Method and control device for operating a magnetic resonance system

Abstract

In a magnetic resonance imaging procedure, multiple slices are initially spatially selectively excited in a first time interval by respective RF pulses followed by at least one RF refocusing pulse that causes one echo signal from each slice, with a time interval of two consecutive echo signals equal to the first time interval. A second RF refocusing pulse is emitted at a second time interval from the last echo signal that causes, one further echo signal per slice, with the time interval of two consecutive echo signals equal to the first time interval. At least one further RF refocusing pulse is emitted in a third time interval following the preceding RF refocusing pulse producing multiple temporally separated echo signals per refocusing pulse. The third time interval is selected so that the number of echo signals per RF refocusing pulse is twice the number of excited slices.

Claims

1. A method for acquiring magnetic resonance (MR) data from an examination subject, comprising: operating an MR data acquisition unit, in which a subject is situated, according to a pulse sequence module in which a plurality of spatially selective radio-frequency (RF) slice excitation pulses are radiated that respectively excite nuclear spins in a plurality of respective slices in the examination subject, whereby the time between two consecutive of these RF slice excitation pulses defines a first time interval; operating said MR data acquisition unit in said sequence module to execute a preparation block following a last of said plurality of RF slice excitation pulses and, in said preparation block, radiating at least one RF refocusing pulse that causes each of the excited slices to produce one respective echo signal, with a time interval of two consecutive ones of said echo signals equaling said first time interval; operating said MR data acquisition unit to radiate a second RF refocusing pulse at a time interval after a last of said echo signals produced by said preparation block, and configuring said second RF refocusing pulse to cause each of the excited slices to produce one further echo signal, with a time interval between two consecutive ones of said further echo signals equaling said first time interval; operating said MR data acquisition unit to radiate at least one further RF refocusing pulse at a third time interval after a respective preceding RF refocusing pulse, and configuring said at least one further RF refocusing pulse to cause the excited slices to produce a plurality of temporally separated echo signals for each further RF refocusing pulse, and setting a duration of said third time interval to cause the plurality of echo signals caused by each further RF refocusing pulse to be twice as high as said plurality of excited slices; and operating said MR data acquisition unit to readout raw data representing each of said echo signals and entering said raw data into an electronic memory organized as k-space, thereby producing an electronic memory filled with data in a form allowing reconstruction of images of said plurality of slices of the subject.

2. A method as claimed in claim 1 comprising operating said MR data acquisition unit to activate gradient pulses in slice selection direction between consecutive RF slice excitation pulses such that the total accumulated moment in slice selection direction between two consecutive ones of said RF slice excitation pulses is zero.

3. A method as claimed in claim 1 comprising reading out said plurality of echo signals for each RF refocusing pulse by operating said MR data acquisition unit to activate a readout gradient pulse during the reading out of each of said echo signals and an additional gradient pulse in a readout direction between two consecutive ones of said RF slice excitation pulses, with the total gradient moment of said additional readout gradient pulse having an absolute value equal to a gradient moment accumulated between two consecutive ones of said echo signals by their respective readout gradient pulses.

4. A method as claimed in claim 3, comprising: operating said MR data acquisition unit to activate another gradient pulse in said readout direction between said last of said RF slice excitation pulses and said preparation block, such that said another gradient pulse exactly compensates the gradient moment in readout direction that is accumulated by nuclear spins in the slice that was excited by said last of said RF slice excitation pulses, between a start of the preparation block and the first echo signal produced after said preparation block by said last of said RF slice excitation pulses and said preparation block; and operating said MR data acquisition unit to activate another gradient pulse in said readout direction between a last echo signal produced by said preparation block, said another gradient pulse exactly compensating the moment in the readout direction that is accumulated by nuclear spins excited in a slice by a first of said RF slice excitation pulses, between said second RF refocusing pulse and the first echo signal of said slice excited by said first of said RF slice excitation pulses produced after said second RF refocusing pulse.

5. A method as claimed in claim 1 comprising operating said MR data acquisition unit to radiate each of said RF slice excitation pulses for a duration that is shorter than a duration at which any of said RF refocusing pulses is radiated.

6. A method as claimed in claim 1 comprising operating said MR data acquisition unit with said preparation block configured to attenuate transverse magnetization of said nuclear spins in a region of the subject from which the echo signals in said preparation block originate, dependent on diffusion properties of said tissue.

7. A method as claimed in claim 1 comprising entering said raw data into said electronic memory organized as k-space by, for each of said slices, acquiring said raw data twice with a single echo train in said sequence module and entering the twice-acquired raw data into a respective k-space region of said electronic memory for the respective slice.

8. A method as claimed in claim 1 comprising operating said MR data acquisition unit to acquire said raw data in a pulse sequence comprising a plurality of sequence modules, each identical to said sequence module and entering raw data into two segments of said k-space for each slice acquired with an echo train in each of said sequence modules.

9. A method as claimed in claim 8 comprising entering said raw data into k-space by segmenting k-space according to a PROPELLER trajectory.

10. A method for producing magnetic resonance image data of an examination subject, comprising: operating an MR data acquisition unit, in which a subject is situated, according to a pulse sequence module in which a plurality of spatially selective radio-frequency (RF) slice excitation pulses are radiated that respectively excite nuclear spins in a plurality of respective slices in the examination subject, whereby the time between two consecutive of these RF slice excitation pulses defines a first time interval; operating said MR data acquisition unit in said sequence module to execute a preparation block following a last of said plurality of RF slice excitation pulses and, in said preparation block, radiating at least one RF refocusing pulse that causes each of the excited slices to produce one respective echo signal, with a time interval between two consecutive ones of said echo signals equaling said first time interval; operating said MR data acquisition unit to radiate a second RF refocusing pulse at a time interval after a last of said echo signals produced by said preparation block, and configuring said second RF refocusing pulse to cause each of the excited layers to produce one further echo signal, with a time interval between two consecutive ones of said further echo signals equaling said first time interval; operating said MR data acquisition unit to radiate at least one further RF refocusing pulse at a third time interval after a respective preceding RF refocusing pulse, and configuring said at least one further RF refocusing pulse to cause the excited slices to produce a plurality of temporally separated echo signals for each further RF refocusing pulse, and setting a duration of said third time interval to cause the plurality of echo signals caused by each further RF refocusing pulse to be twice as high as said plurality of excited slices; operating said MR data acquisition unit to readout raw data during a scanning window around each of said echo signals and entering said raw data into an electronic memory organized as k-space, thereby producing an electronic memory filled with data in a form allowing reconstruction of images of said plurality of slices of the subject; from a processor, accessing said raw data in said electronic memory and operating on said raw data in said processor according to an image reconstruction algorithm to calculate a plurality of separate magnitude images from the raw data acquired during each of said scanning windows; and in said processor, combining the respective magnitude images from each slice to form a single MR image for that slice, and making said magnitude image for each slice available at an output of said processor in electronic form.

11. A method as claimed in claim 10 comprising combining said magnitude images by executing a sum of squares algorithm in said processor.

12. A method for producing magnetic resonance image data of an examination subject, comprising: operating an MR data acquisition unit, in which a subject is situated, according to a pulse sequence module in which a plurality of spatially selective radio-frequency (RF) slice excitation pulses are radiated that respectively excite nuclear spins in a plurality of respective slices in the examination subject, whereby the time between two consecutive of these RF slice excitation pulses defines a first time interval; operating said MR data acquisition unit in said sequence module to execute a preparation block following a last of said plurality of RF slice excitation pulses and, in said preparation block, radiating at least one RF refocusing pulse that causes each of the excited slices to produce one respective echo signal, with a time interval between two consecutive ones of said echo signals equaling said first time interval; operating said MR data acquisition unit to radiate a second RF refocusing pulse at a time interval after a last of said echo signals produced by said preparation block, and configuring said second RF refocusing pulse to cause each of the excited layers to produce one further echo signal, with a time interval between two consecutive ones of said further echo signals equaling said first time interval; operating said MR data acquisition unit to radiate at least one further RF refocusing pulse at a third time interval after a respective preceding RF refocusing pulse, and configuring said at least one further RF refocusing pulse to cause the excited slices to produce a plurality of temporally separated echo signals for each further RF refocusing pulse, and setting a duration of said third time interval to cause the plurality of echo signals caused by each further RF refocusing pulse to be twice as high as said plurality of excited slices; operating said MR data acquisition unit to readout raw data during a scanning window around each of said echo signals and entering said raw data into an electronic memory organized as k-space, thereby producing an electronic memory filled with data in a form allowing reconstruction of images of said plurality of slices of the subject; and from a processor, accessing said raw data from said electronic memory and generating an MR image therefrom by executing an algorithm in said processor in which image data for a respective slice are formed as a complex-valued combination of raw data acquired for that slice in different ones of said scanning windows.

13. A method as claimed in claim 12 comprising, before implementing said complex-valued combination, computationally removing any spatially slowly changing phase in the image domain from said image data.

14. A computerized control device for operating a magnetic resonance (MR) data acquisition unit in which an examination subject is situated, said control device being configured to: operate said MR data acquisition unit, in which a subject is situated, according to a pulse sequence module in which a plurality of spatially selective radio-frequency (RF) slice excitation pulses are radiated that respectively excite nuclear spins in a plurality of respective slices in the examination subject, whereby the time between two consecutive of these RF slice excitation pulses defines a first time interval; operate said MR data acquisition unit in said sequence module to execute a preparation block following a last of said plurality of RF slice excitation pulses and, in said preparation block, radiating at least one RF refocusing pulse that causes each of the excited slices to produce one respective echo signal, with a time interval of two consecutive ones of said echo signals equaling said first time interval; operate said MR data acquisition unit to radiate a second RF refocusing pulse at a time interval after a last of said echo signals produced by said preparation block, and configure said second RF refocusing pulse to cause each of the excited layers to produce one further echo signal, with a time interval between two consecutive ones of said further echo signals equaling said first time interval; operate said MR data acquisition unit to radiate at least one further RF refocusing pulse at a third time interval after a respective preceding RF refocusing pulse, and configure said at least one further RF refocusing pulse to cause the excited slices to produce a plurality of temporally separated echo signals for each further RF refocusing pulse, and setting a duration of said third time interval to cause the plurality of echo signals caused by each further RF refocusing pulse to be twice as high as said plurality of excited slices; and operate said MR data acquisition unit to readout raw data during a scanning window around each of said echo signals and to enter said raw data into an electronic memory organized as k-space, thereby producing an electronic memory filled with data in a form allowing reconstruction of an images of said plurality of slices of the subject.

15. A magnetic resonance (MR) imaging apparatus comprising: an MR data acquisition unit comprising a basic field magnet that generates a basic field in which an examination subject in the MR data acquisition unit i situated, a radio-frequency (RF) transmission system, a gradient system, an RF reception system, and a control unit; said control unit being configured to operate said MR data acquisition unit, in which a subject is situated, according to a pulse sequence module in which a plurality of spatially selective RF slice excitation pulses are radiated by said RF transmission system that respectively excite nuclear spins in a plurality of respective slices in the examination subject, whereby the time between two consecutive of these RF slice excitation pulses defines a first time interval; said control unit being configured to operate said MR data acquisition unit in said sequence module to execute a preparation block following a last of said plurality of RF slice excitation pulses and, in said preparation block, radiating at least one RF refocusing pulse by said RF transmission system that causes each of the excited slices to produce one respective echo signal, with a time interval of two consecutive ones of said echo signals equaling said first time interval; said control unit being configured to operate said MR data acquisition unit to radiate a second RF refocusing pulse by said RF transmission system at a time interval after a last of said echo signals produced by said preparation block, with said second RF refocusing pulse being configured to cause each of the excited slices to produce one further echo signal, with a time interval between two consecutive ones of said further echo signals equaling said first time interval; said control unit being configured to operate said MR data acquisition unit to radiate at least one further RF refocusing pulse by said RF transmission system at a third time interval after a respective preceding RF refocusing pulse, and with said at least one further RF refocusing pulse being configured to cause the excited slices to produce a plurality of temporally separated echo signals for each further RF refocusing pulse, and to set a duration of said third time interval to cause the plurality of echo signals caused by each further RF refocusing pulse to be twice as high as said plurality of excited slices; and said control unit being configured to operate said gradient system of said MR data acquisition unit to execute a readout gradient during the readout of raw data to spatially encode said raw data, and to enter said raw data into an electronic memory organized as k-space, thereby producing an electronic memory filled with data in a form allowing reconstruction of images of said plurality of slices of the subject.

16. A non-transitory, computer-readable data storage medium encoded with programming instructions, said data storage medium being loaded into a control and evaluation system of a magnetic resonance (MR) apparatus that also comprises an MR data acquisition unit, said programming instructions causing said control and evaluation system to: operate said MR data acquisition unit, in which a subject is situated, according to a pulse sequence module in which a plurality of spatially selective radio-frequency (RF) slice excitation pulses are radiated that respectively excite nuclear spins in a plurality of respective slices in the examination subject, whereby the time between two consecutive of these RF slice excitation pulses defines a first time interval; operate said MR data acquisition unit in said sequence module to execute a preparation block following a last of said plurality of RF slice excitation pulses and, in said preparation block, radiating at least one RF refocusing pulse that causes each of the excited slices to produce one respective echo signal, with a time interval of two consecutive ones of said echo signals equaling said first time interval; operate said MR data acquisition unit to radiate a second RF refocusing pulse at a time interval after a last of said echo signals produced by said preparation block, with said second RF refocusing pulse being configured to cause each of the excited slices to produce one further echo signal, with a time interval between two consecutive ones of said further echo signals equaling said first time interval; operate said MR data acquisition unit to radiate at least one further RF refocusing pulse at a third time interval after a respective preceding RF refocusing pulse, with said at least one further RF refocusing pulse being configured to cause the excited slices to produce a plurality of temporally separated echo signals for each further RF refocusing pulse, and set a duration of said third time interval to cause the plurality of echo signals caused by each further RF refocusing pulse to be twice as high as said plurality of excited slices; and operate said MR data acquisition unit to readout raw data during a scanning window around each of said echo signals, and enter said raw data into an electronic memory organized as k-space, thereby producing an electronic memory filled with data in a form allowing reconstruction of images of said plurality of slices of the subject.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) FIG. 1 is a schematic representation of a magnetic resonance imaging system according to one embodiment of the invention.

(2) FIG. 2 shows an example of a pulse pattern for a sequence of a conventional single slice TSE pulse sequence in prior art.

(3) FIG. 3 shows the combination of FIGS. 3A and 3B.

(4) FIGS. 3A and 3B show a pulse pattern for a multi-slice TSE pulse sequence according to one embodiment of the invention having two simultaneously refocused slices.

(5) FIG. 4 shows a phase diagram for the second slice when performing a pulse sequence with a pulse pattern according to FIGS. 3A and 3B.

(6) FIG. 5 shows a phase diagram for the first slice when performing a pulse sequence with a pulse pattern according to FIGS. 3A and 3B.

(7) FIG. 6 is a flowchart of an example of a sequence of a method for a complex combination of the raw data acquired in the different scanning windows.

(8) FIG. 7 is a flowchart for a sequence of a method for combining the raw data acquired in the different scanning windows from a PROPELLER TSE sequence designed according to an inventive pulse pattern.

DESCRIPTION OF THE PREFERRED EMBODIMENTS

(9) FIG. 1 shows an approximate invention-based magnetic resonance imaging system 1 (subsequently also called MR system). On the one hand, it comprises the actual magnetic resonance scanner 2 with an examination room 3 or tunnel, into which it is possible to insert an examination object O, or in this case a patient or test person in whose body the examination object, for example a specific organ, is located.

(10) In conventional manner, the magnetic resonance scanner 2 is equipped with a basic field magnet system 4, a gradient system 6, as well as an RF antenna transmission system 5 and an RF antenna reception system 7. In the embodiment shown, the RF antenna transmission system 5 is a whole body coil permanently installed in the magnetic resonance scanner 2, while the RF antenna reception system 7 is composed of local coils that can be arranged on the patient or test person (symbolized in FIG. 1 only by a single local coil). Basically, it is also possible to use the whole body coil as RF antenna reception system and the local coils as RF antenna transmission system, provided these coils can be switched to different operational modes. Here the basic field magnet system 4 is designed in conventional manner in such a way that it produces a basic magnet field in longitudinal direction of the patient, i.e., along the longitudinal axis of the magnetic resonance scanner 2, which axis extends in z direction. In conventional manner, the gradient system 6 has individually controllable gradient coils in order to connect independently from one another gradients in x, y or z direction.

(11) The MR system shown in FIG. 1 involves a whole body system with a tunnel into which a patient can be completely inserted. Basically, it is also possible to use the invention on other MR systems, for example, with laterally open, C-shaped housings, particularly with smaller magnetic resonance scanners, in which, for example, only 1 body part can be positioned.

(12) Furthermore, the MR system 1 has a central control device 13 used for controlling the system. This central control system 13 has a sequence control unit 14 for performing measuring sequence control. It is used to control within a measuring session the sequence of radio-frequency pulses (RF pulses) and gradient pulses depending on a selected pulse sequence PS in a test person's volume range of interest. At the same time, several sequence modules can be used to build a pulse sequence. Each sequence module acquires data from one and/or several slices. For example, such a pulse sequence PS can be predetermined and parameterized within a measurement or control protocol P. Usually, different control protocols P are stored in a memory 19 for different measurements or measuring sessions and can be selected by an operator (and changed, if required) and then used for performing the measurement. In the case at hand, among other things, the control device 13 includes pulse sequences that operate in accordance with the invention-based method. Subsequently, by means if FIGS. 3A and 3B, an example of such a pulse sequence is described in more detail.

(13) To emit the individual RF pulses of a pulse sequence PS, the central control device 13 has a radio-frequency transmission system 15, which produces, amplifies and supplies via a suitable interface (not shown) the RF pulses to the RF antenna transmission system 5. For controlling the gradient pulses of the gradient system 6 in order to appropriately switch the gradient pulses according to the predetermined pulse sequence, the control device 13 comprises a gradient system interface 16. The sequence control unit 14 communicates in an appropriate manner, for example, through transmission of sequence control data SD, with the radio-frequency transmission system 15 and the gradient system interface 16 in order to perform the pulse sequences. In addition, the control device 13 comprises a radio-frequency reception system 17 (that also communicates in an appropriate manner with the sequence control unit 14), in order to receive within the scanning window predetermined by the pulse sequence PS and coordinated by means of the RF antenna reception system 7 magnetic resonance signals, in the context of the present invention echo signals (described below) and thus, after digitalization, demodulation and low pass filtering, acquire the complex raw data for the individual slices.

(14) Here, a reconstruction device 18 accesses the acquired raw data and reconstructs magnetic resonance image data for the slices. Usually, also this reconstruction is performed based on parameters that are predetermined in the respective measurement protocol. For example, these image data can be stored in a memory 19. In the case at hand, the reconstruction device 18 is designed in such a way that it can operate in accordance with the invention-based method which is subsequently described in more detail by means of FIGS. 6 and 7. In particular, it is possible to combine the raw data and/or image data of a slice in a specific data combination unit 20 of the reconstruction device 18. The data combination unit 20 can be formed by software modules and usually does not require any additional hardware component.

(15) The central control device 13 can be operated via a terminal having an input device 10 and a display device 9 by means of which an operator can operate the entire MR system 1. It is also possible to display MR images on the display device 9 and measurements can be planned and started by means of the input device 10, optionally in combination with the display device 9. In particular, as described above, it is possible to select and optionally modify control protocols P with appropriate pulse sequences PS.

(16) Moreover, the invention-based MR system 1 and especially the control device 13 can comprise a plurality of additional components, which are here not particularly shown but which are usually available in such systems, for example, a network interface to be able to connect the entire system with a network and to be able to exchange raw data and/or image data or parameter cards, even additional data, for example, patient-relevant data or control protocols.

(17) Those of ordinary skill in MRI know how to acquire appropriate raw data by radiation of RF pulses and switching of gradient pulses and how to reconstruct MR images from the acquired raw data. Therefore, the process does not require any further description herein. The basic principle of different slice-measuring sequences, for example, the TSE pulse sequences described above is also known. Nevertheless, a typical conventional TSE sequence will be explained using FIG. 2 in order to demonstrate the differences to an invention-based pulse sequence which is described below in more detail by means of an example shown in FIGS. 3A and 3B. In a conventional manner, the arrangement of the RF and gradient pulses, magnetic resonance signals (echo signals) and scanning windows over the time t (from left to right) is displayed on respectively different axes in the pulse diagrams of FIG. 2, as well as FIGS. 3A and 3 B. On the top axis, the RF pulses and echo signals are represented, on the second axis the gradient pulses in slice selection direction, on the third axis the gradient pulses in readout direction, on the fourth axis the gradient pulses in phase encoding direction, and on the bottom axis the scanning windows. The following applies to the three gradient axes: the horizontal axis represented by a dotted line is the respective zero line. The height of the signals represents (not necessarily true to scale), respectively, the relative amplitude. The sign of the amplitude (in relation to the zero axis) corresponds to the direction of the gradient field. Some of the gradient pulses shown in the figures have one or several capital letters. These letters represent the zero moment of a gradient pulse or the zero moment accumulated by the gradient pulse during a time interval. These moment specifications have the purpose of facilitating the understanding of the pulse sequence. In particular, different gradient pulses or partial intervals of different gradient pulses which accumulate the moment equal to zero are provided with the same capital letter.

(18) FIG. 2 shows the first two echoes E of a conventional TSE sequence in prior art. The sequence starts with a slice-selective 90 RF slice excitation pulse a (subsequently called excitation pulse), succeeded by a series of slice-selective RF refocusing pulses .sub.1, .sub.2, .sub.3 (subsequently called refocusing pulse). After each refocusing pulses .sub.1, .sub.2, .sub.3 exactly one echo is formed, which is read is a particular scanning window (readout interval) AQ. At the same time, the duration T.sub.ACQ of the scanning window AQ is determined by the number of data points that have been read and the time interval of two data points, known as dwell time. FIG. 2 shows only three refocusing pulses .sub.1, .sub.2, .sub.3, in T2-weighted imaging the number is usually considerably higher (even in the invention-based pulse sequence shown below and any variations thereof) and varies, depending on the application, between three and several hundred refocusing pulses. The number of refocusing pulses is also called echo train length (ETL).

(19) The represented sequence fulfills the so-called Carr Purcell Meiboom Gill condition (CPMG condition), which ensures that echo signals E of spins that follow different coherent echo paths are constructively superimposed at the echo moment. Among other things, the CPMG condition requires that each phase which a spin acquires between any two consecutive refocusing pulses .sub.1, .sub.2, .sub.3 has to be equal. For example, the CPMG condition is described in more detail in the Handbook of MRI Pulse Sequences, Elsevier Academic Press; illustrated edition (Sep. 21, 2004); ISBN-10: 0120928612; ISBN-13: 978-0120928613 by Matt A. Bernstein, Kevin E. King, Xiaohong Joe Zhou.

(20) First, the time interval T.sub. between the refocusing pulses .sub.1, .sub.2, .sub.3 is selected to be twice as long as the time interval between the isodelay point of the excitation pulse and the center of the first refocusing pulse .sub.1.

(21) Secondly, the phase of the refocusing pulses is rotated by 90 compared to the phase of the excitation pulse (i.e., when, for example, the B.sub.1 field of the excitation pulse extends along the x-axis in a coordinate system rotating about the z-axis in which the B.sub.0 field is located, the B.sub.1 field of the refocusing pulse is applied parallel or anti-parallel to the y-axis).

(22) At the excitation pulse , as well as the refocusing pulses .sub.1, .sub.2, .sub.3, the width of the excited slice is, respectively, controlled by the bandwidth of the RF pulse and by a slice selection gradient pulse GS.sub.1, GS.sub.4, GS.sub.7, GS.sub.10, which is present during the radiation of the excitation or refocusing pulse .sub.1, .sub.2, .sub.3. Directly prior to or directly following each refocusing pulse , a respective crusher gradient GS.sub.3, GS.sub.5, GS.sub.6, GS.sub.8, GS.sub.9, GS.sub.11 is switched which have the function of dephasing the FID of the refocusing pulse .sub.1, .sub.2, .sub.3 prior to the subsequent scanning window AQ, so that it does not supply any signal contribution. FID (free induction decay) signifies the transient signal of a spin system induced by a single RF pulse. In other words, it involves the signal emanating from spins for which the refocusing pulse acts as an excitation pulse.

(23) At the same time, the left and right crusher gradient GS.sub.3, GS.sub.5 are supposed to have the same zero moment. In the figures, the absolute value of the moment of a gradient pulse (which corresponds through the surface below the pulse to the amplitude of the gradient integrated by time) is represented by capital letters in the respective surface area of the pulse. Consequently, the capital F in FIG. 1 show that the left and the right crusher gradient GS.sub.3, GS.sub.5 have the same moment. Furthermore, the crusher gradients GS.sub.3, GS.sub.5, GS.sub.6, GS.sub.8, GS.sub.9, GS.sub.11 of different refocusing pulses .sub.1, .sub.2, .sub.3 also have again the same moment F. Any different selection infringes the CPMG condition.

(24) After the excitation pulse , a slice re-phasing gradient pulse GS.sub.2 is required, the moment A of which is equal to the negative of the moment A accumulated through the slice selection gradient pulse GS.sub.1 between the isodelay point of the excitation pulse and the end of the slice selection gradient pulse GS.sub.1.

(25) The temporal sequence of the PF pulses is arranged in the manner that a spin echo signal E is formed at the moment T.sub./2 following each refocusing pulse a.

(26) Each spin echo signal E is frequency-encoded by a readout gradient pulse GR.sub.2, GR.sub.3. A readout pre-phasing gradient pulse GR.sub.1 between the excitation pulse a and the first refocusing pulse .sub.1, the moment B of which coincides with the moment B which accumulates a spin from the start of the readout gradient pulse GR.sub.2, GR.sub.3 until the center of the echo signal E, makes sure that the total moment is zero at the time of the echo signal E.

(27) The second part of the readout gradient GR.sub.2, GR.sub.3 following the echo also has the surface B and is therefore used as pre-phasing gradient for spins that follow coherent echo paths, which are located between more than one pair of refocusing pulses .sub.1, .sub.2, .sub.3 in the transverse plane.

(28) For phase-encoding the echo signal E, provision has been made for a phase-encoding gradient pulse GP.sub.1, GP.sub.3 which is, respectively switched between the end of the refocusing pulse .sub.1, .sub.2, .sub.3 and the start of the scanning window AQ. After the end of the readout interval AQ and prior to the start of the next refocusing pulse .sub.2, .sub.3, the moment D, E acquired by said gradient pulse GP.sub.1, GP.sub.3 has to be compensated with a moment D, E by a phase refocusing gradient pulse GP.sub.2, GP.sub.4, in order to fulfill the above-mentioned CPMG condition.

(29) For drawing-related reasons, FIG. 2 shows only the first two echo signals E. By repeating the framed sequence block SB, it is possible to obtain the sequence diagram for the complete sequence variation which consists of the excitation block AB with the excitation pulse and the subsequent echo train. The echo train is configured of several series connected sequence blocks SB. Each sequence block SB includes a refocusing pulse with subsequent echo signals E, a slice selection gradient, the right crusher gradient of the inherent refocusing pulse, as well as the right crusher gradient of the following refocusing pulse, a readout gradient, a readout interval AQ, a phase-encoding gradient and an appropriate phase refocusing gradient. When different echo signals E encode different k-space lines, the moment of the phase-encoding gradient GP.sub.1, GP.sub.3 and the phase-encoding refocusing gradient GP.sub.2, GP.sub.4 is varied between the repetitions of the sequence block SB. All other pulses do not change their value so as not to infringe the CPMG condition.

(30) FIGS. 3A and 3B (combined as in FIG. 3) show the start of a pulse sequence for simultaneous refocusing of several excited slices according to a possible embodiment of the invention. To provide a better representation, the pulse diagram was divided in two partial FIGS. 3A and 3B, wherein FIG. 3A shows an excitation block AB in which the slices are excited for the first time, a preparation block and a first sequence block SG.sub.1. This is followed by a number of sequence blocks SB. Each sequence block SB.sub.1, SB starts with a refocusing pulse , , , . FIG. 3B shows the first two of these further sequence blocks SB, as well as the start of a subsequently following similar sequence block SB with a refocusing pulse . At the same time, the pulse diagrams shown in FIGS. 3A and 3B connect seamlessly to one another at the dash-dotted line. As with the conventional pulse sequence shown in FIG. 2, the number of the sequence blocks SB or refocusing pulses within the pulse sequence can be varied in virtually any way and, for example, depending on the application, it can amount to between three and several hundred refocusing pulses. The subsequent description shows that the following sequence blocks SB differ only with regard to the height of the phase-encoding gradient pulses GP.sub.5, GP.sub.7, GP.sub.9 and the phase-encoding refocusing gradient pulses GP.sub.6, GP.sub.8.

(31) Based on the conventional TSE sequence according to FIG. 2, the following changes have to be made in order to achieve such an invention-based pulse sequence:

(32) Instead of only one slice excitation pulse , at least two slice excitation pulses .sub.1, .sub.2 are now emitted within the excitation block AB in a (first) time interval T.sub.from one another. The two slice excitation pulses .sub.1, .sub.2 are exciting the spins in different, usually parallel slices. When thickness of the slices and the radio-frequency bandwidth of both slice excitation pulses .sub.1, .sub.2 coincide, they only differ with regard to their carrier frequency. The length of the time interval T.sub.between the slice excitation pulses .sub.1, .sub.2 is subsequently described in more detail. To provide a better representation, only the excitation and simultaneous refocusing of m =2 slices are shown. However, it is certainly possible to use the sequence for a simultaneous refocusing of more than two slices (i.e., m =2).

(33) Furthermore, for the preparation of magnetization, so-called diffusion gradient pulses GS.sub.3, GS.sub.5 are inserted left and right of the refocusing pulse R. The time T.sub./2 between the last slice excitation pulse .sub.2 and the refocusing pulse, and thus also the time T.sub. is extended depending on the desired maximum diffusion weighting (and thus with the period of the diffusion gradient). The diffusion preparation can be performed simultaneously on one, two or all three spatial axes. To provide a clearer view, in FIG. 3 it is shown only on the slice selection axis. When the diffusion gradients have sufficient moment, the crusher gradients GS.sub.3, GS.sub.5 shown in FIG. 2 can be eliminated about the first refocusing pulse , because their function is assumed by the diffusion gradient pulses GS.sub.3, GS.sub.5. The preparation block shown in FIG. 3A is only to be viewed as an example. It can be substituted by a number of other blocks, depending on the desired preparation. Subsequently, it is only required that the preparation block is designed in such a way that it forms an echo with respect to its temporal symmetrical axis. Thus it will usually have at least one refocusing pulse . In other words, when an excitation pulse is emitted in a temporal interval T.sub./2 prior to the temporal center of the preparation block , said excitation pulse forms an echo in the temporal interval T.sub./2 following its temporal center. Therefore, the simplest embodiment involves a single symmetrical refocusing pulse , the temporal center of which coincides with the temporal center of the preparation block , flanked by two crusher gradients.

(34) Furthermore, the refocusing pulses , , , , are designed in such a way that they respectively acquire or impact at least partially the slice excited by the excitation pulse .sub.1, as well as the slice excited by the excitation pulse .sub.2. In the simplest case, this is achieved by elevating the width of the refocused slice in relation to the width of the excited slices. When z depicts the width of an excited slice and d is the interval between the two slices, the width of the slice achieved by a refocusing pulse .sub.1, .sub.2, .sub.3 is selected at least d+z (or with m slices in a more general form (m1)d+z). For example, when the radio-frequency bandwidth of the refocusing pulses .sub.1 .sub.2, .sub.3 from the sequence shown in FIG. 2 is maintained, this can be achieved in that the amplitudes of the slice selection gradient pulse GS.sub.4, GS.sub.7, GS.sub.10, GS.sub.13, GS.sub.16 respectively emitted with the refocusing pulse , , , , is reduced in relation to the slice selection gradient pulse GS.sub.4, GS.sub.7. Furthermore, when the carrier frequency of the refocusing pulse , , , , is selected, respectively, in such a way that the center of the refocusing slice is located exactly in the center of the m excited parallel slices. The crusher gradient pulses GS.sub.6, GS.sub.8, GS.sub.9, GS.sub.11, GS.sub.12, GS.sub.14, GS.sub.16, GS.sub.17 do not have to be adjusted.

(35) While in the conventional pulse sequence according to FIG. 2 exactly one echo is formed after each refocusing pulse .sub.1, .sub.2, .sub.3, the invention-based sequence produces up to 2 m echo signals E.sub.1a, E.sub.1b, E.sub.1a, E.sub.2b after each refocusing pulse , , , , , wherein m represents the number of the simultaneously refocused slices. Therefore, in FIGS. 3A and 3B, m=2.

(36) If all readout parameters would maintain the pulse sequence shown in FIG. 2 and thus also the readout duration T.sub.ACQ, the (third) time interval T.sub. between two refocusing pulses , , , of the further sequence blocks SB would have to be appropriately adjusted.

(37) Alternatively, it is also possible to reduce the readout duration T.sub.ACQ of the scanning windows AQ, for example, by reducing the dwell time by a factor a. When the field of view (FoV) and the number of read data points remain unchanged, it means that the amplitude of the readout gradient pulses GR.sub.2, GR.sub.3, GR.sub.4, GR5 has to be increased by the same factor a in relation to the readout gradient pulses GR.sub.2, GR.sub.3 according to FIG. 1 so that the k-space interval of the readout points remains unchanged. At the same time, said factor a is limited by the maximum gradient amplitude of the gradient system and the minimum dwell time of the analog to digital converter.

(38) The time interval between the m first consecutive echo signals E.sub.1b, E.sub.2b and the m last consecutive echo signals E.sub.1a, E.sub.2a after each refocusing pulse , , , , is equal to the time interval T.sub. of the slice excitation pulse .sub.1, .sub.2. To avoid that the different scanning windows AQ are overlapping with the different echo signals E.sub.1a, E.sub.2a, E.sub.1b, E.sub.2b, the duration T.sub.ACQ of the scanning windows AQ has been determined so that the time interval T.sub. between two consecutive echo signals E.sub.1a, E.sub.2a, E.sub.1b, E.sub.2b, is limited downward:
T.sub.T.sub.ACQ(1)

(39) In a preferred embodiment of the invention, provision has been made that spin echoes and gradient echoes coincide. In the context of the present invention, this coincidence of spin echoes and gradient echoes has to be understood as follows: gradient echoes involve that the phase accumulated as a result of the switched gradient fields (with the exception of the phase encoding gradients) have to be zero at the echo moment. This is a vital prerequisite for forming an echo. The spin echo involves that the phase accumulated as a result of a local off-resonance is zero at the echo moment. For example, a CPMG-TSE sequence (described, for instance, in the context of FIG. 2) fulfills at each echo moment this gradient condition, as well as the spin echo condition. Consequently, in this case, spin echo and gradient echo coincide.

(40) By this coincidence of spin echoes and gradient echoes, it is achieved that the sequence is robust against local inhomogeneities of the B.sub.0 field. The first spin echo is formed after the time interval T.sub. following the refocusing pulse of the first sequence block SB.sub.1, wherein T.sub. represents the time interval between the last echo signal E.sub.1a formed by the preparation module and the isodelay point of the refocusing pulse . As described below, the position of the first gradient echo can be determined by selecting the readout pre-phase gradient pulse GRP between the last slice excitation pulse .sub.2 and the preparation block .

(41) Preferably, the (second) time interval T.sub. is selected at first as short as possible, in order to implement the shortest possible echo spacing T.sub.. (The time interval between two consecutive refocusing pulses, i.e., in this case the space between two sequence blocks is described as echo spacing T.) As a result, it is possible to keep at a minimum the so-called T2 blurring, i.e., blurring artifacts due to an unavoidable T2 dissolution along the echo train. FIG. 3A shows that the time interval T.sub. is limited downward by the duration T.sub.REF of the refocusing pulse , , , , , the duration T.sub.ACQ of an excitation interval AQ and a period TG:
T.sub.T.sub.REF/2+TG+T.sub.ACQ/2(2)

(42) The period TG is the time required for phase encoding, a crusher gradient or for accelerating the readout gradient. Since the above-mentioned gradient pulses are usually switched in parallel, the longest one determines the period TG.

(43) With the selection of the time intervals T.sub.ACQ, T.sub. and T.sub., it is also possible to determine the time interval T.sub. of two consecutive refocusing pulses , , , , of the sequence blocks SB.sub.1, SB:
T.sub.2(m1)T.sub.+2T.sub.+T.sub.ACQ(3)

(44) With these time periods T.sub. and T.sub.ACQ, in turn, the flattop duration TGRO.sub.2 (the period of the medium range of a trapezoidal pulse in which the amplitude does not vary) of the readout gradient GRO.sub.2 of the first sequence block SB.sub.1 is limited downward:
TGRO.sub.2T.sub.ACQ+(2m2)T.sub.+(T.sub.(2m2T.sub.)=T.sub.ACQ+T.sub.2T.sub.(4)

(45) Again, in order to keep the T2 blurring as low as possible, it is preferred to select the time period T.sub. as short as possible which, in turn, results in the fact that TGRO.sub.2 is limited upwardly:
TGRO.sub.2T.sub.T.sub.REF2TGRO.sub.RT.(5)

(46) At the same time, TGRO.sub.RT represents a ramp time of the symmetrical readout gradient pulse, which is limited downward as follows by the maximum gradient rise time S and the amplitude of the readout gradient A.sub.GRO:
TGRO.sub.RTSA.sub.GRO(6).

(47) The same time conditions apply to the readout gradients GRO.sub.3, GRO.sub.4, GRO.sub.5 of the further sequence blocks SB or their flattop durations TGRO.sub.3, TGRO.sub.4, TRGO.sub.5.

(48) Even if the echo signals E.sub.2a, E.sub.1a formed by the preparation block are read, it is possible to switch a readout gradient pulse GRO.sub.1 at the appropriate point in time. Usually, said readout gradient pulse GRO.sub.1 has the same amplitude A.sub.GRO and the same ramp time as the following readout gradient pulses GRO.sub.2, GRO.sub.3, GRO.sub.4, GRO.sub.5 of the sequence blocks SB.sub.1, SB. If, in addition, the flattop duration is selected prior to the first echo signal E.sub.2 and after the last echo signal E.sub.1a corresponding to the design of the readout gradient pulses GRO.sub.2, GRO.sub.3, GRO.sub.4, GRO.sub.5, it results in a total moment of the first readout gradient GRO.sub.1 of 2B+(m1)C. However, this is not a mandatory selection. It is also possible to omit completely the readout gradient GRO.sub.1, if the echo signals E.sub.2a, E.sub.1a formed by the preparation block should not be read. However, such deviations have to be taken into consideration when determining the zero moment of the readout pre-phase gradient pulse GRP. Optionally, it is possible that the echo signals E.sub.2a, E.sub.1a are phase encoded by switching a phase encoding gradient and a phase refocusing gradient with opposite signs and equal absolute zero moment (in FIG. 3A zero moment G).

(49) Between the slice selection gradient GS.sub.1,1, GS.sub.1,2 of two consecutive slice excitation pulses .sub.1, .sub.2, a gradient GS.sub.2,1 is switched in slice selection direction, the zero moment of which is equal to the negative of the sum of the moment accumulated between the isodelay point of the first excitation pulse .sub.1 and the end of the first slice selection gradient GS.sub.1,1 and the moment accumulated between the start of the second slice selection gradient GS.sub.1,2 and the isodelay point of the second excitation pulse .sub.2. Consequently, its sign is opposite to the sign of the slice selection gradients GS.sub.1,1 and GS.sub.1,2. When using symmetrical slice selection gradients GS.sub.1,1, GS.sub.1,2 and excitation pulses .sub.1, .sub.2 and centering the isodelay points in the center of the flattops of the slice selection gradient GS.sub.1,1, GS.sub.1,2, as shown in FIG. 3A, the moment of this gradient GS.sub.2,1 amounts to 2A, and thus it is equal to the negative moment 2A of a slice selection gradient GS.sub.1,1 and GS.sub.1,2. Under the above-mentioned conditions, the actions of the gradients can also be interpreted for easier understanding as follows: anticipatory, the gradient GS.sub.2,1 compensates the moment that the spins of the first slice excited by the first excitation pulse .sub.1 will be accumulated in sequence of the slice selection gradient GS.sub.1,2 of the second excitation pulse .sub.2. The gradient pulse GS.sub.2,2 immediately following the last excitation pulse .sub.2 in slice selection direction operates as mutual slice refocusing pulse of the first excitation pulse .sub.1 and the second excitation pulse .sub.2.

(50) Between two consecutive excitation pulses .sub.1, .sub.2, a further gradient pulse GRO.sub.0 is switched in readout direction the zero moment C of which corresponds according to amount exactly with the zero moment which is accumulated in readout direction between two consecutive echo signals (from the group of the first m echo signals E.sub.1b, E.sub.2b or from the group of the last m echo signals E.sub.2a, E.sub.1a). Therefore, FIG. 3A reads as follows:
|C|=|TA.sub.GRO|(7),
wherein A.sub.GRO represents the amplitude of a readout gradient pulse GRO.sub.2, GRO.sub.3, The sign of the gradient GRO.sub.0 depends on the number of refocusing pulses within the preparation module. When that number is unequal, the gradient GRO.sub.0 has the same sign as the readout gradients GRO.sub.1, GRO.sub.2, . . . , otherwise it has an opposite sign.

(51) Furthermore, for spin echoes to coincide in the desired manner with gradient echoes, the temporal position and the zero moment of the readout pre-phase gradient pulse GRP have to be appropriately selected. This allows for great freedom. Subsequently, a distinction is made between two cases that are especially important in practice. For each case, sufficient design criteria are provided that would result in a concurrence of spin echoes and gradient echoes. However, these design criteria do not claim completeness. It is also possible to find different equivalent solutions.

(52) In the first case shown in FIG. 3A, the first m echoes formed by the preparation block are read. In this case, it is preferred (but not required) to apply readout pre-phase gradient pulse GRP between the last slice excitation pulse .sub.2 and prior to the preparation block . The zero moment of the readout pre-phase gradient pulse GRP is now selected in such a way that the phase that accumulates the spins of the last excited slice .sub.2 between the start of the preparation block and the first echo E.sub.2a of the slice following the preparation module is exactly compensated. Because of the fact that each refocusing pulse negates the phase accumulated by the preceding gradients, it is particularly important to consider the number of refocusing pulses of the preparation block . In the example depicted the moment GRP is equal to the moment B, which is accumulated by the readout gradient GRO.sub.1 between start and spin echo E.sub.2a.

(53) Furthermore, in this case, the moment that is accumulated in readout direction between the last echo E.sub.1a (of the slice .sub.1 excited first) formed by the preparation block and the first sequence block SB.sub.1 is selected equal to the moment that is accumulated in readout direction within the sequence block SB.sub.1 between refocusing pulse and first echo E.sub.1b. The symmetrical form of the readout gradient GRO.sub.1 shown in FIG. 3A fulfills this criterion in a natural manner.

(54) Therefore, at the echo moment of the echo signal E.sub.2b shown in FIG. 3A, the total moment amounts to zero:
B(2B+C)+(B+C)=0.(8)

(55) In a second case (not shown), the first m echoes formed by preparation block are not read (i.e., the readout gradient GRO.sub.1 is omitted here, if readout is not performed but the readout gradient GRO.sub.1 is switched, the first case applies) and the preparation block is self-refocusing. The term self-refocusing means that all. gradients switched by the preparation block are compensated in readout direction within the preparation block . Because of the symmetrical design of the preparation block with respect to the temporal symmetrical axis, this requirement is usually fulfilled in a natural manner or it is easy to fulfill.

(56) In this second case, it is preferred (but not required) to apply the readout pre-phase gradient pulse GRP between the last echo E.sub.1a formed by the preparation block and the start of the first sequence block S.sub.B1. Its moment is selected equal to the moment B that is acquired in readout direction within the sequence block SB.sub.1 between refocusing pulse and first echo E.sub.1b. In this procedure, m1 further pre-phase gradients are switched, respectively, with a moment C between the echoes formed by the preparation block .

(57) Subsequently, for easier understanding, the function of the invention-based pulse sequence described above is explained in more detail, wherein these explanations apply in general to such pulse sequences with m excited and simultaneously refocused slices.

(58) For this purpose, it is first of all necessary to consider the spins in the slice which are impacted by the last excitation pulse .sub.m (in FIG. 3A the second slice excited by the excitation pulse .sub.2). All preceding excitation pulses .sub.1, . . . , .sub.m-1 (in FIG. 3A the first excitation pulse .sub.1) do not impact these spins because the resonance condition has not been fulfilled. As a result, all preceding gradient pulses do not have an impact on the spins because the longitudinal magnetization is not impacted by gradient fields.

(59) A first echo is formed by the effect of the preparation block at the temporal interval T.sub. following the excitation pulse .sub.m. At this moment, the phase accumulated by the readout pre-phase gradient GRP (with the total moment B) is exactly compensated by the first part of the readout gradient GRO.sub.1 and thus produces a simultaneous gradient echo, which can be read in the echo signal group E.sub.ma (E.sub.2a in FIG. 3A). (here and subsequently, the echo signals are described as echo signal groups because, as explained above, for the most part several signals coincide.)

(60) The spins involved in this echo signal E.sub.2a are again refocused by the refocusing pulse of the first sequence block SB.sub.1 and form a spin echo following the refocusing pulse at a time interval T.sub.2ato2b/2=T.sub.+T.sub.. Again, at the echo moment, the phase that has been accumulated by the second half of the readout gradient GRO.sub.1 (total moment C+B) since the first echo is exactly balanced by the phase (total phase B+C) accumulated as a result of the second readout gradient GRO.sub.2, thus making it possible to read the echo in the echo group E.sub.mb of the first sequence block SB.sub.1.

(61) This signal (or the impacted spins) is again refocused by the refocusing pulse of the second sequence block SB (i.e., of the first of the further sequence blocks SB) and forms a second spin echo after a time interval T.sub.2bto2a/2=T.sub.+T.sub.). This involves the signal directly refocused twice of the first echo signal of the second slice formed by the preparation block (In the example depicted in FIGS. 3A and 3B, the echo formed by the preparation block is already a spin echo. Here, counting starts only after the preparation block , in order to keep to the extent possible the description independent of the special embodiment of the preparation block.) Again, at this further echo moment, the phase that was accumulated as a result of the readout gradient GRO.sub.2 between the first spin echo and the refocusing pulse (i.e., the total moment D=C+B) has been exactly compensated by the phase that is accumulated by the third readout gradient GRO.sub.3 between the refocusing pulse and the second spin echo. Therefore, it is possible to use also the second spin echo in the echo signal group E.sub.2a of the second sequence block SB. At the same time, the echo spacing T.sub.2ato2b describes the time interval between an echo of the echo signal group E.sub.2a and the following echo of the echo signal group E.sub.2b. Correspondingly, the echo spacing T.sub.2bto2a depicts the time interval between an echo of the echo group E.sub.2b and the following echo of the echo group E.sub.2a.

(62) In the second sequence block SB, the first stimulated echo of the second slice is also formed by the mutual impact of the refocusing pulses and . Again, the first radio-frequency pulse .sub.m operates as a slice excitation pulse which folds magnetization in the transverse plane. When the echo has been stimulated, the refocusing pulse of the first sequence block SB.sub.1 operates as a so-called restore pulse, i.e., it folds back part of the transverse magnetization in longitudinal direction which is then folded again into the transverse plane by the refocusing pulse of the second sequence block SB. It is said that said magnetization is stored in longitudinal direction between refocusing pulse and refocusing pulse because, as longitudinal magnetization, it is not impacted by the gradient fields and only subjected to the relatively slow T1 relaxation. Consequently, this first stimulated echo is formed at the time interval T.sub.2ato2b/2=T.sub.+T.sub. following the refocusing pulse in the echo group E.sub.2b of the second sequence block SB. Also at this moment, the phase acquired as a result of readout gradient GRO.sub.1 is exactly compensated by the phase acquired as a result of readout gradient GRO.sub.3 (the total moment amounts to B+C), resulting in the fact that the stimulated echo and the gradient echo coincide. The first stimulated echo and the second spin echo (the directly refocused first spin echo) of the latest excited slice do not coincide temporally and can be read separately. This distinguishes the sequence from a sequence block SB of a CPMG sequence shown in FIG. 2 or the sequence block SB disclosed in DE 10 2012 204 434. This is advantageous because, for the reasons described above, the phase position of the echo formed by the preparation block is not known and therefore the B.sub.1 vector of the refocusing pulse cannot be applied parallel or anti-parallel to said phase position. However, this is a prerequisite for constructive overlapping of the second spin echo and the first stimulated echo.

(63) Subsequently, the spins in the first slice which are impacted by the first excitation pulse .sub.1 are considered. The bandwidth of the slice excitation pulses .sub.1, .sub.2 and the amplitude of the slice selection gradients GS.sub.1,1, GS.sub.1,2 are selected in such a way that the spins of a slice are not impacted by the following excitation pulses .sub.2, ..., .sub.m, (in the simplified example shown in FIG. 3A only .sub.2). However, because of the fact that all gradient fields that are switched after a slice excitation pulse impacts the spins that have been tipped into the transverse plane by the slice excitation pulse, the spins of the first slice accumulate, among other things, a zero moment by the slice selection gradient GS.sub.2,1 of all following slice excitation pulses .sub.2. As described above, in order to avoid that the signal is de-phased by the slice selection gradients GS.sub.2,1 of the following slice excitation pulses .sub.2 , a negative gradient pulse GS.sub.2,1 with a total moment of -2A was switched in slice selection direction between two consecutive slice excitation pulses .sub.1, .sub.2. One half of the moment (A) is used as an ordinary slice refocusing moment for the slice selection gradient GS.sub.1,1 of the preceding excitation pulse .sub.i, (here i =1). The other half of the moment (A) is used as pre-phase gradient which compensates the positive moment accumulated by the slice selection gradient GS .sub.1,2 of a subsequent slice excitation pulse .sub.i+1 (here .sub.2) between the start of the slice selection gradient GS.sub.1,2 and isodelay point of the subsequent slice excitation pulse .sub.2. The moment (again A) accumulated by the second half of the slice selection gradient GS.sub.1,2 of the subsequent excitation pulse is compensated in a conventional manner by the slice refocusing gradient GS.sub.2,2 of the subsequent slice excitation pulse .sub.2. Because of the fact that a gradient filed does not impact spins that are aligned in longitudinal direction to the basic magnetic field B.sub.0, the negative gradient GS.sub.2,1 between the slice excitation pulses .sub.1, .sub.2has no impact on the spins that are excited by all subsequent excitation pulses (also here .sub.2).

(64) By the above-mentioned measures for selecting the parameter (thickness of the slices, bandwidth, etc.) of the slice excitation pulses .sub.1, .sub.2, the refocusing pulse S of the preparation block simultaneously refocuses the signal of all spins that were excited by one of the m excitation pulses .sub.1, . . . , .sub.m. The spins of the first slice, which were excited by the excitation pulse .sub.1, are refocused to a first echo at the moment T.sub.+T.sub./2 following the first refocusing pulse (i.e., 2T.sub.+T.sub. following the excitation pulse .sub.1). Switching the gradient pulse GRO.sub.0 in readout direction with the zero moment C between two consecutive slice excitation pulses .sub.1, .sub.2 compensates, together with the readout pre-phase gradient pulse GRP, the moment B between the latest slice excitation pulse .sub.m (here .sub.2) and the preparation block , the moment that is acquired as a result of the first readout gradient GRO.sub.1 between the preparation block and the first echo signal E.sub.1a of the first slice. Gradient echo and spin echo of the spins of the first slice are produced simultaneously and can be read as echo group E.sub.1a at the moment T.sub.+T.sub./2 following the refocusing pulse of the preparation block . It should be noted that the spins of previously considered slice m (here slice 2) have again accumulated the moment (m1)C (here only C) at the moment of the first echo of the first slice and are therefore de-phased. This is the case because these spins have not seen the gradient between the excitation pulses because it was temporally switched prior to its excitation. Conversely, this gradient impacts the spins of the preceding slice (here the first slice). Therefore, at the moment of the spin echo of slice m, they are still de-phased by a moment (m1)C and do not contribute a signal for the echo group E.sub.m (here E.sub.2), despite adequately great moment C. Switching the gradient pulse GRO.sub.0 in readout direction with zero moment C between two consecutive slice excitation pulses .sub.1, .sub.2, together with the temporal sequence of the radio-frequency pulses, serves the purpose of securely separating the scanning windows AQ of different slices.

(65) The first echo signal E.sub.1a of the first slice is also refocused again by the refocusing pulse of the first sequence block S.sub.1 and forms a first spin echo at the moment T.sub.1ato1b/2=T.sub. following the refocusing pulse . (This involves the directly refocused signal of the first echo of the first slice formed by the preparation block . In the example depicted in FIG. 3a, the echo formed by the preparation block is already a spin echo. Also in this case, counting starts only after the preparation block , in order to keep to the extent possible the description independent of the special embodiment of the preparation block.) At this moment T.sub.1ato1b/2=T.sub. following the refocusing pulse , the zero moment B accumulated between the echo of the first slice formed by the first readout gradient pulse GRO.sub.1 and the refocusing pulse is exactly compensated so that the echo in the echo signal group E.sub.1b can be read. The signal of the other slices, however, is still de-phased by the first readout gradient pulse GRO.sub.1 (for m=2 by the moment C).

(66) Then the signal of the first spin echo of the first slice is again refocused by the refocusing pulse of the second sequence block SB and also forms a second spin echo at the moment T.sub.1bto1a/2=T.sub.T.sub.. Again, at this moment, the phase that was accumulated as a result of the readout gradient pulse GRO2 between first spin echo and refocusing pulse (total moment C+D+C+B) is exactly compensated by the phase that is accumulated by the third readout gradient pulse GRO3 between refocusing pulse and second spin echo of the first slice. As a result, it is possible to use also the second spin echo in the echo group E.sub.1a of the second sequence block SB.

(67) The first stimulated echo of the echo of the first slice formed by the preparation block is produced at the moment T.sub. following the refocusing pulse . The spins of the first slice, which contribute to said stimulated echo, were located in the transverse plane between preparation block and refocusing pulse of the first sequence block SB.sub.1 and, at the same time, accumulated the moment B in readout direction. Their signal was stored in longitudinal direction between the two refocusing pulses and and therefore the readout gradient GRO.sub.2 was ineffective. By the refocusing pulse , they were folded back into the transverse plane. At the moment of the stimulated echo, the readout gradient GRO.sub.3 of the second sequence block SB exactly compensated the zero moment accumulated prior to the refocusing pulse . As a result, it is possible to read the first stimulated echo in the echo group E.sub.1b of the second sequence block SB. At this moment, the spin echo of the first slice read in the echo group E.sub.1a is still de-phased by a moment 2C+D (in general mC+D), and the signal of the second slice read in the echo signal groups E.sub.2a or E.sub.2b is still de-phased by a zero moment D+C or C.

(68) Consequently, it is possible, by means of zero moment C, to control the separation of signals of different slices and, by means of moment D, it is possible to control the separation of signals of different echo paths of the same slice. For example, moment C can be adjusted by selecting T.sub. in step 2 (C=T.sub.A.sub.GRO). Subsequently, moment D (D=(T.sub.2T.sub.2(m1)T.sub.)A.sub.GRO) can be adjusted by selecting T.sub. according to formula (3). In this way, it is also possible to safely eliminate radio-frequency artifacts. If, in the process, the time interval T.sub. or (T.sub.2T.sub.2(m1)T.sub.), which is required for a complete separation of the signals, exceeds the time T.sub.ACQ, the amplitude could be elevated in readout direction at the time intervals in which no reading is done, in order to minimize the echo spacing at the particular separation moment C or D. However, at the same time, it is important to pay attention to eddy currents resulting from additional gradient circuits. Which procedure would result in an improved image quality depends on a plurality of parameters and is easiest determined in empiric manner.

(69) As previously described, the signals of different slices are separated by means of the temporal sequence of radio-frequency pulses and the gradient moment C. The aspect of separating signals of a slice is further discussed below.

(70) Characteristic for a coherent echo path is the time (subsequently depicted as transverse time) in which the spins following said echo path were situated in the transverse plane. For the above-mentioned echo paths the transverse times are as follows:

(71) TABLE-US-00001 Following the . . . E.sub.1a E.sub.1b E.sub.2a E.sub.2b Preparation block T.sub. + 2T.sub. T.sub. Sequence block SB.sub.1 T.sub. + 2T.sub. + 2T.sub. T.sub. + 2T.sub. + 2T.sub. 2.sup.nd Sequence block SB T.sub. + 2T.sub. + 2T.sub. T.sub. + 2T.sub. + 2T.sub. T.sub. + 2T.sub. T.sub. + 2T.sub. + 2T.sub.

(72) Starting with the third sequence block SB, also signals originating from spins that followed different echo paths overlap within an echo signal group.

(73) In general, it applies that coherent echo paths of the first slice with a transverse time of
T.sub.+2T.sub.+gT.sub.,(9)
are acquired in the echo signal group E.sub.1a (i.e., in the example of FIGS. 3A and 3B in the respectively fourth scanning window AQ of a sequence block SB), wherein g represents an even whole number, and that echo paths of the first slice, in which the transverse time amounts to
T.sub.+2T.sub.+2T.sub.+gT.sub.,(10)
wherein g represents an even whole number, are acquired in the echo group E.sub.1b (i.e., in the example of FIGS. 3A and 3B in the respectively first scanning window AQ of a sequence block SB). The same applies to the coherent echo paths of the second slice: the time, in which spins that form echoes within the same echo signal group of the slice were located in the transverse plane differs, respectively, by an even multiple of the echo spacing T.sub..

(74) For example, signals of an echo group of a slice can reach the same echo group of the slice (of a subsequent sequence block SB) with direct double refocusing (and at the same time extend their transverse time with double echo spacing T.sub.). However, with direct double refocusing the spins do not acquire a phase because of the spin echo principle. The same applies to fourfold, sixfold, and other multiple direct refocusing. Moreover, the sequence design achieves, with single direct refocusing, that the signal is received in the respectively different echo group of the same slice. The same applies to a stimulated echo: the singular storage of the signal in longitudinal direction with subsequent folding back in the transverse plane after any number of echo spacing T.sub. exacts the group transition.

(75) The general validity of the above-mentioned principle can be easiest verified by a phase diagram.

(76) FIG. 4 shows the phase diagram of the second slice for the first three sequence blocks SB1, SB. FIG. 5 shows the respective diagram for the first slice. The vertical axis is always the phase ph of the signal for any position, as long as it is fixed. The horizontal axis is the time t. Each line of the phase diagram corresponds to a coherent echo path. When the spins that follow a particular echo path are located in the transverse plane, they accumulate in linear manner with the time t phase ph. The reason for the phase accumulation is the readout gradient which simplification is assumed to be constant between the refocusing pulses, and/or a local B.sub.0 inhomogeneity at the position observed. The phase accumulated as a result of the phase encoding gradient has not been acquired so as not to overcomplicate the representation. The spoiler gradients have been taken into consideration to the extent that refocusing pulses are not the source of new echo paths (with an initial phase of zero), because they suppress these echo paths.

(77) A transverse signal (characterized by an inclined section with phase accumulation) is divided in four new branches (see the respective signal splitting of the refocusing pulses , , ), by a non-perfect refocusing pulse. A portion of the signal remains unaffected, one portion is refocused and one portion of the signal is stored in longitudinal direction. The non-affected and the refocused branch continue to accumulate as transverse magnetization after the refocusing pulse. The refocused branch is characterized by a change in sign of the phase accumulated prior to the refocusing pulse (i.e., this branch starts in a space under the zero axis in FIGS. 4 and 5, respectively). The two horizontal branches correspond to the signal stored in longitudinal direction (without phase accumulation).

(78) When the signal again impacts a (subsequent) non-perfect refocusing pulse, it is divided in two branches. Either it remains unaffected (horizontal branch which as longitudinal magnetization still does not accumulate phase), or it is folded back in the transverse plane (inclined branch). By maintaining the previously accumulated phase, the latter continues phase accumulation after the refocusing pulse.

(79) When an echo path intersects with the zero axis, an echo is formed at this moment.

(80) It is possible to read the transverse time of a specific echo path by adding up the inclined sections from the diagram.

(81) When all k-space lines of one of the excited slices required for image reconstruction are encoded with a sequence module (i.e., a singular excitation of each of the m slices and the subsequent echo train) or several sequence modules (i.e., with repeated excitation of the m slices, each of which followed by the subsequent echo train) of a pulse sequence according to FIGS. 3A and 3B, a complete raw data set for each of the 2 m (here four) echo signal groups is obtained.

(82) Consequently, the splitting in different echo signal groups results in the fact that two complete raw data sets are obtained, respectively, for each slice.

(83) Using the example of echo signal groups E.sub.1a, E.sub.1b of the first slice, we are subsequently describing different possibilities of processing these raw data. All other slices can be processed in the same manner.

(84) If only magnitude images are required, it is possible in a first embodiment to reconstruct one magnitude image, respectively, from the raw data set of the echo signal group E.sub.1a and the raw data set of the echo signal groups E.sub.1b (for example, if acquired k-space points are located on the grid points of a Cartesian grid, in the usual way by a two-dimensional Fourier transformation from the k-space occupied with these raw data in the image space) and subsequently both magnitude images can be added in order to improve the signal to noise ratio. Because of the preceding absolute-value generation, the incoherent phase information of both data sets does not lead to signal erasion. This procedure is analogous to a procedure that is described by Fritz Schick in the article SPLICE: Sub-second diffusion mode, published in October 1997 in the journal Magnetic Resonance in Medicine, volume 38, issue 4, pages 638-644. The article describes a TSE sequence in which the CPMG condition is not fulfilled and the slices are separately refocused.

(85) With an alternative method that is based on the sum of squares, it is possible to obtain an image with an improved signal to noise ration. With said method, the combined image M.sub.1 (x, y) of the first slice (i.e., the pixel values M.sub.1 (x, y) of the first image) is calculated as follows:
M.sub.1(x,y)={square root over (|I.sub.1a(x,y)|.sup.2+|I.sub.1b(x,y)|.sup.2)}(11)

(86) At the same time, I.sub.1a(x,y) is the complete pixel of the image with the spatial image coordinates .sub.(x,y) reconstructed from the raw data set of the echo signal group E.sub.1a, and I.sub.1b (x,y) is the corresponding complex pixel of the image reconstructed from the raw data set of the echo group E.sub.1b. |I.sub.1a(x,y)| depicts the amount of the complex pixel:
|I.sub.1a(x,y)|={square root over (Re{I.sub.1a(x,y)}.sup.2+Im{I.sub.1a(x,y)}.sup.2)}(12)
and, correspondingly, |I.sub.1b(x,y)| depicts the amount of the complex pixel I.sub.1b(x,y):
|I.sub.1b(x,y)|={square root over (Re{I.sub.1b(x,y)}.sup.2+Im{I.sub.1b(x,y)}.sup.2)}(13)

(87) In a further preferred embodiment, both complex images I.sub.1a (x,y) and I.sub.1b (x,y) are first subjected to a phase correction
.sub.1a(x,y)=I.sub.1a(x,y)e.sup.i{circumflex over ()}.sup.1a.sup.(x,y)(14)
.sub.1b(x,y)=I.sub.1b(x,y)e.sup.i{circumflex over ()}.sup.1b.sup.(x,y)(15)

(88) As subsequently shown by means of FIG. 6, the exponents {circumflex over ()}.sub.1a(x,y) and {circumflex over ()}.sub.1b(x,y) are so-called phase maps which can be calculated from the acquired data. Subsequently, the phase corrected images .sub.1a(x,y), .sub.1b(x,y) are added in the complex number range to a complex combined image for the respective slice according to
.sub.1(x,y)=.sub.1a(x,y)+.sub.1b(x,y)(16)

(89) Based on this combined image, it is possible to produce a magnitude image of the respective slice according to
{tilde over (M)}.sub.1(x,y)={square root over (Re{.sub.1(x,y)}.sup.2+Im{.sub.1(x,y)}.sup.2)}(17)
and to produce real part images according to
{tilde over (R)}.sub.1(x,y)=Re{.sub.1(x,y)})(18)
and to produce real part magnitude images according to
{tilde over (R)}.sub.B1(x,y)=|Re{.sub.1(x,y)}(19)
or phase images according to

(90) $\begin{array}{cc}{\stackrel{~}{}}_1\left(x,y\right)~\mathrm{atan}\left(\frac{\mathrm{Im}\left\{{\tilde{I}}_1\left(x,y\right)\right\}}{\mathrm{Re}\left\{{\tilde{I}}_1\left(x,y\right)\right\}}\right)& \left(20\right)\end{array}$

(91) FIG. 6 is a flowchart showing an example of how the phase maps {circumflex over ()}.sub.1a (x,y) and {circumflex over ()}.sub.1b(x,y) required in the equations (14) and (15) can be calculated from the acquired data.

(92) For this purpose, in step Ia, first the raw data set S.sub.1b(k.sub.x,k.sub.y) of the first echo group is duplicated and then in step Ib the raw data set S.sub.1b(k.sub.x,k.sub.y) of the second echo group is duplicated.

(93) Like in a conventional standard reconstruction, in step III.2a or III.2b, from one copy a respective complex image S.sub.1b(k.sub.x,k.sub.y) or I.sub.1b(x,y) is obtained with the aid of a two-dimensional complex Fourier transformation.

(94) The other copy is filtered with a low pass in step IIa or IIb, respectively.) Subsequently, in step III.1a or III.1b, the filtered raw data sets .sub.1a(k.sub.x,k.sub.y) of the first echo group or .sub.1b(k.sub.x,k.sub.y) of the second echo group are transformed in the image space with a two-dimensional Fourier transformation, in order to obtain images with a lower spatial resolution .sub.1a(x,y) or .sub.1b(x,y).

(95) The required phase maps {circumflex over ()}.sub.1a(x,y) and {circumflex over ()}.sub.1b(x,y) could now be directly calculated through phase extraction from the images with the lower spatial resolution according to
.sub.1a(x,y)=|.sub.1a(x,y)|e.sup.i{circumflex over ()}.sup.1a.sup.(x,y)(21)
and
.sub.1b(x,y)=|.sub.1b(x,y)|e.sup.i{circumflex over ()}.sup.1b.sup.(x,y)(22)

(96) However, mathematically it is usually more advantageous to conjugate complexly each pixel of the images with the low spatial resolution .sub.1a(x,y) or .sub.1b(x,y) and to divide them by their amount. The correction maps thus obtained are then multiplied pixel by pixel in step Iva or IVb with the images with high spatial resolution I.sub.1a(x,y) or I.sub.1b(x,y), thus directly attaining from the equations (14) and (15) the phase corrected images

(97) $\begin{array}{cc}{\tilde{I}}_{1a}\left(x,y\right)=I_{1a}\left(x,y\right)\frac{{\hat{I}}_{1a}^{*}\left(x,y\right)}{.Math.{\hat{I}}_{1a}\left(x,y\right).Math.}\mathrm{or}& \left(23\right)\\ {\tilde{I}}_{1b}\left(x,y\right)=I_{1b}\left(x,y\right)\frac{{\hat{I}}_{1b}^{*}\left(x,y\right)}{.Math.{\hat{I}}_{1b}\left(x,y\right).Math.}& \left(24\right)\end{array}$

(98) In step V, it is then possible to perform a complex addition according to equation (16), in order to attain the combined image .sub.1(x,y) of the respective slice.

(99) At this point, it should be added that in the context of the present patent application the term complete raw data set describes a data set with which it is possible to reconstruct an image according to prior art. Therefore, it includes data sets in which individual raw data lines, which are, for example, required for image reconstruction by means of fast Fourier transformation, were not acquired and have to be substituted, for example, with parallel reconstruction technologies.

(100) Furthermore, a complete raw data set can be acquired with a single echo train, such as shown in FIGS. 3A and 3B, or with multiple repetition of the sequence shown in FIGS. 3A and 3B, wherein generally different k-lines are acquired with different repetitions. The first procedure corresponds to the so-called single shot variations HASTE or RARE in conventional turbo spin echo technology, the second the so-called multi shot variations with the respective advantages and disadvantages.

(101) The inventive sequence is compatible with the most important non-Cartesian k-space trajectories known, for example, as PROPELLER sequences, spiral sequences, sequences with concentric rings or radial sequences.

(102) A PROPELLER sequence is a turbo spin echo sequence known from the article Motion and Free-Breathing Cardiac Imaging by James Pipe, published in the journal Magnetic Resonance in Medicine 42:963-969 (1999), which sequence acquires with each echo train a Cartesian k-space segment of a slice that involves the k-space center. A PROPELLER variation of the invention-based pulse sequence acquires with each echo train for each of the m simultaneously refocused slices two Cartesian k-space segments, each of which involves the k-space center. The k-space segments acquired in different echo trains are respectively rotated against one another about the k-space center.

(103) Preferably, in a PROPELLER/BLADE variation, the above-mentioned complex combination of both echo groups is performed segment by segment. Here the term segment involves the data that are read after a single excitation pulse. In the PROPELLER/BLADE variation, each k-space segment represents a Cartesian sub-space by means of which the above-mentioned algorithm can be applied. The flow diagram shown in FIG. 7 displays the modified PROPELLER reconstruction for the first slice. The reconstruction for the other slices is performed in the same manner. At the same time, changes compared to a conventional PROPELLER reconstruction are respectively marked with a dash-dotted frame.

(104) As in prior art, the different slices are reconstructed independently from one another. Therefore, the representation shows the reconstruction of a single slice. Contrary to prior art, part of the propeller blades of the slice are acquired twice in different echo groups. It is the goal of modified reconstruction to combine the doubly acquired propeller blades in the same direction according to a number of procedural steps with the result that, like in prior art, exactly one segment data set is available for each direction and the remaining procedural step can be performed in conventional manner.

(105) Usually, a PROPELLER reconstruction starts with several procedural steps which, respectively, operate only based on the data of a segment.

(106) If parallel reconstruction technology with several receiving coils has been used, the respective lines of the segment data set not acquired (for example, of the coil sensitivities of the single coils) are substituted in the procedural steps P.Ia, P.Ib. In the simplest case, this procedural step does not differ from the respective procedural step in conventional PROPELLER reconstruction. Optionally, the dual presence of the data set can be used advantageously, for example, for achieving an improved signal-to-noise ratio, for reducing remaining artifacts or for saving computing capacities.

(107) Propeller blades of a specific slice with the same rotation angle, which are acquired twice in two scanning windows, can subsequently be combined in complex-valued manner in step P.III, after the slowly varying phase in the image space has been removed mathematically in steps P.IIa or P.IIb. For details of the procedural steps P.IIa, P.IIb and P.III, see the method described above by means of FIG. 6. The only difference is that the operations are performed on the bass of individual segment data sets and not on the complete doubly acquired k-space data set of a slice.

(108) After the complex-valued combination of the doubly acquired propeller blades, for each direction a complete, phase-corrected propeller blade segment data set B.sub.1x is available for each adjustment (rotation angle of the propeller blade). Consequently, the remaining procedural steps can be performed in the manner. of conventional PROPELLER reconstruction. Here, these remaining procedural steps comprise an optional motion detection (step P.IV), a density compensation (for example, in step P.V) and finally a combination of propeller blades with different adjustment in the k-space, a final two-dimensional Fourier transformation in the image space and further optional steps, for example, filter operations (all steps represented mutually by the block step P.VI). Usually, the combination of the propeller blades with different adjustment is implemented as so-called gridding operation. Optionally, as described in DE 10 2005 046 732, this step can also be implemented as rotation with subsequent accumulation. Details of conventional PROPELLER reconstruction are included in the above-mentioned journal article by James Pipe.

(109) Density compensation is advantageous because the central regions of the k-space are repeatedly acquired by different propeller blades, while the peripheral regions are usually acquired only once.

(110) The subsequent description includes further preferred embodiments.

(111) In turbo spin echo imaging, short echo spacing usually has a positive effect on the image quality. In the invention-based method, the number of scanning windows per refocusing pulse has been increased by 2 m compared to a conventional one-slice turbo spin echo sequence which (as shown in FIG. 2) has only one scanning window per refocusing pulse. In order to be still able to realize short echo spacing, it is preferred to use the invention-based pulse sequence with a large readout gradient so as to traverse in readout direction the k-space to be acquired in the shortest period possible. However, the maximum gradient amplitude A.sub.max is technically limited by the gradient system of the magnetic resonance system. Furthermore, in the invention-based pulse sequence, in the period T.sub. between two consecutive slice excitation pulses .sub.1, .sub.2, the same gradient moment C=T.sub.A.sub.GRO as between two consecutive echo signals has to be switched in readout direction. However, the time available for this process is shorter by the duration of a slice excitation pulse .sub.1, .sub.2, than the time period T.sub.. Consequently, the maximum readout gradient is always lower than the maximum amplitude A.sub.max of the gradient system and can be selected the closer to the maximum amplitude A.sub.max the shorter the duration of an excitation pulse .sub.1, .sub.2. Therefore, in a preferred embodiment, in consideration of the maximum B.sub.1 amplitude to be realized by the radio-frequency transmission system of the magnetic resonance imaging system and in consideration of SAR limitations, the duration of a slice excitation pulse .sub.1, .sub.2 is selected to be as short as possible. Because of the fact that the tilting angle of 90 to be implemented by an excitation pulse .sub.1, .sub.2 usually is smaller than the tilting angle of a refocusing pulse , , , , , it is usually possible, when a maximum .sub.B1 amplitude is supplied, to select the duration of an excitation pulse .sub.1, .sub.2 particularly shorter than the duration of a refocusing pulse , , , , .

(112) In conclusion, it should be noted that the previously described detailed methods and designs involve embodiments and the basic principle can be applied by an expert in a variety of areas without leaving the range of the invention to the extent provided by the claims.

(113) For example, by means of a sequence of readout gradients with alternating amplitude, such as an EPI sequence, it is possible to form several echoes per echo group and these, like in a GRASE method (gradient and spin echo method, as described in GRASE (Gradient and Spin Echo) Imaging: A Novel Fast MRI technique; Magnetic Resonance in Medicine, 20, 1991, pp. 344-349), separately encode phases for a reduction of acquisition time. Alternatively, the temporal interval of the readout gradient of an echo group can be selected in such a way that it is possible to achieve a desired phase shifting between water and fat components of the read signal. From the different images of an echo group thus obtained, it is possible with the aid of so-called Dixon reconstruction to reconstruct images which respectively represent only the fat components or only the water components of the tissue to be examined.

(114) The inventive pulse sequences are able to maintain a sufficiently long echo train for fast T2-weighted imaging even when flip angle of the refocusing pulses is significantly reduced compared toward 180. This is especially advantageous when used in high-field systems with a basic magnetic field of 3 tesla or more in order to achieve an adequate reduction of SAR exposure with a moderate slice number m (and thus a moderate extension of echo spacing). Therefore, for SAR reduction, the inventive sequence is often preferably used with reduced flip angles of the refocusing pulses. Although all refocusing pulses in the figures are displayed in the same manner, in particular different refocusing pulses can also have different flip angles, for example, =180, =120, =120, . . . .

(115) The sequence is also compatible with so-called variable rate (VR) or variable rate selective excitation (VERSE) pulses by means of which it is possible to achieve a reduction of the radiated RF power by reducing the peak amplitude of the radio-frequency pulses compared to a respective SINC pulse.

(116) Although modifications and changes may be suggested by those skilled in the art, it is the intention of the inventor to embody within the patent warranted hereon all changes and modifications as reasonably and properly come within the scope of his contribution to the art.