IEEE Transactions on Nuclear Science, Vol. NS-29, No. 3, June 1982
QUANTIFICATION OF OBSTRUCTIONS IN VESSELS BY NUCLEAR MAGNETIC RESONANCE (NMR)
Lawrence Crooks, Phil Sheldon, Leon Kaufman, and William RowanUniversity of California, San Francisco, Radiologic Imaging Laboratory
and Theodore MillerUniversity of California, San Francisco, Department of Pathology
Abstract
The sequence of radiofrequency and magnetic fieldgradients used to form an NMR image produces changesin the signals emitted by moving blood when comparedto the signals from stationary blood. These changescan be used to measure the open area of vessels. Inaddition, specific relaxation time signatures oflesions may help in identifying the nature of theobstructions.
Introduction
NMR imaging is based on the ability to induce andmonitor resonance of the magnetic moment of nuclei inthe presence of magnetic fields. By the use of mag-netic fields whose strength varies with position, itis possible to define both the location andconcentration of resonant nuclei, and, thereby tocreate images that reflect their distribution intissue. Hydrogen, because it is the most sensitiveof the stable nuclei to NMR and because it is alsothe most abundant nucleus in the body, is ideallysuited for NMR imaging.
The basis of nuclear magnetic resonance is theproperty of all nuclei with an odd number of protonsor neutrons, or both, to act as magnets. In the ab-sence of a magnetic field, the axes of these nucleipoint in random directions; however, when placed in astrong magnetic field, they align with the field. Ifweak radio-waves of the proper resonant frequency arethen applied, the nuclei will flip and alignthemselves against the magnetic field. In a fixedmagnetic field, as the nuclei flip back to theiroriginal alignment, they send out radiowaves of thesame frequency as the ones applied. This sequence isknown as nuclear magnetic resonance (NMR).
In any given magnetic field the frequency of theradio-wave that will cause resonance is specific foreach element. Hydrogen in a magnetic field of 3.5 KGwill resonate at 15MHz. For all elements, theresonant frequency changes in direct proportion tothe strength of the surrounding field. For instance,hydrogen in a magnetic field of 7 KG will resonate at30 MHz. This frequency is related to the magneticfield at a nucleus, HLOCI by f=yHLOC' where f isfrequency, and Y is a constant, the magnetogyricratio of the particular type of nucleus.
NMR ImagingTwo concepts are particularly important for anunderstanding of NMR imaging. First, because theradio-waves emitted by nuclei in the magnetic fieldscommonly used are between 10 and 100 meters inlength, images cannot be made through opticalprocesses, such as those used with ultrasound,x-rays, and gamma-rays. Instead, since the frequencyof the radio-waves is proportional to magnetic fieldstrength, if the latter's distribution in space isknown, the frequency spectrum of NMR signals encodesthe spatial distribution of the nuclei emitting thesesignals. Thus, identical hydrogen nuclei placed in amagnetic field whose strength varies in space willemit radio-waves of different frequencies-- thestronger the field, the higher the frequency. Since
the field variation is known, a plot of signalintensity vs. frequency is equivalent to a plot ofintensity vs. position, the two factors necessary tocreate an image.
The second concept important for understanding NMRimaging is that the intensity of the signal is notsimply a reflection of hydrogen density. Becausethe signals are being produced by hydrogen nuclei,it might be expected that they would be reflectionsonly of the distribution of hydrogen. In actuality,the observed intensity is hydrogen density (H)strongly modulated by local physical and chemicalfactors, including molecular structure, elementalcomposition, temperature, and viscosity. All thepreceding factors affect the rate at which nucleialign with the external magnetic field (T1) and therate at which nuclear energy emission decays (T2).Commonly referred to as "magnetic relaxation times",Ti and T2 are exponential time constants that differin different tissues. For example, very pureliquids, in general, align less quickly and emitenergy for a longer time than liquids with proteins.
Varying the interval between the successiveexcitations of a region of the sample willselectively enhance tissues according to Ti. If theinterval between successive excitations is short,the tissues with longer Ti will yield relativelyless signal than those with short Ti since theformer have less of a chance to become fully alignedbefore a new excitation is started. Varying thetime interval between nuclear excitation andobservation of the signal will selectively enhancetissues according to T2, the tissues with longer T2values providing relatively larger signals thanthose with short T2.
Another factor will change the NMR intensity in alive object: If the hydrogen nuclei move throughthe volume being imaged in the time it takes to per-form one data acquisition sequence (approximately 50msec), their signal is lost. The reduction inintensity depends on the fraction of hydrogen atomsthat are moving and on their velocity.
Thus, NtMR intensity images, because they represent acomplex folding of physical characteristics of thetissue and of the instrument parameters, can bemanipulated to highlight a particular type oftissue.
These concepts are expanded upon in reference 1.
Methods
The results presented here were obtained with an NMRimager of 6.5 cm aperture operated at a 3.5 KGaussfield. Selective irradiations (2) are used todefine the plane of the image, and up to fivecontiguous planes are obtained simultaneously. Thestandard imaging mode used for these investigationshas a spatial resolution of 0.5 x 0.5-mm in a4mm-thick section, with images of the full fieldobtained in approximately 4 minutes.
0018-9499/82/0600-1181$00.75© 1982 IEEE 1181
Any one image is a two dimensional representation ofthe NMR intensity (I) under the given imagingconditions. In the imager used here, I is given bythe expression
I = H f(v) exp(-a/T2)[1-exp(-b/T1)] (eq. 1)
where I is the NMR intensity in a particular regionof the image; H is the local hydrogen density; a isthe T2 parameter of the instrument, measured inmilliseconds and varied within a broad range undercomputer control; b is the Ti parameter of theinstrument, measured in seconds and alsocomputer-controlled; f(v) is a function of both thespeed with which hydrogen nuclei move through theregion being imaged and of the fraction of the totalnumber of nuclei that are moving. At present, theeffects of H and f(v) are not separated, thushydrogen images (H) are modulated by f(v), as Figure1 demonstrates.
Results
AreaA semi-automated boundary definite program is used tomeasure the area where flow occurs.
The ability of NMR to recover area imaging isexcellent. A phantom consisting of plastic rods of1.55 to 15.75 mm-diameter (which simulate areas offlow) was imaged and the area measured by NMR wascompared to actual measurements. The results areshown in Figure 2. These are shown for the casewhere the boundary of the area of "flow" is set atthe point where the NMR intensity reaches a valuew=50% of the intensity between the inside and outsideregions (which are marked by the operator). Thesetting of w is not critical except for the smallest"vessels", as shown in Figure 3. The ability tomeasure open and blocked areas of tubes simulatingflow in vessels was assessed by the use of aninflatable balloon catheter containing the same typeof liquid as being flown in the tube. Results areshown in Figure 4.
FlowSince signal intensity is so critically dependent onimaging sequence, it would be expected that movingnuclei will produce little or no signal, and, indeed,this is what is observed in vessels in animals andhumans (Figures 1 and 5). In practice, the use ofselective irradiations results in a more complexbehaviour. At very low velocities the nuclei movinginto the volume being imaged have not undergoneprevious irradiation sequences, and are consequentlypolarized to the full equilibrium value at the givenmagnet field strength. These "fresh" nuclei producemore signal than those that are stationary becausethe latter are partially depolarized by repeatedradio frequency excitations. Therefore, the signalfirst increases with increasing flow, a phenomenon wecalled paradoxical enhancement. As the flowcontinues to increase, the disruption of the imagingsequence becomes the dominant factor and intensitydecreases in a near linear manner with velocity. Inmulti-section imaging, paradoxical enhancement issignificant mostly for the outside sections, since bythe time the fluid reaches an inner section itsnuclei have undergone at least one irradiationsequence. Figure 6 shows a graph of NMR signalintensity vs. velocity. Figures 7 shows paradoxicalenhancement in a 9.66 mm-diameter tube. The effectsobserved at higher velocities are shown in Figures 8and 9. Decrease of signal intensity with velocity is
also observed for motion in the plane of the image,except that in these cases paradoxical enhancementdoes not come into play.
Lesion CharacterizationWork with excised human vessels indicates that inaddition to outlining areas where blood flows invessels, NMR imaging may be able to characterize thenature of lesions. Figure 10 shows across-sectional image of a lesion in an iliacartery. The components of the lesion are welldifferentiated.
Discussion
Because of its apparently benign nature (3) NMRimaging may provide a convenient and safe method fordetecting and quantifying the degree of obstructionof vessels deep in the body. Whether the coronaryvessels can similarly be studied will depend on thesuccessful application of gating or stop-motiontechniques. Although lmm-diameter vessels arevisible in small animals, the size of the smallestvessels that will be accessible to study in humansis still to be defined. The use of contrast agentscould add a further dimension of capabilities(Figure 11).
Acknowledgements
These investigations are supported in part by USPHSContract N01 HV02928 from the NHLBI and by Diasonics(NMR), Inc.
References
1. NMR Imaging in Medicine. L Kaufman, LE Crooksand AR Margulis, Editors. Igaku Shoin, NY, NY,1981.
2. LE Crooks. Selective Irradiation Line ScanTechniques for NMR Imaging. IEEE Trans Nucl SciNS-27: 1239, 1980.
3. TF Budinger. Potential Medical Effects andHazards of Human NMR. In reference 1, chapter10.
Figure 1. NMR transverse section through the upperabdomen of a supine rat. The top row is of the liverat. The rat was killed in place and imaged again(bottom row). Images at left were obtained with aand b parameters as indicated (see equation [1]).Note the hepatic vein (HV), inferior vena cava(IVC), and spinal cord (SC). In the dead animal theHV and IVC blend into the liver, and the aortaappears bright against the darker back muscle. Theliver appears blotchy in the live animal, and quiteuniform in the dead, as evidenced by hydrogenimages.
1182
20
10
Z 5
2
1 2 5 10 20 S0 100 200
ACTUALAREA mni2I
Figure 2. Solid plastic rod surrounded bystationary fluid is used to simulate vessels in thisphantom. The area recovered from the NMR image ofthe phantom (inset) matches well the actualcross-sectional area of the rods.
so
70
so
I
4Uaz
50
40
30
20 -
10
a
.. 4.9 mm
*3.24'ty'nU U UU
0 0.1 0.2 0.3
VOLUME (ml)
Figure 4. An inflatable balloon catheter containingthe same fluid as being circulated in the tubes isused to measure by NMR the open area of tubes ofdifferent diameters. Because the length of theobstruction is reasonably constant, its area isapproximately proportional to its volume. The solidwalls of the catheter decrease the open area of thetube even when the balloon volume is zero.Consequently, the intercept is at less than the fulltube area. The inset shows four contiguous NMRimages of the phantom, obstructed in section 2.Coronal reconstruction of the four sections shows thedirection of flow.
Figure 6. NMR intensity obtained with a and bparameters as indicated using a fluid of .52 sec and230 msec Ti and T2 values, respectively. The sectionin which the fluid enters the imaged volume shows theeffects of paradoxical enhancement. The exit sectionand horizontal flow do not show enhancement, and aresimilar to each other.
-0 DOterd-4 1S.75o- w ~~~~~~12.77
o 00.55-12 ° 7.95
+~~~+IS
t...
30 40 50 60 70
W (%)
Figure 3. Area of a vessel is measured by a programthat automatically finds the boundary where the NMRintensity reaches a value w between the averageintensity inside and outside the vessel wall (bothindicated by the operator). Except for the smallestvessels, the setting of w is not critical, since theNMR area changes by less than +10% over a w range of±20%.
Figure 5. NMR image of the mid section of a humantorso, showing spinal cord, aorta and inferior venacava (IVC). Flow permits visualization of thesevessels without the administration of contrastagents.
VELOCITY (cm/sec)0 0.5 1 1.5 2 3 4 S 7.5 10 12.5
120 . .. ..
Z s b0.5e X
FLOW (mUmin)
1183
r0 00° °0000 oQO
CIc
Figure 7. Paradoxical enhancement is evidenced inthe NMR intensity image of a 9.6-mm inside diametertube through which fluid flows at a velocity of 1.3cm/sec. Note the brighter appearance of the movingfluid compared to that of the stationary fluidoutside the tube. Contour map shows intensityincreasing from the rim of the tube (where flow is atlower velocity) to its center, where velocity ishighest.
Figure 8. Five NMR images of a flow phantomconsisting of three plastic tubes that run throughits length (top row), and a U-shaped tube with ahorizontal arm. The ends of the U are seen in section1, and the walls of the horizontal arm in sections 2and 4. The center of the horizontal tube is imagedin section 3. The coronal reconstruction of theU-tube shows the horizontal tube walls withintermediate intensity because of partial volumeeffects. The fluid inside and outside the tubes(which are connected in series) is stationary.
Figure 9. For a flow of 355 cc/min the NMR Intensityinside the tubes drops to background levels, exceptalong the inner wall of the largest tube, where theslower boundary layer still produces some signal.
Figure 10. Five contiguous sections of an iliacartery with a lesion occupying approximately 50% ofthe lumen. This lesion consists of a large plaquewith orderly connective tissue around thecircumference. The central area (seen in section 2)contains amorphous debris, which contains somelipids. Acconpanying photograph shows this lesion.
1184
Figure 11. Because of blood flow and heart motionthe chest of the live rat normally appears with lowintensity in NMR images obtained with our system(left). The use of contrast agents designed for NMRimaging allows visualization of the left and rightventricles as well as of vasculature in the rat chest(center and right). Top and bottom row are fordifferent values of a in equation 1. These imageswere obtained using contrast agents developed byMilos Sovak of the University of California, SanDiego.
1185