WO1987001208A1

WO1987001208A1 - Measurement of capillary flow using nuclear magnetic resonance

Info

Publication number: WO1987001208A1
Application number: PCT/US1986/001693
Authority: WO
Inventors: H. Samuel Patz; Robert C. Hawkes
Original assignee: Brigham And Women's Hospital
Priority date: 1985-08-14
Filing date: 1986-08-13
Publication date: 1987-02-26
Also published as: IL79691A0; AU6222886A; EP0232387A1

Abstract

An improved method for measuring very slow flow rates using nuclear magnetic resonant techniques. The basic technique is that of steady state free precession, in which a sequence of radio frequency pulses are applied to nuclei in a magnetic field having a substantial gradient, so that a spatial periodicity in the magnetization of the nuclei is established. The nuclei reach a state of driven equilibrium by application of radio frequency pulses to the sample. Two images are generated, using different time intervals between the application of the radio frequency pulses. One image is subtracted from the other, which cancels out any static nuclei in the signal, while relatively fast flowing nuclei, namely in the larger blood vessels or the like, never reach the equilibrium state. The remainder is therefore proportional only to nuclei which are part of relatively slowly flowing liquids, such as in capillary blood flow in organs.

Description

MEASUREMENT OF CAPILLARY FLOW USING NUCLEAR MAGNETIC RESONANCE

TECHNICAL FIELD

This invention relates to improvements in nuclear magnetic resonance measurement techniques. More particularly, this invention relates to methods for using nuclear magnetic resonance techniques for deter¬ mination of extremely low flow rates, such as blood flow in capillaries.

BACKGROUND ART

The techniques of nuclear magnetic resonance are well known to the art. In general, nuclear magnetic resonance involves aliqning the magnetic moments of atomic nuclei in a sample by exposing the sample to a relatively strong external magnetic field. A pulse of radio frequency energy is then applied to the sample, to cause the moments of the nuclei to be aligned along a particular axis, typically 90° to the axis of the external magnetic field. Over time, the nuclei will return to alignment with the external field. As they do so, they emit electromagnetic radiation which can be detected. The rate at which the moments of the nuclei return to alignment with the external field is characteristic of the nuclei and the nuclear site.

SUBSTITUTE SHEET This fact is used in a variety of methods for generat¬ ing images of the density of the nuclei, for example, in a "slice" of human body or other sample in which they are located. In this way, a detailed cross-sec¬ tional view of the body is provided, in a non-invasive technique.

A number of different types of NMR-produced images are available, in which the intensity of each element of the image varies with a different parameter. Tissue type is perhaps the most usual. Images in which the intensity of each element varies with blood flow rate have also been provided, as discussed below.

One image which is not presently available is one in which the intensity of the elements of the image is proportional to flow at very low rates, specifically the rate of blood flow in capillaries of organs. This would be of great relevance in determination of the health of organs, and also to determine whether ade- uate blood is being supplied to them.

The use of nuclear magnetic resonance for measure¬ ment of flow of liquids is well known in the art, and several different techniques have been proposed. Moore _____ l« U.S. Patent 4,015,196 suggests the application of a technique known as steady state free precession to the study of flow. The steady state free preces¬ sion technique is defined generally at column 7, lines 15-57, of the Moore et al. patent and is related to flow among other uses at column 7, lines 58-61. How- ever, for a variety of reasons, which are discussed in detail below, the steady state free precession (SSFP) technique has not been extremely popular in the art. Broadly, the main reason for this is that the SSFP technique involves repetitive supplying of bursts of

ISA®

SUBSTITUTE SHEET radio frequency (RP) energy to a sample in a magnetic field having a strong gradient. The amplitude of the 5 relaxation signal is proportional to the degree to which the magnetic moments are in phase. Upon appli¬ cation of the RF pulse, they are completely in phase. Thereafter, the relaxation signal from the excited nuclear magnetic moments first decreases, then in- 0creases as the moments become rephased, such that the signal reaches a peak synchronized with the applica¬ tion of the radio frequency energy. Because the RF energy is synchronized with the peak in the relaxation signal, the peak—the most "informational" portion of 5the signal—cannot be detected. Accordingly, the SSFP techniques proposed to date are not capable of provid¬ ing the complete picture of the signal resulting from the relaxation of the spins towards the external mag¬ netic field. Furthermore, SSFP techniques provide oonly certain image generation possibilities, as dis¬ cussed in detail below.

The art has accordingly gone to a somewhat differ¬ ent technique, not requiring the repetitive RF pulses of SSFP, referred to as "spin echo imaging." Many 5references discuss spin echo techniques generally. The applications of spin echo techniques to flow are dis¬ cussed in a number of articles. See "Nuclear Magnetic Resonance Blood Flow Measurements in the Human Brain" by Singer and Crooks, Science, Vol. 221, pp. 654-656 0(1983); "NMR Diffusion and Flow Measurements: An Introduction to Spin Phase Graphing" by Singer, J. Phys. B: Scientific Instruments, Vol. 11, pp. 281-291 (1978); "The Spatial Mapping of Translational Diffu¬ sion Coefficients by the NMR Imaging Technique," 5Taylor and Bushell, Phys. Med. Biol., Vol. 30, No. 4,

^' -4-

pp. 345-349 (1985); Modern Developments in Flow Measurement, Ch pter?--2-1: "Recent Measurements of Flow Using Nuclear Magnetic Resonance Techniques" by Singer and Grover, pp. 38-47 (1971); "Using NMR to Measure Blood Flow Volume and Velocity" by Singer and Crooks, Barrington Publications, Inc.; and U.S. Patent 4,520,828 to Burl et al. All of these references, as well as the Moore et al. patent, are incorporated by reference herein.

An example of the prior art flow measurement technique used is generally as follows. An initial magnetic field H_Q is supplied to cause all the nuclear spins to line up. They are then in-phase in one direction by the application of a RF pulse, using what is referred to as a 90° pulse, to indicate that the spins are rotated 90° with respect to the applied magnetic field H«. A magnetic field having a gradient (that is, the magnetic field varies in the plane of the "slice" in which is to be measured) is applied, which causes the polarized spins to dephase as a function of time. After a specific time f, a second radio frequency pulse is applied, which again tips the spins in the slice. Nuclei of, for example, blood, which have flowed into the slice of the body during the period " are tipped by application of the second radio frequency pulse, as are any which had become realigned with the external field H_Q in the interval. Ordinarily, the interval ~C will be much less than the typical relaxation time T,, such that the contribution of the latter effect is small. Accordingly, the signal detected due to subsequent return of the nuclei which

ISA/OS. were tipped by application of the^" second pulse con¬ tains a component-proportionate to blood flow, as well as a smaller component proportional to the amount of nuclei which relaxed during the time T.

According to the usual technique, a number of different measurements using a number of gradually increased times X. are performed. At some point, a maximum signal value will be reached, indicating that all the blood in the slice has been replaced during the period "C.

It will be appreciated by those skilled in the art that the technique just described, as is true of other NMR techniques (which provide signals proportional to nuclear density or^' the like), provides the data for a single projection of the image slice for each experi¬ ment, an experiment being the sequence of events just described. If an image proportional to flow is desired, the gradient is then varied and the experi- ment is reperformed a number of times. The number determines the ultimate resolution of the image. The signals are functions of the relaxation time of the various nuclei. The time-dependent signals are Fourier-transformed, to become functions of frequency. As is well known, the nuclei precess about the external field at a frequency referred to as the Larmor frequency, which is proportional to the mag¬ netic field at their location. The signal emitted is a function of the precession frequency. Since a gradi- ent has been imposed upon the external field, the Larmor frequencies of nuclei at different positions within the slice vary. Accordingly, the Fourier

ISAjUJ. SUBSTITUTE SHEET -6-

transform, providing as it does a frequency distribu¬ tion of the nuclei,-'Simultaneously provides a spatial distribution of the nuclear density within the matrix. Hence, the Fourier-transformed data can be used to directly form an image. This imaging technique, of course, is well known to the prior art, and is referred to here only to provide a basis for the subsequent discussion.

The Singer and Crooks technique just described requires knowledge of vein volume in order to generate an actual flow rate value. This requires that the vein or artery through which flow is to be measured be large enough that it can actually be measured on the image, and a value for volume thus calculated. In practice, this means that the vein must be at least 2 or 3 millimeters in diameter. Capillary flow is therefore not measurable using this technique, because capillaries are typically too small to show up indi¬ vidually on the image. Furthermore, the technique i pliedly assumes that the flow is essentially perpen¬ dicular to the plane of the,slice. Capillary flow may be thought as a convoluted flow, and cannot be assumed to be in any given direction at any given time, such that this assumption will not hold true. Furthermore, the best result reported by Singer and Crooks is measurement of flow at a rate of about 2 centimeters per second: typical rates of capillary flow are lower by a factor of at least about 20. For all these reasons, the Singer and Crooks technique is not appli¬ cable to very low flow rates in very small vessels, such as capillary flow.

^SUBSTITUT -7-

DESCRIPTION OF THE INVENTION

It is accordingly an object of the invention to provide an NMR flovr^'rate measurement technique which provides reasonably accurate values for the rate of flow in extremely small conduits, such as blood flow in capillaries.

It is a further object of the invention to provide a method for measuring flow rates using nuclear mag¬ netic resonance techniques in which a steady state free precession, driven equilibrium state is maintain¬ ed, whereby contributions to the spin-relaxation signal from static matter and relatively fast flowing matter is disregarded, due to the specific technique employed, and wherein only very slow flow contributes to the signal, such that accurate imaging of low flow rates ca.n be provided.

It is a further object of the invention to provide a nuclear magnetic resonance technique for physical imaging using steady state free precession techniques, in which the radio frequency pulses used to drive the system into a driven equilibrium state are prevented from substantial interference with reception of the received signal, such that the full information con¬ tent is retrieved from the relaxation signal.

It is a further object of the invention to provide an NMR imaging system in which the relative intensity of elements in the image relates to the relative rate of capillary blood flow, whereby variations in flow rate in tissues can be observed, so as to enable evaluation of the relative health of the tissues as measured by the blood supply thereto.

-8-

Other aspects and objects of the invention will be apparent to those of skill in^' the art. The present invention achieves the needs of the art and objects of the invention mentioned above by its provision of an improved NMR technique in which two images are generated using steady state free pre¬ cession techniques. Subtraction of one image from the other provides an image in which the intensity of each image element is proportional only to the density of nuclei flowing slowly therethrough. An image of capillary flow rates is thus provided.

The steady state free precession technique, as is understood in the prior art, involves application of repetitive pulses of radio frequency energy to a sample which is in^* a magnetic field of relatively high gradient, such that the nuclear magnetization which is established in this "driven-equilibrium" state has a spatial periodicity along the direction of the mag¬ netic field gradient. The spatial periodicity varies as well with the interpulse ti e^' . The magnetization of the nuclei oscillates in a periodic manner with motion along the direction of the gradient. The length over which one oscillation of magnetization occurs is the spatial periodicity. The spacing can be selected to be equal to the resolution of the image generating device. Thus, for example, in a sample having an overall dimension of 20 centimeters, used with an image production device having 256 pixels each in the X and Y dimensions, the resolution is selected to be about 0.8 mm. In this way, the signals from nuclei in a full spatial interval of the magnetization are mapped to a single pixel in each of X and Y. According to the invention, two images are produced, each with different interpulse spacings ~C, andχ₂ ~-ⁿ t-i-ne of the RF pulses, such that the spatial periodicity of the data used to generate the images differs. When the Fourier transform process is applied to transform the data from time-based samples to frequency-based samples, the spatial periodicity is automatically compensated for, such that the images can be directly subtracted from one another. When this is done, fixed components (i.e., due to non- moving nuclei in the tissues) disappear from the signal, leaving a portion of the signal which is proportional only to flow.

By careful selection of the spatial periodicity by control of the interpulse interval, nuclei flowing at relatively high flow rates, such as blood in veins, arteries and the like, do not appear in the signal because they do not remain in the spaced locations long enough to arrive in the driven equilibrium state.

The result is that only slow flow rate nuclei in capillaries and the like, which move a distance much less than the spatial interval in the interpulse period, contribute to the signal. The difference between the two images is thus proportional to the flow within certain velocity limits. These limits encompass the range of capillary flow.

BRIEF DESCRIPTION OF THE DRAWINGS

The invention will be better understood if reference is made to the accompanying drawings in -10-

which:

Figure 1 shows^' schematically the experimental apparatus;

Figure 2 shows the relative orientation of the external magnetic field H« and of the gradient G through a slice of tissue S to be imaged;

Figure 3 shows the relative orientation of mag- netization of nuclear spins across the slice S, and defines the spatial periodicity thereof;

Figure 4 shows prior art steady state free pre¬ cession (SSFP) techniques, and comprises:

Figure 4a, showing the amplitude of the radio frequency pulses as a function of time;

Figure 4b, showing the relaxation signal as a function of time;

Figure 4c, showing the gradient which is applied as a function of time; and Figure 4d, showing the detected signals a function of time;

Figure 5 shows the SSFP technique according to the invention, and comprises:

Figure 5a, showing the sequence of radio frequency pulses applied;

Figure 5b, showing the gradient applied; Figure 5c, showing the detected signal, all as functions of time; and

Figure 6 shows the stages in processing of the signals generated according to Figure 5, and com¬ prises:

Figure 6a, showing the detected signal as a function of time for a first interpulse intervalT -, ;

E SHEET Figure 6b, showing the detected signal as a function of time for. a second .interpulse interval 7_T₂7 Figure 6c, showing the signal of Figure 6a after Fourier transformation;

Figure 6d, showing the signal of Figure 6b after Fourier transformation; and

Figure 6e, showing the difference between the signals of Figures 6c and 6d, which is the signal proportional to slow flow through the slice, as a function of location.

BEST MODE OF CARRYING OUT THE INVENTION As discussed briefly above, the basic techniques of NMR imaging are well known to the art. Therefore, when a particular technique or method is not described in detail herein, it is to be understood that it is generally within the prior art; reference is made generally thereto and specifically to the publications incorporated by reference above, for more detailed discussions of such matters. The disclosure herein is meant to support the claims hereof, and not to provide an exhaustive review of NMR techniques generally. Referring now to Figure 1, a schematic view of experimental apparatus for performing NMR imaging is shown. A sample is confined within a pair of coils 10 and 12. Coil 12 is connected to radio frequency oscillator 14 by a switch 16 controlled by control device 24, so that when switch 16 is closed, a pulse of radio frequency energy is applied to the sample therewithin. Coil 10 is connected to detector and processing devices indicated generally at 18, which may be all as disclosed in the prior art, as is

ISA _. control device 24. See, for example, the Singer review article referred to ,a_bove. The sample and the two coils 10 and 12 (which may be the same) are all confined within a large-valued magnetic field H_Q shown schematically as being pro¬ vided by the two poles of a large magnet 20. An addi¬ tional field is provided by additional magnets 22. These are controlled by control device 24 to provide a magnetic field having a gradient G which may be con¬ trolled individually in the X, Y and Z directions to yield any desired net field. All this is well under¬ stood by the prior art. In particular, the gradient G of the field is provided by additional magnets shown schematically at 22, which are programmable to vary the gradient through the number of different gradients needed to complete the image. In a typical system, 256 different gradients are provided to provide 256 different X and Y values, such that a total image of approximately 64,000 pixels can be separately gene¬ rated. (The field H_Q and the gradient G are of course vectors. The conventional notation for vector fields

-_. -*

H_Q and G is not used herein for reasons of conveni- ence, and should be understood.)

As is generally the rule in NMR processing, the external field H_Q causes nuclei in the sample to take a given angle with respect to the direction of the lines of the field H_Q, and to precess therearound at the Larmor frequency. When a subsequent pulse of radio frequency energy is applied, the nuclei are caused to take a new alignment with respect to this external field, and then continue to precess at the Larmor frequency. As is well known, this precession

ISA/U5. T frequency is a function of the applied magnetic field and of the gyromagnetic ratio .of the nucleus, which in turn is a function of the species of the nucleus.

After the radio frequency pulse ceases, the nuclei return one by one to alignment with the field H_Q over a period of time. The "relaxation time" T., within which a predetermined portion of the nuclei return to alignment with H_Q, is a characteristic of the material and is used in conventional imaging techniques to identify the material. The relaxation of the nuclear magnetic moments perpendicular to H_Q is characterized by a different relaxation time T~. The times T,, ~ are used in conventional imaging techniques to iden¬ tify the materials of the sample, and to determine the actual images observed by a researcher or physician in conventional NMR use.

Figure 2 shows the gradient G of the magnetic field through a slice of a sample to be measured, e.g. a section through the torso of a patient. The slice is not necessarily perpendicular to the external field H_Q. S shown, the variation in the magnetic field, of which the gradient G is the measure, is quite substan- tial across the slice S. This is done purposely, such that the Larmor frequency of nuclei in the sample varies quite widely across the extent of the slice S. Such variation is necessary to provide images using the SSFP techniques according to the invention. Throughout the following discussion, the contribu¬ tion made by the magnetic field H_Q to the precession of the nuclear spins will be ignored, as is conven¬ tional in the art, and the discussion will be from the point of view of the rotating reference frame, that

ISA/US

SUBSTITUTE SHEET is, that of the nuclei in H_Q, again as is entirely conventional in the 'art. As also conventional in the art, terminology such as "a 90° pulse" will be used to refer to a radio frequency pulse supplied by energiza¬ tion of the coil 12 for a length of time sufficient to cause a number of nuclei to be polarized about axes 90° to the magnetic field H_Q. Figure 3 shows schematically what occurs when the radio frequency oscillator 14 and switch 16, under control of control device 24, are operated to provide a regular sequence of pulses of radio frequency energy in the slice S, disposed in a relatively strong gradi- ent field. (As a matter of interest to those of skill in the art, the gradients required by the present invention are those typically available in clinical NMR machinery; the high gradients presently found pri¬ marily in laboratory equipment, as used for example in diffusion studies, are not required by the techniques of the invention.) Regular, extremely short 90° pulses are applied at intervals ^*C . Application of a brief radio frequency pulse of energy will result in the radio frequency energy being of relatively wide band- with; as is well understood, the shorter the pulses the wider their bandwidth must necessarily be. Accord¬ ingly, the RF energy will be at the Larmor frequencies of a wide variety of nuclei. Therefore, nuclei substantially throughout the sample will be excited. Application of additional gradient fields, such that there is a gradient in each of the three dimensions, as is conventional in the art, defines the slice S. (The terms "protons" and "nuclei" are used substan¬ tially interchangeably throughout this portion of this

ISA/US_.

SUBSTITUTE SHEET specification as the nuclei involved in practical techniques are generally hydrogen nuclei, which are simply protons).

Application of a sequence of +90° and -90° pulses will cause the nuclei at intervals r to establish a magnetization which is periodic in space, as shown in Figure 3. This is referred to as the driven equili- brium state. Experimental results indicate that the driven equilibrium state is established after appli¬ cation of a number of RF pulses; stabilization is deemed to occur after a time on the order of T,.

Figure 3 shows this schematically. The nuclear spins are caused to enter a state of driven equilibri¬ um in both X and Y directions by application of a radio frequency pulse at an interval t . The magneti¬ zation (which is proportional to the alignment of the spins) in X and Y, Mx and My, respectively, is shown. The signals shown are the response measured with respect to +90 and -90° RF pulses in the X direction of the rotating reference frame. The magnetization of the nuclei in the driven equilibrium state has a peri¬ odicity t r_r which is equal to the quantity 1/GtfT. In this relation, G is the gradient and y is the gyromag- netic ratio of the nucleus (42.576 MHz/Tesla for hydrogen nuclei). Because the gradient G and the time constant ^~Z are independently controllable, any desired spacing Λr can be obtained according to the invention. What has been described thus far is the conven¬ tional steady state free precession technique, as described in the Moore e_t al. patent referred to above. Figure 4 shows signals generated using these prior art techniques. Figure 4a shows the sequence of

LSAjUS application of the radio frequency pulses in time; Figure 4b shows the. tesultant -signal; Figure 4c shows the gradient; and Figure 4d shows the detected signal, illustrating the difficulty with prior SSFP techni¬ ques.

As shown in Figure 4a, the radio frequency pulses are relatively short in time, and are separated by a time spacing T. Figure 4b shows the signal which is generated. As is understood generally in NMR, the amplitude of this signal is indicative of the degree to which the nuclei are in phase, i.e., the degree to which their spins are aligned. At the time of appli- cation of the RF pulse, nuclei throughout the slice are aligned with one another in the driven equilibrium state. In the intervals "T between application of the pulses of radio frequency energy, the nuclei will naturally precess at varying frequencies according to their location with respect to the gradient. As they precess, the nuclei emit radiation which is detected by coil 10. Upon application of the RF pulse, the nuclei at the spaced locations are in phase with one another, and the radiation is at a maximum value. Over time, the nuclei are dephased. The signal drops off accordingly. The in-phase bulk transverse magnetiza¬ tion begins to increase as the spins rephase. See Hinshaw, J. App. Phys., 47: 3709 (1976). As shown in Figure 4b, the emitted signal, which again is pro- portional to the degree of transverse in-phase magne¬ tization across the entire slice, thus shows peaks coincident with application of the RF pulses. This is the driven equilibrium state. The RF pulses ensure that nuclear magnetization is maintained at each^"C^"; if

ISA/US,

SUBSTITUTE SHEET the RF pulses were not applied, the signal would drop off relatively quicxly, although the peaks would continue to occur. Figure 4c shows the gradient in the prior SSFP techniques. As shown, the gradient did not vary during a given experiment; naturally, the gradient was varied from experiment to experiment, in order to generate a complete image. Finally, Figure 4d shows the detected signal. This is the same as the signal shown in Figure 4b, the emitted signal, but the peaks are missing because the radio frequency energy applied (as shown in Figure 4a) effectively prevents detection of the signal at those times, due to the much greater intensity of the applied radio frequency energy as compared to the detected signal, which is relatively weak, and of substantially the same bandwidth.

The fact that the peak signal could not be detected was a substantial limitation on prior art SSFP techniques; while a number of expedients to remove this deficiency was tried, none were completely successful. As discussed above, to provide a signal detection technique which avoids this problem is an object of the invention, which is achieved as will now be discussed in connection with Figure 5.

Figure 5a is identical to Figure 4a and shows the sequence of application of radio frequency pulses at intervals ^*f to a sample for SSFP imaging. In this case, however. Figure 5b shows time variation of the gradient, which is varied in synchronism with the radio frequency pulses, so as to include an opposite portion just prior to the application of the radio frequency pulse. The net gradient is still positive, however, so as to provide the spatial periodicity

ISA/US

UTE SHEE -18-

discussed above. The gradient reversal shown in Figure 5b has the effect of generating an additional peak in the detected signal (comparable to Figure 4b) earlier in the period defined by "C. The additional peak occurs between application of the radio frequency pulses of energy, as shown in Figure 5c. The peaks of the signal can now be readily detected. Those skilled in the art will understand why application of a short period of negative gradient has the effect of creating an additional peak. At any given moment, the amplitude of the detected signal, shown in Figure 5c, is a measure of the transverse nuclear magnetization, that is, the extent to which the nuclei are in phase with one another. After application of an RF pulse, whereupon the nuclei are in phase, the gradient causes them to precess at varying frequencies. As understood by those skilled in the art, the relative phase of each nucleus is the integral of its precession frequency over time. When the integral is taken, one is effectively integrating the gradient function, as the precession frequency is a function of the gradient in a given point. Hence, the phase is a function of the integral of the gradient. The net change of phase will be equal to the total integral of the gradient. Therefore, if a portion A- of the gradient is made negative, as indicated in Figure 5b, in effect, one subtracts an area A. from the positive going portion of the gradient curve, equal in area to the negative portion A₂. Since the nuclei are brought into phase upon subsequent application of the RF pulse, the point on the gradient curve at which A,

ISA/US

SUBSTITUTE SHEET -19-

begins is therefore an additional point at which the nuclei are in phase- with one another; as this is spaced in time from the application of the radio frequency pulses, the peaks in the detected signal (Figure 5c) can be accurately detected without inter¬ ference from application of the radio frequency energy. It will be appreciated by those skilled in the art that this improvement in SSFP detection techniques has application beyond the flow rate studies which are discussed in detail below, and that it will be gene¬ rally applicable to a wide variety of SSFP imaging techniques.

SSFP techniques were originally very popular when

^" nuclear magnetic resonance first became a research and medical diagnostic reality. However, SSFP provides a signal which is proportional to the ratio of T₂/T, , where, as discussed above, T₂ and T^ are time relaxa¬ tion constants which vary from material to material. ^ and ₂ are very adequately described in the litera¬ ture references referred to above and need not be further detailed here. As mentioned briefly in the BACKGROUND OF THE INVENTION section of this speci¬ fication, however, it later became more popular to use spin echo imaging techniques, which provide differen- tiable values for T₂ and T,, such that useful images showing different tissues (e.g., gray and white matter) could be more readily made using these techni¬ ques. These techniques, however, are much slower than SSFP, typically on the order of three times slower, such that SSFP has some advantages. It is possible.

SUBSTITUTE SHEET therefore, that the method of the invention, compris¬ ing the step of reversing the gradient for a portion of the interpulse interval, so as to cause the SSFP image to be formed other than at application of the radio frequency pulse, may allow SSFP to be of significant resumed interest.

Figure 6 shows how the SSFP techniques of the invention just described can be used to generate an image which shows relatively slow flow as a physical characteristic of a sample. A series of experiments as described in connection with Figure 5 (all having different gradients) are performed. The number of experiments provides the limit on the resolution of the eventual image. Typically, for an image of 256 x 256 image elements, 120 separate experiments may be performed. In each, the time constant ^"t» is identi¬ cal, but the direction of the gradient applied varies somewhat, so as to define a spatial periodicity in different directions, in effect defining a series of projections. Then a second series of experiments is done. Figure 6b, in which a different interpulse interval T₂ is employed, which is different from f , and hence provides a different grid spacing.. Thus, a second set of 120 experiments, each again having a different gradient orientation, is performed.

The result of each of the two sets of 120 experiments is a digitized record of the SSFP signals' dephasing and rephasing as a function of time during the intervals'!.. and^"C^* ₂. Each record is then Fourier- transformed to yield data which are functions f^ and f of frequency. Figures 6c and 6d, respectively.

ISA/US.

SUBSTITUTE SHEET Because the gradients provide frequency variation across the slice, such that the frequency varies with spacing, the Fourier-transformed signals provide image information (relative to the density of the nuclei and the ratio T_j/ ^) directly. One of the Fourier-trans¬ formed signals is then subtracted from the other; the remainder is shown in Figure 6e. This subtraction removes all static nuclei from the signals. Moreover, fast flowing nuclei, such as in blood veins and arteries, move too fast to attain driven equilibrium according to the SSFP technique employed according to the method of the invention and hence do not appear in the frequency-transformed data. The result is that the amplitude of the signal shown in Figure 6e relates only to slpw flowing nuclei, such as in capillaries and the^" like, which are not static, but which flow slowly enough that they are driven into equilibrium by the sequential application of the radio frequency pulses, all as discussed above in connection with Figure 5.

The individual data elements then are supplied to conventional display means for image generation, for example, using the conventional back-projection technique, as indicated schematically on Figure 6. Alternatively, the images could be formed prior to subtraction.

Those skilled in the art will have no difficulty in implementing the invention given the above disclo- sure; apparatus for performing the method of the invention is commercially available from a number of manufacturers.

SUBSTITUTE SHEET It will be appreciated that there has been described a method for nuclear magnetic resonant ixnag- ing of a physical body using data relating only to nuclei which flow slowly, and that a method has further been described for improved SSFP nuclear magnetic resonant measurement techniques. While a preferred embodiment of the invention has thus been described, it will be appreciated by those skilled in the art that further modifications and improvements can be made thereto and therefore that the invention should not be limited by the above exemplary disclo¬ sure, but only by the following claims.

SUBSTITUTE SHEET

Claims

CLAIMS:

1. A method for providing an image of a tissue sample, wherein individual elements of said image have intensities corresponding to the rate of flow of very slowly flowing liquids therein, comprising the steps of:

(a) placing said sample in a first magnetic field;

(b) establishing a first magnetic field gradient through said sample;

(c) applying a series of repetitive radio-fre¬ quency pulses of electromagnetic energy to said sample at intervals "^i, said radio frequency energy, said gradient and said interval \ being chosen to create a spatial periodicity of magnetization of nuclei in said sample;

(d) detecting the signal emitted by said nuclei in the intervals ^*f. between application of said radio-frequency pulses;

(e) recording the signal thus detected;

(f) iteratively varying the gradient applied and repeating said steps (c), (d), and (e); (g) after said gradient has been varied through a predetermined range, generating an image using the signal values recorded with respect to each value of the gradient;

(h) changing the interval to a new value T and repeating said steps (b), (c), (d) , (e), (f) and (g) to generate a second image; and

(i) subtracting said second image from said first image to obtain a difference image, in which the image

SUBSTITUTE SHEET elements are each determined solely by the nuclear magnetic resonance of nuclei 'in slowly flowing fluids in said sample.

2. The method of claim 1 wherein said gradient is reversed for a portion of each interpulse period "T^ and^*C₂*

3. The method of claim 1 wherein said detected signals are functions of time, and are converted to functions of frequency prior to formation of said images.

4. The method of claim 1 wherein the interpulse intervals U.,^₂ are significantly less than either of the relaxation times T^, τ₂ of any of the nuclei the flow velocities of which are imaged in said process.

5. In a method for steady state free precession nuclear magnetic resonant analysis of a sample, com¬ prising the step of applying a regular sequence of radio-frequency pulses to a sample, said pulses being separated in time by an interval £., to cause^" nuclei in said sample to assume a state of driven equilibrium, said sample being disposed in a magnetic field having a gradient whereby spatial periodicity of the nuclear magnetization is established, the improvement which comprises reversing the gradient for a period of time less than half of the interpulse interval ■ , whereby the signal detected reaches a maximum value at an additional point other than at the time of application of the radio-frequency pulses, and detecting the

ISA/US

SUBSTITUTE SHEET signal emitted by said nuclei during the interpulse interval -

6. The improvement of claim 5, comprising the steps of generating said improvement is used in a method for providing a nuclear magnetic resonant image of a sample, each element of said image corresponding to a value for flow of liquids at very low flow rates, comprising the steps of generating plural images of the same sample using different time constants f^T_jr and subtracting one image from the other, whereby the result is a signal proportional only to density of nuclei in said slowly moving fluid sample.

7. Apparatus for generating nuclear magnetic resonant images of a slice through a body, each element of said image being proportional to the density of slowly flowing fluids in said slice, comprising: means for applying a first magnetic field to said sample; means for applying a variable gradient field to said sample; means for applying a repetitive series, of radio frequency pulses to said sample, to cause nuclei in said sample to establish a driven equilibrium state of spatially periodic magnetization; means for detecting a signal emitted by said nuclei in the interpulse interval caused by their dephasing; means for varying said gradient and for producing images corresponding to the various gradients for a

ISAIUS

SUBSTITUTE SHEET first interpulse interval "C , and for a second interpulse intervalf -; and means for subtracting the image generated with respect to~f^ from that generated with respect to ₂, whereby an image is generated having elements the value of which are proportional substantially only to the density of nuclei of slowly flowing liquids in said sample.

8. The apparatus of claim 7 further comprising means for varying said gradient field during the interpulse intervals ^*£, and ₂, whereby an additional peak of the detected signal is formed, displaced in time from the radio frequency pulses.

9. The apparatus of claim 7 further comprising means for taking the Fourier transform of said signals received prior to assembling them into an image.

10. In apparatus for nuclear magnetic resonance exper¬ iments, comprising first means for applying a magnetic field to a sample, means for applying a magnetic field of variable gradient to said sample, means for detect¬ ing a signal emitted by nuclei in said sample, means for imparting a regular sequence of radio frequency pulses of energy to said sample, and means for con¬ trolling the gradient of the field, the improvement comprising said means for controlling the gradient of the field comprising means for varying said gradient of the field during said interpulse interval so as to enable detection of a peak in a detected signal varying with dephasing of said nuclei.

ISA/US

SUBSTITUTE SHEET

11. The improvement of claim 10, as applied to a system in which a first set <of experiments are per- formed with varying gradients using a first interpulse interval^•* . and a second set of experiments are subse¬ quently performed using the same variety of gradients in a second interpulse interval ^*T₂, and further com¬ prising means for comparing an image generated in the first set of experiments to an image generated in the second set of experiments, to generate an image each element of which is substantially proportional to the density of slowly flowing nuclei at specific physical positions within said sample.

12. The system of claim 10, wherein said detected signals are functions of time., and further comprising means for Fourier-transforming the detected signals to functions of frequency.

ISA

SUBSTITUTE SHEET