Method and Apparatus for the Formation of a Plurality of NMR Images, Each Having a Different Characteristic, of the Same Slice, in a Single Sequence

This application is submitted with a microfiche appendix, having Exhibit B containing copyrighted material, Copyright 1991, Toshiba American (MRI), Inc. The appendix consists of one (1) microfiche with 34 frames. The copyright owner has no objection to the facsimile reproduction by any one of the patent document or the patent disclosure, as it appears in the Patent and Trademark Office patent file or records, but otherwise reserves all copyright rights whatsoever in th appendix.

The present invention relates to a method and apparatus for the formation of a plurality of NMR images produced from a single sequence. Each of the NMR images has a different characteristic of the same selected nuclei.

The formation of NMR images based upon excited nuclei is well known in the art. In general, for a spin echo image, the MRI signal received from a selected set of nuclei is governed approximately by the following relationship:

where

I is the signal strength

.rho.(h) is the proton density of the selected nuclei,

T.sub.1 and T.sub.2 are respectively, the longitudinal and transverse relaxation times of the selected nuclei,

T.sub.R is the repetition time; and

T.sub.E is the echo time.

The parameters T.sub.R and T.sub.E are operator adjustable parameters, while T.sub.1, T.sub.2 and .rho.(h) are magnetic parameters intrinsic to each tissue.

In any NMR imaging process, the investigation seeks to determine the characteristics of the tissue under examination. For an image which is T.sub.1 weighted, it means that the image has more contribution from the T.sub.1 relaxation time than from the T.sub.2 relaxation time or the proton density. Similarly, for a T.sub.2 weighted image, it means that the image has more contribution from T.sub.2 relaxation time than from T.sub.1 relaxation time or the proton density.

In order to produce a T.sub.2 weighted image, the parameters T.sub.R and T.sub.E must be adjusted such that T.sub.R is relatively long and T.sub.E is relatively long. For a T.sub.1 weighted image, T.sub.R must be relatively short and T.sub.E must be relatively short. For a proton density weighted image, the variable T.sub.R must be relatively long and T.sub.E must be relatively short.

Finally, for clinical diagnosis, it is often desirable to obtain different weighted images simultaneously, of the same tissue or selected nuclei under examination.

In the prior art, to accomplish the formation of a plurality of images having different weighted characteristics of the same selected nuclei, a plurality of separated sequences of the MRI excitation and image re-construction is performed. For example a first sequence may be based upon T.sub.R at approximately 3,000 milliseconds and T.sub.E1 at 30 milliseconds, and T.sub.E2 at 80 milliseconds. A second sequence may have T.sub.R set at approximately 500 milliseconds and T.sub.E1 at 30 milliseconds.

A sequence contains a plurality of contiguous T.sub.R time periods. In the case of a multi-slice/multi-echo sequence, each T.sub.R time period is further divided into a plurality of slice time intervals, during each of which a different physical slice is excited and echo signals are acquired. Prior to the start of each sequence, the MRI instrument must be prepared for the acquisition of the echo signals during that sequence. This is termed the overhead of signal acquisition. Examples of such overhead include: opening and closing of files, checking for sufficient disk space, creating and loading the actual sequence, and adjusting the acquisition hardware such as RF level, receiver gain, and sampling rate. Since MRI instruments are relatively expensive and it is highly desirable to increase the throughput of patient imaging through an MRI instrument, it is desired to cut down the amount of time it takes to generate different images of the same selected nuclei.

In the prior art, it is also known to generate an image of a selected nuclei wherein the repetition time is varied within the single sequence, in order to reduce acquisition time. Thus, see for example, "T.sub.2 Weighted Spin Echo Pulse Sequence With Variable Repetition and Echo Times for Reduction of MR Image Acquisition Time", by R. Kim Butts, et al., RSNA, 1991. See also "Intra-Acquisition Modification of Pulse Sequence Parameters" by F. Farzaneh et al., SMRM, 1989.

Accordingly, in the present invention, a method of NMR imaging in a single sequence for the formation of a plurality of images of the same selected nuclei is disclosed. Each of the images of the same selected nuclei has a characteristic different from the other. The method comprises the steps of eliciting and receiving a first plurality of MRI signals from a plurality of selected nuclei during a first time interval. The first time interval comprises a plurality of first intervals T.sub.R1 with each first interval T.sub.R1 in the first time interval being substantially the same. The selected nuclei have their phase encoded by a different magnetic phase encoding gradient during each of the first intervals T.sub.R1. The method continues to elicit and to receive a second plurality of MRI signals from the same plurality of selected nuclei during a second time interval. The second time interval also comprises a plurality of second intervals T.sub.R2. However, each of the second interval T.sub.R2 is different from the first intervals, T.sub.R1. During the second time interval, the selected nuclei are phase encoded by a different magnetic phase gradient in each of the second intervals T.sub.R2. The magnetic phase encoding gradients used during the second intervals are the same as or are a subset of those during the first intervals. Finally, the acquired MRI signals are regrouped and processed to generate a plurality of images of the selected nuclei with each image being of the same plurality of selected nuclei but with different characteristics. During the process, portions of the data are used repeatedly.

FIG. 1 is a schematic block diagram of an MRI instrument with the present invention.

FIG. 2 is a schematic timing diagram of an MR imaging sequence representing the various parameters that are found in a typical MR pulse sequence of the prior art.

FIGS. 3a is a schematic timing diagram of two examples of MR imaging sequences of the prior art, and the echo signals collected therefrom represented as a plurality of K space diagrams (FIG. 3b), prior to each K space data set being fourier transformed to an image of a slice.

FIGS. 4a is a schematic timing diagram of a single MR imaging sequence of the prior art, and a K space diagram of the echo signals of the ith slice collected therefrom (FIG. 4b), prior to transformation into an image of the ith slice.

FIG. 5 is a K space diagram of the data collected from one slice, of one embodiment of the MR imaging sequence of the present invention, prior to transformation into an image of the slice.

FIG. 6a is a K space diagram of the data collected from one slice, of another embodiment of the MR imaging sequence of the present invention, and the regrouping of the collected data (FIG. 6b), prior to transformation into an image of the slice; FIG. 6c is a refinement of the method shown in FIG. 6a, with the regrouped data shown in FIG. 6d.

FIG. 7a is a K space diagram of the data collected from one slice, of yet another embodiment of the MR imaging sequence of the present invention, and the regrouping of the collected data (FIG. 7b) prior to transformation into an image of the slice. FIG. 7c is a refinement of the method of FIG. 7a.

FIG. 8a is a timing diagram showing forward and backward read-out truncation. FIG. 8b is a K space representation of the data for a slice in which read-out truncation is applied.

FIG. 9a is a K space diagram of the data collected from one slice, of still yet another embodiment of the MR imaging sequence of the present invention, and the regrouping of the collected data (FIG. 9b) prior to transformation into an image of the slice. FIG. 9c is a refinement of the method of FIG. 9a.

Referring to FIG. 1 there is shown an MRI instrument 10 suitable for use with the present invention. As is well known, the instrument 10 comprises a magnet 12, such as a superconductive magnet. The magnet 12 is typically cylindrically shaped, in which a patient is placed. The magnet 12 generates a static magnetic field, e.g. along the longitudinal axis of the patient.

The instrument 10 also comprises a plurality of magnetic gradient coils 11. The magnetic gradient coils 11 are driven by a plurality of magnetic gradient coil drivers 14, which supply the electric current to the magnetic gradient coils 11 to generate the necessary magnetic gradient fields. To select a particular "slice" or a group of selected nuclei, then to receive the MRI signal therefrom, the instrument 10 also comprises an RF transmitter and receiver 16. The RF transmitter and receiver 16 and the magnetic gradient coil drivers 14 are under the control of a control computer 18. The control computer 18 communicates with a host computer 22 which contains the programmed variable control. The operator interfaces with the keyboard/display terminal 20, which is connected to the host computer 22.

The timing of the energization of the gradient coils 11, and the RF transmitter and receiver 16 is controlled by a sequence, which is contained in the programmed variable control 22. A protocol file, shown in Exhibit A together with the sequence shown in Exhibit B generate the necessary controls for the host computer 22 to perform the present invention. Exhibit B is in the nature of a microfiche index, and is used by the computer 18 to control the timing of the energization of the gradient coils 11, and of the RF transmitter and receiver 16.

In the preferred embodiment, the MRI instrument 10 is a Toshiba MRT-35, containing a micro VAX computer made by the Digital Equipment Corporation.

The Toshiba MRT-35 MR instrument 10 can perform the present invention which is in the nature of NMR imaging in a single sequence. Each sequence typically comprises a plurality of T.sub.R intervals. A single T.sub.R interval (out of, for example, 256 PE cycles per slice volume in a typical single MRI sequence) is shown in FIG. 2. Each PE cycle is sometimes referred to as a "subsequence" of the overall sequence.

As shown in FIG. 2, a T.sub.R interval comprises a plurality of N slices, or subintervals. At the beginning of each cycle, an RF nutation pulse of .theta..degree. (e.g., 90.degree.) is applied in the presence of a slice select (SS) magnetic gradient pulse (e.g. G.sub.2) so as to selectively address a given slice volume of hydrogen nuclei. In that slice subinterval, a phase encoding (or PE) magnetic gradient pulse (e.g. G.sub.y) is imposed (e.g., having a magnitude/duration). In that same slice subinterval, a slice selective 180.degree. RF refocusing pulse is applied to produce a first spin echo signal. The spin echo signal is received during the application of the read out (or RO) magnetic gradient (e.g. G.sub.x). The first spin echo signal occurs at time T.sub.E1 from the center of the RF nutation pulse (i.e. 90.degree.).

An A/D converter (not shown) is activated prior to T.sub.E1 and is turned off shortly after T.sub.E1, from the center of the nutation pulse. Thus, the RF signal which is centered about T.sub.E1 is sampled. The digital data is collected and stored in a file containing other first echo signals from that same slice, and with different PE gradient.

If a second spin echo signal is desired to be received, in that same cycle of the same slice, another 180.degree. refocusing pulse is applied. A second read out gradient pulse is applied at or about the time T.sub.E2 from the center of the nutation pulse. The A/D converter is activated prior to T.sub.E2 and is turned off shortly after T.sub.E2, from the center of the nutation pulse. Thus, the RF signal which is centered about T.sub.E2 is sampled and is digitized. The digital data is collected and is stored in a file containing other second echo signals from that same slice.

After the collection of the second spin echo signal, if any, the time interval for the slice subinterval 1 includes a brief waiting period wait.sub.1. During the entirety of the slice subinterval 1, the phase encoding gradient (PE) is set at a first value. As shown in FIG. 2 the phase encoding is set at -128.

During the second slice subinterval of the subsequence T.sub.R, the frequency of the RF pulse is varied by a delta amount in order that a different physical slice of selected nuclei is excited and the signals therefrom are read out. The phase encoding gradient, however, is set the same as are the parameters T.sub.E1 and T.sub.E2. This continues for N slice subintervals. The sum of all slice subintervals, including the waiting period associated with each slice subinterval equals the desired time duration of T.sub.R. Of course, within each subsequence T.sub.R, the slice subintervals can proceed from one to another without a waiting period, with one waiting period at the end of the last slice subinterval N.

After the N slice subintervals, the sequence continues with the next subsequence T.sub.R. In the next subsequence T.sub.R, the above steps are repeated with the phase encoding gradient changed to a different value, e.g. -127.

After all the spin echo signals are collected, the stored spin echo signals for a particular slice, but with different phase encoding gradients thereto, are processed, by for example, using 2DFT or two dimensional fourier transform, to generate an image of the slice. Such processing is well known in the art.

The sequence shown hereinabove is termed a multislice/multiecho sequence wherein a plurality of slices with a plurality of echoes in each slice is read out. Thus, as can be seen from FIG. 2, in a single sequence of the prior art, N.times.M images (N=number of slices; M=number of echoes in each slice) with each image representing a slice of different selected nuclei and a T.sub.E value can be produced.

As previously stated, however, it is highly desirable in certain aspects to form a plurality of images having different weighted characteristics (T.sub.1, T.sub.2 or .rho.(h)) of the same selected nuclei. Referring to FIG. 3a, there is shown a time sequence of the prior art.

In the prior art, this was accomplished by a first sequence wherein, for example, the T.sub.1 weighted characteristics of the N slices of images are produced in a first sequence. Thereafter, after a period of delay due to acquisition overhead, as discussed heretofore, a second sequence of NMR imaging is produced. In the second sequence, the appropriate variables such as T.sub.2 and T.sub.R are adjusted in order to produce a plurality of N slice images having different weighted characteristics.

Referring to FIG. 3b there is shown a well known depiction of a plurality of files of data, with file representing data from each slice, obtained from the sequence shown in FIG. 3a. The data is represented as a K space representation. In FIG. 3b, for sequence #1, there are N slices; hence there are N K space representations, with a K space representation for each slice. The data contained in each of the K space representations is based upon a value of T.sub.R (small, on the order of 500 milliseconds) and a single value of T.sub.E (also small, on the order of 30 milliseconds). For sequence #2, there are two K space representations for each slice. The K space representations are both based upon a value of T.sub.R (large, on the order of 3,000 millisecond). The K space representations for each slice differ as they are based upon different T.sub.E values.

Referring to FIG. 4a there is shown another prior art sequence. In this sequence, a single sequence is shown wherein the operator adjustable variables T.sub.R is varied within the single sequence. During a portion of the sequence, the echo signals are received with T.sub.R =T.sub.R1. In another portion of the sequence, the echo signals are received with T.sub.R different from T.sub.R1, with T.sub.R making a gradual transition from T.sub.R1 to T.sub.R2, and then back to T.sub.R1. Thus, the repetition time T.sub.R and the echo time T.sub.E can be both varied in the same single sequence. This sequence is discussed hereinabove in the Background, by Butts et al. See, also, U.S. patent application Kaufman et al., entitled "MRI Using Variable Imaging Parameter(s) Within a Single Image Sequence", assigned to The Regents of the University of California (Attorney's Reference No. 89-171). Such a sequence generates a single image of a group of selected nuclei for each echo.

Referring to FIG. 5 there is shown a K space representation of the data for a slice of selected nuclei acquired in one method of the single sequence of the present invention. The single sequence comprises two time intervals (A and B). In the embodiment shown in FIG. 5, the first time interval (A) is comprised of a plurality of subsequences, e.g. 256, with PE ranging in value from -128 to +127 gradient unit, with each subsequence being T.sub.R1 in duration, and with only one echo signal T.sub.E1 being received.

The second time interval (B) excites and receives a second plurality of MRI signals from the excitation of the same plurality of selected nuclei as during the first time interval (A). During this second time interval, again, the phase encoding gradient would be cycled through 256 possible values from -128 to +127. For each of the 256 subsequences, the variable T.sub.R would be different, i.e. T.sub.R2 and with two echo signals T.sub.E1 and T.sub.E2 obtained in each subsequence.

After the MRI signals from the first interval and from the second interval are collected, they are separated and processed by conventional processing means such as 2DFT to generate a plurality of images of the same slice but with each image being of different weighted characteristics.

Thus, in the example shown in FIG. 5 there are three possible weighted images for each slice of selected nuclei. The images are: ##EQU1##

As can be seen from the sequence shown in FIG. 5, with the single sequence of the present invention, the "set up time" between sequences is eliminated. This speeds up the acquisition of different weighted MR images of the same selected nuclei.

Referring to FIG. 6a, there is shown another K space representation of the data for a slice of selected nuclei acquired in another embodiment of the method of the present invention. In the embodiment shown in FIG. 6a, the sequence again comprises two contiguous time intervals (A and B). In the time interval (A), the phase encoding magnetic gradient is cycled through the entire range from -128 to +127, thereby generating 256 subsequences. Each of the subsequences is T.sub.R1 in duration. However, two echo signals T.sub.E1 and T.sub.E2 are acquired in each of the subsequences.

During the second time interval (B), however, the phase encoding magnetic gradient is cycled only through values -.alpha. to +.alpha.. These values represent a subset of the phase encoding values during the first time interval, and center about the value of phase encoding equal 0. Thus, only 2.alpha. subsequences of MRI signal acquisition is performed in the second time interval (B). During the 2.alpha. subsequences, T.sub.R is gradually changed from T.sub.R1 to T.sub.R2 and then gradually back to T.sub.R1, as shown in FIG. 6a.

Once the data for each slice is collected, they are regrouped. Data is regrouped in accordance with that shown in FIG. 6b. Three files of new, regrouped K space data are formed. The T.sub.E1 echo signals for each slice, for the first time interval, i.e. acquired when T.sub.R =T.sub.R1, is divided into three parts:

a.sub.1 for the signals when PE=.alpha.+1 to PE=+127;

b.sub.1 for the signals when PE=-.alpha. to PE=+.alpha.;

c.sub.1 for the signals when PE=-128 to PE=-.alpha.-1.

Similarly the echo signals for the T.sub.E2 signals are divided into three parts: a.sub.2, b.sub.2, and c.sub.2. The echo signals acquired during the second time interval (B) are B.sub.1, and B.sub.2 signals for echo signals T.sub.E1 and T.sub.E2 respectively.

The regrouping produces three files of data:

with a.sub.1 and c.sub.1 used repeatedly (twice). Exhibit C contains software routines to accomplish the regrouping of the data. These three data files are then reconstructed using fourier transformation to produce three images of the particular slice, having weighted characteristics.

In addition, the following data files, when fourier transformed to images, may provide clinically significant images:

Referring to FIG. 6c, there is shown a variation of the embodiment of the method of the present invention to increases the number of slices. During the transition from T.sub.R1 to T.sub.R2 either

a) two echoes can be acquired with T.sub.E varying with T.sub.R ; or

b) only the first echo is acquired and the data is shared by both echoes during reconstruction.

In this method, the second echo signals T.sub.E2 acquired during T.sub.R1, i.e. a.sub.2, b.sub.2, and c.sub.2 are eliminated. Thus, during the first time interval (A) when T.sub.R =T.sub.R1, only one echo signal is acquired during each subsequence. The regrouping of the three data files, when fourier transformed to produce the three different weighted images, would be as follows:

with a.sub.1 and c.sub.1 used repeatedly (thrice). When these data files are fourier transformed, the images would again be representative of the weighted images described heretofore. Image transformed from the data file, a.sub.1 +B.sub.2 +c.sub.1, is similar to the image transformed from the data file a.sub.2 +B.sub.2 +c.sub.2, because the phase encoding signals, near PE=0 contains the most of the contrast data.

Since it is well known that MRI signals received can be used to calculate the conjugate of the MRI signals, another embodiment of the present invention utilizes this in order to reduce yet further the time required to acquire the MRI signals in order to produce the plurality of images per slice, or to increase the number of slices in the same acquisition time period.

Referring to FIG. 7a there is shown another embodiment of the single sequence of the present invention. Again, similar to the foregoing discussion, the single sequence is divided into two contiguous time intervals A and B. During the first interval (A), however, the MRI signals are not acquired by cycling the phase encoded magnetic gradient through the entire range of values. Instead, during the first interval, only .alpha.+128 subsequences are generated. Each of the subsequence, again, is of T.sub.R1 in duration, with two echo signals acquired during each subsequence. During the second time interval (B), however, the MRI signals are again cycled through the phase encoding range -.alpha. to +.alpha.. In this time interval (B), 2.alpha. subsequences are formed, with T.sub.R different from T.sub.R1, with T.sub.R making a gradual transition from T.sub.R1 to T.sub.R2 and then back to T.sub.R1. Again during this time interval B, two read out echo signals per subsequence are acquired.

In the regrouping of the data, since the MRI data between PE=-128 and PE=-.alpha.-1 can be recovered by complex conjugate of the data from +.alpha.+1 to +128, the following data files are formed:

where is the conjugated data. The fourier transformation of these data files would produce the weighted images.

In addition, similar to the embodiment shown in FIG. 6, the following alternative data files, when fourier transformed would produce images which may have clinical significance:

Similar to the embodiment shown and described in FIGS. 6c and 6d, the second echo signal T.sub.E2 during the first time interval A, i.e. during subsequences having T.sub.R1 in duration, can be skipped to accommodate more slices. The resultant data files are shown in FIG. 7(c).

In the regrouping of the data files from the data acquired in the method shown in FIG. 7(c), the following data files are regrouped:

here echo data are mixed, i.e. data from the first echo, a.sub.1, are grouped in a file with echo data from the second echo, B.sub.2.

It is also well known that the echo signal may be truncated on one side in the read-out direction, with the truncated data, either being restored by using conjugated data, as described heretofore, or simply zeroed for fourier transformation purpose. Referring to FIG. 8(a) there is shown a timing diagram of the acquisition of one echo signal, during one subsequence, with truncated read-out gradient, and truncated sampling by an analog-to-digital (A to D) converter (not shown). Applying this truncation throughout the sequence would result in a K space representation of one slice, as shown in FIG. 8(b), with missing data (TD). The missing data (TD) can be calculated by the complex conjugate of the acquired data, as shown, or it can be simply left as zero fill, as the fourier transform thereof would not significantly affect th image contrast.

Referring to FIG. 9a there is shown another embodiment of the single sequence of the present invention, employing data in both the readout and phase encoding direction. Again, similar to the foregoing discussion, the single sequence is divided into two contiguous time intervals A and B. During the first interval (A), similar to that shown in FIG. 7(a), .alpha.+128 subsequences are generated. Each of the subsequence, again, is of T.sub.R1 in duration, with two echo signals acquired during each subsequence. In addition, each of the echo signals is truncated forward and backward, respectively.

During the second time interval (B), the MRI signals are cycled through the phase encoding range -.alpha. to +.alpha.. In this time interval (B), 2.alpha. subsequences are formed, with the subsequence duration T.sub.R being gradually changed from T.sub.R1 to T.sub.R2 and then back to T.sub.R1. Again during this time interval B, two echo signals per subsequence are acquired, with read-out gradient truncation for each of the echo signals.

In the regrouping of the data, since the MRI signals from -128 to -.alpha.-1 phase encoded magnetic gradient is the conjugate of those from +.alpha.+1 to +128, the following data files are formed:

where * is the conjugated data. In addition, the truncated read-out signal for each of the echoes, may be zero filled or may be filled with conjugated data. The fourier transformation of these data files would produce the weighted images.

In addition, the following alternative data files, when fourier transformed would produce images which may have clinical significance:

Similar to the embodiment shown and described in FIGS. 7c and 7d, by eliminating the second echo signal T.sub.E2 during the first time interval A, i.e. during subsequences having T.sub.R1 in duration, more slices can be accommodated. The resultant data files are shown in FIG. 9(c).

In the regrouping of the data files from the data acquired in the method shown in FIG. 9(c), the following data files are regrouped:

again with the truncated read-out signals either zero filled, or filled by conjugated data. An advantage of the method of FIG. 9 is that the number of slices can be increased, because T.sub.R per slice is reduced with forward and backward truncation of the first and second echoes.

From the foregoing, it can be seen that a new improved method and apparatus of the present invention is disclosed wherein in a single sequence, a plurality of images for the same group of selected nuclei with each of the images having different contrast characteristics of the group of selected nuclei is disclosed. In several embodiments, the data is regrouped and portions are reused to form multiple data files prior to fourier transformation of each data file to an image. The method and apparatus permits more rapid acquisition of MRI signals and more efficient use of the acquired data, and permits greater patient throughput in MRI instruments. In all of the prior art, each magnetic phase encoding gradient value is used only once in a single sequence. In the present invention, common to all the embodiments, at least a portion of the magnetic phase encoding gradient values are repeatedly used in the same sequence. ##SPC1##