease, as determined atconventional angiography (not shown). An extensive region of perfusion deficit can be seen inthe septal and anterior walls of the myocardium (arrows). Images are shown at every second timepoint (1.9 second per time point).

Volume 219 z Number 1 Interleaved Notched Saturation z 261

each of the two contrast material doses inthe partial saturation protocol is alsoshown in Figure 4. Results from both pro-tocols are summarized in the Table. Sta-tistically significant differences betweencontrast material doses were noted forthe notched saturation data, but no sig-nificant difference was noted for the par-tial saturation data. Furthermore, com-pared with the partial saturation protocol,the notched saturation protocol resultedin a 3.9-fold improvement in the meanpeak enhancement with a contrast mate-rial dose of 0.15 mmol/kg and a 2.7-foldimprovement with a dose of 0.10 mmol/kg. Even with a much lower dose of 0.05mmol/kg, peak contrast enhancementwas still 1.8 times greater than that withpartial saturation at 0.15 mmol/kg. As aresult, lower contrast material doses canbe used with notched saturation, whilestill providing at least twice the peakenhancement of partial saturation. Thenotched saturation results in Figure 4also exhibit a linear relationship betweenpeak enhancement and contrast materialdose, which is not seen in the partialsaturation results. This is due to thelonger magnetization recovery TI andhigher preparation flip angle. These re-sults suggest that the current protocolcould be amenable to quantitative evalu-ation.

Figure 5 shows six sections (of the sevensections acquired) at one phase in a per-fusion MR imaging study performed in apatient under pharmacologic stress. Fig-ure 6 depicts a plot of the percentageenhancement versus time for regions ofnormal and ischemic myocardium fromthe fourth section in Figure 5. Also shownin Figure 6 are enhancement curves fornormal and ischemic myocardium from atypical partial saturation perfusion MRimaging study. As reflected in Figures 5and 6, notched saturation providedgreater enhancement and better contrastbetween enhancing and nonenhancingmyocardium. Figure 7 shows the timecourse of the contrast agent through amidventricular section from anotherstress perfusion MR imaging study.

Discussion

This study evaluated a magnetizationpreparation method for multisection first-pass contrast-enhanced myocardial perfu-sion MR imaging. Use of a radio-frequencysaturation pulse with a notched frequencyprofile enabled interleaving of the prepara-tion and acquisition sequences to pro-duce a long magnetization recovery time.

Unlike with other methods, insertion of aphysical dead time into the sequence toallow relaxation between the preparationand acquisition of each section was notnecessary. Section coverage was therebymaximized for the heart rate and imag-ing parameters. A 90° flip angle was usedin the radio-frequency preparation pulseto provide stronger T1 weighting and in-sensitivity to arrhythmias.

The long recovery time and high prep-aration flip angle also provided better im-age contrast than did short TI saturationrecovery or partial saturation and greateranatomic coverage than did conventionallong TI recovery. Furthermore, becauseblood is saturated on both sides of theimaged section immediately prior to dataacquisition, the presence of the notch pro-vides a potential mechanism for bloodpool suppression. If saturated blood movesinto the imaging plane, it may help dis-criminate the myocardial wall from theventricular space and reduce the inten-sity of flow-related artifacts. The degreeof expected in-plane blood suppression isdetermined by the width of the notch.

Because a given section was preparedbefore acquisition of the preceding sec-tion, the first section represented a spe-cial case. The first phase of the first sec-tion (ie, the very first image to be acquired)had no preparation. Subsequent phases ofthe first section, however, were preparedwith the saturation pulse that precededthe last section of the previous phase. Forexample, phase two of the first sectionwas saturated by the preparation pulsepreceding phase one of the last section.Consequently, the time between thepreparation and acquisition of the firstsection (ie, the TI) spanned an R wave.The TI for the first section was thereforelonger than that for the other sectionsand was also variable, depending on theR-R interval of the particular heartbeat.Although this effect was not detrimentalto image quality or clinical diagnosis, it ispossible to apply a custom preparationscheme for the first section.

Imaging was performed through the Rwave of the second cardiac cycle for tworeasons. First, it maintained the prepara-tion-acquisition timing for all sectionsexcept the first, as discussed previously. Ifhalf the sections were acquired in each oftwo cardiac cycles, the middle sectionwould also have a heart-rate–dependentTI. Second, imaging through the R waveobviated two trigger windows. This af-forded more time for data collection, of-ten allowing the acquisition of an addi-tional section. It should be noted thatregistration of the sections acquired dur-

ing the second R-R interval was not ap-preciably affected by modest changes inheart rate.

In light of the results of this study, weare continuing to optimize the sequenceto further improve image quality, bycharacterizing the blood pool suppres-sion and minimizing any image artifactsdue to non–steady-state effects that canresult from imaging with higher flip an-gles. Bloch simulations of a variable flipangle excitation (17,18) indicate thatpoint-spread-function side lobes can besubstantially reduced without loss ofoverall signal intensity. We confirmedthese results in preliminary studies in an-imal models (unpublished data).

This study evaluated an alternativeway of performing magnetization prepa-ration with saturation recovery. The in-terleaved preparation and acquisition se-quences and the overlapping recoverytimes allowed the most time-efficient im-plementation of long TI saturation recov-ery for multisection myocardial perfu-sion MR imaging. This implementationresulted in substantially better imagecontrast than can be achieved with shortTI saturation recovery or partial satura-tion and greater section coverage thancan be achieved with conventional longTI saturation recovery or inversion recov-ery.

References1. White CW, Wright CB, Doty DB, et al.

Does visualization of the coronary arte-riogram predict the physiologic impor-tance of a coronary stenosis? N EnglJ Med 1984; 310:819–824.

2. Atkinson DJ, Burstein D, Edelman RR. First-pass cardiac perfusion: evaluation with ul-trafast MR imaging. Radiology 1990; 174:757–762.

3. Manning WJ, Atkinson DJ, Grossman W,Paulin S, Edelman RR. First-pass nuclearmagnetic resonance imaging studies us-ing gadolinium-DTPA in patients withcoronary artery disease. J Am Coll Cardiol1991; 18:959–965.

4. Schaefer S, van Tyen R, Saloner D. Evalu-ation of myocardial perfusion abnormal-ities with gadolinium-enhanced snapshotMR imaging in humans. Radiology 1992;185:795–801.

5. Wilke N, Simm C, Zhang J, et al. Contrast-enhanced first pass myocardial perfusionimaging: correlation between myocardialblood flow in dogs at rest and during hy-peremia. Magn Reson Med 1993; 29:485–497.

6. Eichenberger AC, Schuiki E, Kochli VD,Amann FW, McKinnon GC, von Schul-thess GK. Ischemic heart disease: assess-ment with gadolinium-enhanced ultra-fast MR imaging and dipyridamole stress.J Magn Reson Imaging 1994; 4:425–431.

7. Tsekos NV, Zhang Y, Merkle H, et al. Fastanatomical imaging of the heart and as-sessment of myocardial perfusion with ar-

262 z Radiology z April 2001 Slavin et al

rhythmia insensitive magnetization prep-aration. Magn Reson Med 1995; 34:530–536.

8. Wilke N, Jerosch-Herold M, Wang Y, et al.Myocardial perfusion reserve: assessmentwith multisection quantitative, first-passMR imaging. Radiology 1997; 204:373–384.

9. Sakuma H, Kubo H, Takeda K, et al. As-sessment of myocardial perfusion usingsaturation-recovery echo-planar sequence(abstr). In: Proceedings of the Fifth Meet-ing of the International Society for Mag-netic Resonance in Medicine. Berkeley,Calif: International Society for MagneticResonance in Medicine, 1997; 843.

10. Ding S, Wolff SD, Epstein FH. Improvedcoverage in dynamic contrast-enhancedcardiac MRI using interleaved gradient-echo EPI. Magn Reson Med 1998; 39:514–519.

11. Wolff SD, Day RA, Santiago L, et al. As-sessment of first-pass mycardial perfusionimaging during rest and adenosine stress:

comparison with cardiac catheterization(abstr). In: Proceedings of the Seventh Meet-ing of the International Society for Mag-netic Resonance in Medicine. Berkeley,Calif: International Society for MagneticResonance in Medicine, 1999; 305.

12. Arai AE, Epstein FH, London J, et al. Di-pyridamole stress MRI perfusion: clinicalevaluation using high contrast dose, mul-tislice imaging, and a novel read tech-nique (abstr). In: Proceedings of the Sev-enth Meeting of the International Societyfor Magnetic Resonance in Medicine.Ber-keley, Calif: International Society forMagnetic Resonance in Medicine, 1999;304.

13. Wielopolski PA, Manning WJ, EdelmanRR. Single breath-hold volumetric imag-ing of the heart using magnetization-pre-pared 3-dimensional segmented echo-planar imaging. J Magn Reson Imaging1995; 5:403–409.

14. Slavin GS, Riederer SJ, Ehman RL. Two-dimensional multishot echo-planar coro-

nary MR angiography. Magn Reson Med1998; 40:883–889.

15. Epstein FH, Wolff SD, Arai AE. Segmentedk-space fast cardiac imaging using an echo-train readout. Magn Reson Med 1999; 41:609–613.

16. Farzaneh F, Riederer SJ, Pelc NJ. Analysisof T2 limitations and off-resonance ef-fects on spatial resolution and artifacts inecho-planar imaging. Magn Reson Med1990; 14:123–139.

17. Mugler JP, Epstein FH, Brookeman JR.Shaping the signal response during theapproach to steady state in three-dimen-sional magnetization-prepared rapid gra-dient-echo imaging using variable flip an-gles. Magn Reson Med 1992; 28:165–185.

18. McKinnon G. Ultrafast interleaved gradi-ent-echo-planar imaging on a standardscanner. Magn Reson Med 1993; 30:609–616.

Volume 219 z Number 1 Interleaved Notched Saturation z 263