Method and magnetic resonance system to generate multiple magnetic resonance images

Abstract

In a method and magnetic resonance system to determine multiple magnetic resonance images for respective different echo points in time, k-space is scanned on a segment-by-segment basis with at least two rectangular k-space segments, these being scanned line by line with respective k-space lines oriented parallel to one another. A short side of the rectangular k-space segments is oriented parallel to the k-space lines. First and second gradient echoes are respectively produced by a radio-frequency pulse radiated for each k-space line.

Claims

1. A method to generate multiple magnetic resonance (MR) images of an examination subject, comprising: operating an MR data acquisition unit, in which said examination is situated, according to a multi-echo MR data acquisition sequence to acquire raw MR data from the examination subject; entering said raw MR data acquired from the examination subject into an electronic memory organized as k-space comprising a plurality of rectangular k-space segments each comprising a plurality of k-space lines along which said raw MR data are entered; by operating said MR data acquisition unit with said multi-echo MR data acquisition sequence, producing, for each k-space line, a first gradient echo at a first echo point in time and a second gradient echo at a later second echo point in time; operating said data acquisition unit with said multi-echo MR data acquisition sequence by, for each k-space line, radiating a radio-frequency pulse that manipulates a transverse magnetization of nuclear spins in the examination subject, activating a phase encoding gradient field that phase codes the raw MR data of the respective k-space line, activating a first readout gradient field and a second readout gradient field that, in combination with the manipulation of the transverse magnetization produced by said radio-frequency pulse, respectively produce said first gradient echo and said second gradient echo, and reading out raw MR data, for the respective k-space line, of said first gradient echo during activation of said first readout gradient field in a time interval around said first echo point in time, and reading out raw MR data, for the respective k-space line, of said second gradient echo during activation of said second readout gradient field in a time interval around said second echo point in time; orienting said rectangular k-space segments in k-space with a longer side of each k-space segment oriented along a direction in the respective k-space segment that is defined by said phase encoding gradient field; orienting said rectangular k-space segments in k-space with a short side of each k-space segment oriented along a direction of the respective k-space segment that is defined by said readout gradient fields; and in a processor having access to said memory, reconstructing individual MR images from raw data in said k-space acquired at identical echo times, respectively.

2. A method as claimed in claim 1 comprising: entering said raw data into said k-space segments with said k-space lines oriented parallel to each other; orienting said long side of said k-space segments orthogonally to said k-space lines; and orienting said short side of said k-space segments along said k-space lines.

3. A method as claimed in claim 2, comprising: in said multi-echo MR data acquisition sequence, activating a slice selection gradient field during the radiation of said radiofrequency pulse; and shifting said k-space segments essentially parallel to said short side of said k-space segments in a plane defined by said slice selection gradient field.

4. A method as claimed in claim 1 wherein k-space comprises a k-space center, and comprising: organizing said k-space segments to all include said k-space center; and in said multi-echo MR data acquisition sequence, activating a slice selection gradient field during the radiation of said radiofrequency pulse, and rotating said k-space segments in a plane defined by said slice selection gradient field around said k-space center.

5. A method as claimed in claim 4 said k-space comprising exactly two of said k-space segments; and rotating said k-space segments to each other by an angle of approximately 90 around said k-space center in said plane defined by said slice selection gradient field.

6. A method as claimed in claim 1 wherein k-space comprises a k-space center, said method comprising: organizing said k-space segments in k-space to all include said k-space center; and rotating said k-space segments to each other around said k-space center in order to enter said raw data into a sphere in k-space.

7. A method as claimed in claim 1 comprising: organizing said k-space segments in three-dimensional k-space as cuboid k-space segments that each comprise a plurality of rectangular sub-segments; in said multi-echo MR data acquisition sequence, activating a slice selection gradient field during the radiation of said radiofrequency pulse, and activating an additional phase encoding gradient field that shifts said plurality of rectangular sub-segments to each other along a direction defined by said slice selection gradient field; orienting a long side of the plurality of sub-segments along a direction defined by the phase encoding gradient field or by the additional phase encoding gradient field; and orienting a short side of the plurality of sub-segments along a direction of the respective k-space segment defined by said first and second readout gradient fields.

8. A method as claimed in claim 7 wherein k-space comprises a k-space center, said method comprising: orienting said k-space segments in k-space so as to all include said k-space center; and rotating said k-space segments to each other in a plane defined by the readout gradient field and a plane defined by the phase encoding gradient field or the additional phase encoding gradient field.

9. A method as claimed in claim 7 wherein k-space comprises a k-space center, and said method comprising: organizing said k-space segments to all include said k-space center; and rotating said k-space segments to each other around said k-space center to enter said raw data into a sphere or a cylinder in k-space.

10. A method as claimed in claim 1 comprising forming said gradient echoes by operating said MR data acquisition unit with a spin echo/gradient echo hybrid sequence, as said multi-echo MR data acquisition sequence.

11. A method as claimed in claim 10 comprising, in said spin echo/gradient echo hybrid sequence, radiating said radio-frequency pulse as a refocusing pulse to generate a spin echo of said transverse magnetization, with said first echo point in time and said second echo point in time being within a time duration of said spin echo.

12. A method as claimed in claim 11 comprising: radiating said refocusing pulse as one of a series of multiple refocusing pulses that follow a radio-frequency excitation pulse to excite said transverse magnetization; and after one of said refocusing pulses in said series of multiple refocusing pulses, acquiring raw data for at least one of the k-space lines of at least one of said k-space segments such that all k-space data required for said at least one particular segment is acquired by this series of multiple refocusing pulses.

13. A method as claimed in claim 1 comprising, in said multi-echo MR data acquisition sequence, acquiring raw data from successive gradient echoes during activation of respective readout gradient fields with different polarity.

14. A method as claimed in claim 1 comprising reconstructing said MR images from raw data in said k-space segments acquired at identical echo times using a reconstruction technique selected from the group consisting of regridding in k-space, density compensation in k-space, successive shear operation in k-space, parallel imaging techniques, combining, for each echo time, multiple intermediate MR images respectively reconstructed from each k-space segment, or combining, for each echo time, MR data of multiple k-space segments and reconstructing images from the combined MR data.

15. A method as claimed in claim 1 comprising reconstructing said MR images from raw data of gradient echoes of said k-space segments acquired at identical echo times, according to a GRAPPA (Generalized Autocalibrating Partially Parallel Acquisition) technique.

16. A method as claimed in claim 1 comprising: operating said MR data acquisition unit with said multi-echo MR data acquisition sequence to produce a third gradient echo at a third echo point in time, which follows said first echo point in time and said second echo point in time; and acquiring raw data for a respective k-space line by activating a third readout gradient during a time interval around said third echo point in time.

17. A magnetic resonance (MR) apparatus, comprising: an MR data acquisition unit; a control unit configured to operate said MR data acquisition unit, in which an examination subject is situated, according to a multi-echo MR data acquisition sequence to acquire raw MR data from the examination subject; said control unit being configured to enter said raw MR data acquired from the examination subject into an electronic memory organized as k-space comprising a plurality of rectangular k-space segments each comprising a plurality of k-space lines along which said raw MR data are entered; said operation of said MR data acquisition unit with said multi-echo MR data acquisition sequence, producing, for each k-space line, a first gradient echo at a first echo point in time and a second gradient echo at a later second echo point in time; said control unit being configured to operate said data acquisition unit with said multi-echo MR data acquisition sequence by, for each k-space line, radiating a radio-frequency pulse that manipulates a transverse magnetization of nuclear spins in the examination subject, activating a phase encoding gradient field that phase codes the raw MR data of the respective k-space line, activating a first readout gradient field and a second readout gradient field that, in combination with the manipulation of the transverse magnetization produced by said radio-frequency pulse, respectively produce said first gradient echo and said second gradient echo, and reading out raw MR data, for the respective k-space line, of said first gradient echo during activation of said first readout gradient field in a time interval around said first echo point in time, and reading out raw MR data, for the respective k-space line, of said second gradient echo during activation of said second readout gradient field in a time interval around said second echo point in time; said control unit being configured to enter said raw data into said electronic memory with said rectangular k-space segments oriented in k-space with a longer side of each k-space segment oriented along a direction in the respective k-space segment that is defined by said phase encoding gradient field; said control unit being configured to enter said raw data into said electronic memory with said rectangular k-space segments oriented in k-space with a short side of each k-space segment oriented along a direction of the respective k-space segment that is defined by said readout gradient fields; and a processor having access to said memory, said processor being configured to reconstruct individual MRI images from said raw data in said k-space acquired at identical echo times, respectively.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) FIG. 1 schematically illustrates of an MR system constructed and operating in accordance with the invention.

(2) FIG. 2 illustrates a k-space segment in accordance with the invention.

(3) FIG. 3 shows a sequence for a k-space line of FIG. 2 according to different embodiments.

(4) FIG. 4 shows a sequence with monopolar readout gradients according to different embodiments;

(5) FIG. 5 shows a sequence with three bipolar readout gradients according to different embodiments.

(6) FIG. 6 depicts a short axis PROPELLER-like rotation of multiple k-space segments according to different embodiments.

(7) FIG. 7 depicts a short axis PROPELLER-like rotation of two k-space segments according to different embodiments.

(8) FIG. 8 depicts the shifting of three k-space segments along their short side according to different embodiments.

(9) FIG. 9 depicts a k-space segment according to different embodiments that comprises two parallel 2D sub-segments.

(10) FIG. 10 depicts a sequence scheme for the detection of gradient echoes for a 3D k-space segment.

(11) FIG. 11A depicts a sphere in k-space that is scanned via rotation and/or tilting of 3D k-space segments.

(12) FIG. 11B depicts a cylinder in k-space that is scanned via rotation and/or tilting of 3D k-space segments.

(13) FIG. 12 depicts a spin echo/gradient echo hybrid sequence according to different embodiments.

(14) FIG. 13 depicts a k-space segment that is undersampled in each line.

(15) FIG. 14 is a flowchart of a method to determine an MR image according to different embodiments.

DESCRIPTION OF THE PREFERRED EMBODIMENTS

(16) In the following, the present invention is explained in detail using preferred embodiments with reference to the figures wherein identical reference characters denote identical or similar elements.

(17) In the figures, techniques are explained that relate to a mufti-echo MR measurement (data acquisition) sequence in which gradient echoes are respectively formed for different echo points in time and MR images are determined based on the read-out gradient echoes. For example, by means of such MR images it can be possible to implement what is known as chemical shift imaging, thus for instance a separation of different spin species using the MR images that exhibit contrasts for the different echo points in time or, respectively, echo times.

(18) In particular, the techniques described in the following are characterized in that corresponding k-space lines of different contrasts are acquired at short time intervals, which enables a comparably high insensitivity to movement. In contrast to the n-echo per TR techniques known from the prior art, they simultaneously enable a higher isotropic resolution of the determined MR images. This occurs via the scanning of k-space with multiple k-space segments whose short side is respectively oriented along the k-space lines or, respectively, the direction kx defined by the respective readout gradient field. By scanning a large number of parallel k-space lines per k-space segment, a high resolution can inherently be achieved in the direction ky defined by the respective phase encoding gradient field. Via the combination of the MR data of multiple k-space segments that are shifted and/or rotated counter to one another, it is achieved that the MR images calculated from the superimposed data set are of high isotropic resolution.

(19) In FIG. 1, an MR system 100 is shown that is designed to implement techniques, methods and steps according to the invention. The MR system 100 has a magnet 110 that defines a tube 111. The magnet 110 can generate a basic magnetic field parallel to its longitudinal axis. An examination subject (here an examined person 101) on a bed table 102 can be slid into the magnet 110. The MR system 100 furthermore has a gradient system 140 to generate gradient fields that are used for MR imaging and for spatial encoding of acquired raw data. The gradient system 140 typically comprises at least three gradient coils 141 that can be controlled separately and positioned in a well-defined manner relative to one another. The gradient coils 141 enable gradient fields to be applied and switched along defined spatial directions (gradient axes). The corresponding gradient coils 141 are also designated as channels of the gradient system 140. A machine coordinate system of the MR system 100 can be defined by the windings of the gradient coils 141. The gradient fields can be used for slice selection, for frequency encoding (in the readout direction) and for phase encoding, for example. A spatial encoding of the raw data can thereby be achieved. The spatial directions (which are respectively parallel to slice selection gradient fields, phase encoding gradient fields and readout gradient fields) do not necessarily need to be coincident with the machine coordinate system. Rather, they can, for example, be defined in relation to a k-space trajectory which can in turn be established on the basis of specific requirements of the respective MR measurement sequence and/or be established based on anatomical properties of the examined person 101.

(20) An RF coil arrangement 121 that can radiate an amplitude-modulated RF excitation pulse into the examined person 101 is provided for excitation of the polarization resulting in the basic magnetic field or, respectively, alignment of the magnetization in the longitudinal direction. A transverse magnetization can thereby be generated. To generate such RF excitation pulses, an RF transmission unit 131 is connected via an RF switch 130 with the RF coil arrangement 121. The RF transmission unit 131 can comprise an RF generator and an RF amplitude modulation unit. The RF excitation pulses can flip the transverse magnetization slice-selectively in 1d or spatially selectively in 2D/3D, or globally, out of the steady state.

(21) Furthermore, an RF reception unit 132 is coupled via the RF switch 130 with the RF coil arrangement 121. MR signals of the relaxing transverse magnetization can be acquired via the RF reception unit 132 as raw data, for example via inductive injection into the RF coil arrangement 121.

(22) In general, it is possible to use separate RF coil arrangements 121 for the radiation of the RF excitation pulses by means of the RF transmission unit 131 and for the acquisition of the raw data by means of the RF reception unit 132. For example, a volume coil 121 can be used for the radiation of RF pulses and a surface coil (not shown) which comprises an array of RF coils can be used for the acquisition of raw data. For example, the surface coil for the acquisition of the raw data can comprise 32 individual RF coils and can therefore be particularly suitable for partially parallel imaging. Corresponding techniques are known to those skilled in the art, such that here no additional details need to be explained.

(23) The MR system 100 furthermore has a computer 160. For example, the computer 160 can be set up in order to control the acquisition of MR data within the scope of a multi-echo MR measurement sequence initiated via a user interface 150. Furthermore, the computer can be set up in order to implement a transformation of MR data from k-space into image space to determine an MR image. The computer can furthermore be set up to superimpose acquired MR data for the multiple k-space segments to determine the MR image.

(24) The use of the multi-echo MR measurement sequence means that, for the same k-space points or k-space regions, MR data are respectively acquired at different echo points in time, for example relative to an RF pulse that manipulates the transverse magnetization. Such a scenario is depicted in FIG. 2.

(25) A single k-space segment 200-1 of k-space 210 is scanned by entering raw (acquired MR data therein along multiple k-space lines 220 of the segment 200-1 as shown in FIG. 2. The direction kx (horizontal axis in FIG. 2) is parallel to the direction of the readout gradient field of the k-space segment 200-1. Orthogonal to this is the direction ky, which is oriented parallel to the direction of the phase encoding gradient field of the k-space segment 200-1. The k-space center 211 is located at the origin of the kx and ky coordinates. At the k-space center 211, kx=0 and ky=0.

(26) The spacing between adjacent k-space lines 220 amounts to k.sub.pe. The length of the k-space lines 220 amounts to Naka, wherein Na is the number of readout points and ka is the spacing between adjacent readout points, which is constant in the example of FIG. 2. As is clear from FIG. 2, the length of the k-space lines 229 determines the width 200a of the k-space segment 200-1. As is shown in the following, the length of the k-space lines 200 is limited by the necessity to detect gradient echoes for different and predetermined echo times or, respectively, echo points in time. The extent of the k-space segment in the phase encoding direction NPE*kPE (where NPE is the number of phase encoding steps) is subject to other limitations. For the applications relevant here, these limitations are normally smaller. This means that the extent in the phase encoding direction can be chosen to be greater than the extent in the readout direction.

(27) Therefore, the k-space segment 200-1 is rectangular, with the long side 200b of the k-space segment 200-1 oriented orthogonal to the k-space lines 220 and along the phase encoding direction ky. The short side 200a of the k-space segment 200-1 is oriented along the k-space lines 220 and along the readout direction kx.

(28) An inherently high resolution can therefore be provided along ky by scanning a correspondingly large number of k-space lines 220. Due to the limited width 200a of the k-space segment 200-1, the resolution in the kx direction is initially limited; however, this limited resolution in the kx direction can be corrected according to the invention via the acquisition of MR data for multiple k-space segments (not shown in FIG. 2).

(29) A simplified sequence scheme for a single k-space line 220 is depicted in FIG. 3. The radio-frequency signal 300 is depicted at the top in FIG. 3. The radiation of an RF pulse 401-1 takes place first. For clarity, the application of a slice selection gradient field is not shown in FIG. 2. The RF pulse 401-1 excites the transverse magnetization (RF excitation pulse), i.e. it deflects the longitudinal magnetization of the nuclear spins at least partially out of a steady state produced by the basic magnetic field.

(30) The application of a phase encoding gradient field 402 along the phase encoding direction 302 subsequently takes place. In FIG. 3, the k-space line 220 is scanned twice (double echo gradient echo sequence). For this purpose, a first readout pre-phasing gradient 403-1 is initially switched (activated) along the readout direction 303. The object of the readout pre-phasing gradient 403-1 is to compensate as precisely as possible (by virtue of is gradient moment) for the phase that the spins will acquire as a result of a readout gradient 403-2 activated around the desired first echo point in time 501 (corresponding to duration TE1) of the first gradient echo. The first readout gradient 403-2 is subsequently switched for frequency-encoding of the transverse magnetization. The readout gradient 404-2 forms the second gradient echo at the second echo point in time 502 (corresponding to duration TE2). Since both readout gradients 403-2, 404-2 have the same polarity (monopolar readout scheme) in the shown example, an additional gradient 404-1 is switched between the readout gradients 403-2, 404-2, the moment of which additional gradient 404-1 is chosen such that it compensates the phase which the pins acquire as a result of the first readout gradient 403-2 after the first echo time TE1 and before the second echo time TE2 as a result of the second readout gradient 404-2.

(31) At the bottom of FIG. 3, the k-space trajectory 230 during selected time intervals is illustrated by a thickened arrow. As can be seen from FIG. 3, the k-space line 220 is scanned from left to right during the readout gradients 403-2 and 404-2.

(32) Furthermore, time periods of the rising edge 403-2a, the falling edge 403-2c and the flattop 403-2b are depicted for the readout gradient 403-2 in FIG. 3. If data are read out during the complete duration 403-2b, the length of the k-space line 220 correlates with the moment of the readout gradient 403-2 during the time interval 403-2b. The greater this moment, the greater a spatial resolution that can be achieved by the MR data acquired for the k-space line 220. By means of specific techniques, it is also possible to acquire MR data for the edges 403-2a, 403-2c. The spatial resolution can thereby be additionally increased.

(33) The sequence scheme of FIG. 3 is presented in greater detail in FIG. 4 for multiple k-space lines 220. In particular, in FIG. 4 a first echo point in time 501 at which the first gradient echo is formed and a second echo point in time 502 at which the second gradient echo is formed are depicted. Time intervals 901, 902 are also depicted in which the gradient echoes are read out. A gradient echo always forms at those times t at which it applies for the entire accumulated zeroth gradient moment:

(34) $\begin{array}{cc}{m}_i\left(t\right)={}_{t0}^t{G}_i\left(\tilde{t}\right)\tilde{t}=0;i=R,S,& \left(1\right)\end{array}$
where R, S respectively designate the readout direction 303 and the slice selection direction 301. The start of integration t.sub.0 is what is known as the isodelay point in time of the RF pulse 401-1, which coincides in good approximately with the temporal middle of the RF pulse 401-1 given symmetrical, sinc-shaped RF pulses.

(35) In FIG. 4, three gradients 407a, 407b and 407c are switched in the slice selection direction 301. 407a is the slice selection gradient that is switched during the RF radiation. Gradient 407b has the task of compensating for the phase that the spins have accumulated as a result of the slice selection gradient. With regard to Equation 1: the moment in the slice selection direction that is accumulated between the isodelay time of the RF pulse and the end of the slice selection gradient field 407a is compensated by the slice selection gradient field 407b (slice refocusing gradient) so that m.sub.s(t)=0 for all times t between the end of the slice selection gradient field 407b and the beginning of a spoiler slice selection gradient field 407c (slice spoiling gradient).

(36) In the readout gradient field direction kx, the moment of the readout pre-phasing gradient 403-1 is selected such that it compensates for the moment acquired by the readout gradient field 403-2 at the desired first echo point in time 501.

(37) In the example of FIG. 4, the readout gradients 403-2, 404-2 have the same polarity and the same amplitudes. This selection is not mandatory, but can have advantageous effects given a later reconstruction of the MR image, for example since the shift between fat MR image and water MR image that are respectively obtained for the first echo point in time 501 or, respectively, the second echo point in time 502 is the same in the readout direction 303.

(38) It is also possible that the flattop duration of the readout gradient echoes 403-2, 404-2 is of identical length. For a readout channel 304, the time intervals 901, 902 are respectively illustrated for which gradient echoes are detected and converted from analog to digital in order to obtain sample points of the acquired MR data.

(39) The first echo point in time 501 is defined with regard to the isodelay point in time t0 (labeled with the echo time TE1 in FIG. 4). The second echo point in time 502 is accordingly defined with regard to the isodelay point in time t0 (labeled with the echo time TE2 in FIG. 4). The time difference between the first and second echo point in time 501, 502 is designated with TE. This time period also corresponds to the difference between the echo times TE1, TE2.

(40) A sequence scheme of an alternative embodiment of a multi-echo MR measurement sequence is depicted in FIG. 5. In the sequence scheme of FIG. 5, for a k-space line 220 following the RF pulse 401-1a count of three gradient echoes is detected via application of the bipolar readout gradient fields 403-2, 404-2, 405-2 (in FIG. 5, the three gradient echoes are represented by stars). The three gradient echoes are respectively formed at the echo points in time 501, 502, 503.

(41) While the gradient echo train comprises three gradient echoes in FIG. 5, in different embodiments it would be possible to detect only two gradient echoes or more than three gradient echoes, for example.

(42) Furthermore, in the embodiment of FIG. 5 an acquisition of the MR data takes place at the readout channel 304 and during the ramping of the readout gradient fields 403-1, 404-1, 405-1 (ramp sampling). A relatively higher spatial resolution thus can be achieved.

(43) A count of six k-space segments 200-1, 200-2, 200-3, 200-4, 200-5, 200-6 is depicted in FIG. 6. The k-space segment 200-1 that is discussed with regard to the preceding figures is emphasized by a dashed outline in FIG. 6. In FIG. 6, the direction kx of k-space coincides with the readout direction of the emphasized k-space segment. The same accordingly applies to the direction ky of k-space and the phase encoding direction of the marked k-space segment. The different k-space segments 200-1, 200-2, 200-3, 200-4, 200-5, 200-6 are respectively rotated by 30 relative to their neighbors and around a rotation axis including the k-space center 211, which rotation axis is parallel to a slice selection gradient field direction kz.

(44) Such segmentation of k-space 210 is also used in the aforementioned short axis PROPELLER techniques from U.S. Pat. No. 7,535,222. In contrast to U.S. Pat. No. 7,535,222, here the lines 220 of a k-space segment (propeller blade) are not filled with an EPI trajectory after an excitation pulse. The k-space trajectory 230 for acquisition of a k-space line 220 presently respectively starts in the k-space center 211 and does not successively traverse the k-space lines (see FIG. 3 versus FIG. 1b in U.S. Pat. No. 7,535,222). Moreover, presently each k-space line 220 is traversed multiple times in immediate succession, for example, to acquire different contrasts. The goal that is present in this exemplary embodiment is also different than the goal of U.S. Pat. No. 7,535,222; the teaching there serves to reduce distortions of an echoplanar (EPI) sequence by increasing the speed with which k-space is traversed in the phase encoding direction. One effect of different embodiments of the present invention is to increase the resolution of a multi-contrast sequence at predetermined echo times.

(45) In FIG. 7, a scenario is illustrated in which k-space 210 is scanned by two k-space segments 200-1, 200-2, wherein the two k-space segments 200-1, 200-2 are rotated counter to one another by an angle of 90 around the k-space center 211 in the plane defined by the slice selection gradient field (for example, this plane can have a normal vector or is oriented parallel to kz). In such a scenario, a particularly fast scanning of k-space 210 can take place. Furthermore, sample points 235 that are obtained by digitizing the detected gradient echoes as acquired MR data are depicted in FIG. 7.

(46) A scanning of k-space 210 with only two k-space segments 200-1, 200-2 enables diverse effects. In the phase encoding gradient field direction ky of one of the segments 200-1, 200-2, it applies that:
k.sub.PE=1/FoV.sub.PE=1/(N.sub.PEp.sub.PE) or, respectively, N.sub.PEk.sub.PE=1/p.sub.PE,(2)
where p.sub.PE is the spacing of adjacent lines in the phase encoding gradient field direction ky, and N.sub.PE is the number of phase encoding steps, and FoV.sub.PE designates the dimension of a field of view (FoV) in the phase encoding direction in image space 270 for the MR image 1000 (shown as an inset in FIG. 7).

(47) The smaller that the FoV is chosen to be along the phase encoding gradient field direction ky, the smaller the count of the phase encoding steps N.sub.PE that is necessary in order to realize a desired resolution for the MR image 1000.

(48) In Cartesian imaging, in reference implementations the phase encoding gradient field direction ky is therefore conventionally oriented along the minor [short] body axis of the examined person 101, and the actual field of view in the phase encoding gradient field direction ky (which includes phase oversampling) is chosen to the smaller than the field of view in the readout direction kx. Furthermore, in reference implementations in Cartesian imaging the resolution is often chosen to be smaller in the phase encoding gradient field direction ky than in the readout direction kx.

(49) However, in MR imaging using the aforementioned PROPELLER techniques a non-quadratic field of view can be disproportionately more difficult to realize with conventional techniques, and the efficiency gain can typically be smaller; see in this regard P. E. Larson and D. G. Nishimura Anisotropic Field-of views for PROPELLER MRI in Proc. Intl. Soc. Mag. Reson. Med. 15 (2007) 1726, for instance. In one embodiment with two k-space segments 200-1, 200-2 (as is depicted in FIG. 7), however, a rectangular field of view can be comparably simple to realize and indicate a large gain in efficiency.

(50) Based on two orthogonal directions x and y to be provided by the user; a (normally not quadratic) field of view that is specified by the extent FoVx along the x-direction and the extent FoVy along the y-direction; and a desired pixel size x in the x-direction or, respectively, y in the y-direction and a desired echo time difference TE, the readout gradient field direction kx of the k-space segment 200-1 is oriented along the x-direction and the phase encoding gradient field direction of the k-space segment 200-1 is oriented along the y-direction. It applies that:
FoV.sub.PE,1=FoV.sub.y,p.sub.PE,1=y,FoV.sub.RO,1=1FoV.sub.x,p.sub.RO,1=x
wherein FoV.sub.PE,1 designates the field of view in the phase encoding direction, and FoV.sub.RO,1 designates the field of view in the readout direction, respectively for the first k-space segment 200-1.

(51) The readout gradient field direction kx of the segment 200-2 is placed along the y-direction and the phase encoding gradient field direction of the second segment is placed along the x-direction, such that it applies that:
FOV.sub.PE,2=FOV.sub.x,p.sub.PE,2=x,FoV.sub.RO,2=2FoV.sub.y,p.sub.RO,2=y
wherein FoV.sub.PE,2 designates the field of view in the phase encoding direction, and FoV.sub.RO,2 designates the field of view in the readout direction, respectively for the second k-space segment 200-2.

(52) The factors 1 and 2 are optional additional readout oversampling factors whose value can be set greater than or equal to 1. Therefore, the k-space line spacing k.sub.PE,1, k.sub.PE,2 in the phase encoding direction of the two respective k-space segments 200-1, 200-2 is established with Equation 2. Moreover, the number of phase encoding steps or k-space lines 220 N.sub.PE,1 N.sub.PE,2 is established.

(53) The k-space spacing of two sample points in the readout direction kx is respectively defined as k.sub.RO,1, k.sub.RO,2 for the two k-space segments 200-1, 200-2, and is provided by:
k.sub.RO,i/FoV.sub.RO,i,i=1,2.(3)

(54) In contrast to known solutions, in embodiments according to the invention the number of sample points 235 in the readout direction N.sub.RO,1) N.sub.RO,2 is freely selectable and much smaller than the value calculated from the subsequent equation for Cartesian (or radial) imaging:
N.sub.RO,cartk.sub.RO,i=1/p.sub.RO,i,i=1,2.(4)

(55) For the number of sample points N.sub.RO in the readout direction kx, it thus applies that:
N.sub.RO,i<N.sub.RO,cart=1/(k.sub.RO,ip.sub.RO,i)=FoV.sub.RO,i/p.sub.RO,i,i=1,2(5)

(56) N.sub.RO,1 is preferably selected as large as possible so that the desired echo time difference TE can still be realized between successive echo points in time. The unsampled, peripheral k-space corners can be kept as small as possible in such a manner.

(57) A k-space grid spacing k.sub.x in the x-direction and a k-space grid spacing k.sub.y in the y-direction are subsequently established, and the (k.sub.RO,1, k.sub.PE,1) data matrix of the first segment with grid spacing k.sub.RO,1 in the x-direction and k.sub.PE,1 in the y-direction is interpolated as well as the (k.sub.PE,2, k.sub.RO,2) data matrix of the second segment with grid spacing k.sub.PE,2 in the x-direction and k.sub.RO,2 in the y-direction are interpolated on the (k.sub.x, k.sub.y) grid.

(58) Given suitable selection of the grid spacings, this interpolation can be implemented with a particularly precise and particularly efficient sinc interpolation. The MR data of the k-space segments are complexly added. The determination of the MR image can take place via a 2D FFT of the (kx, ky) data matrix. Due to the linearity of the Fourier transformation, the superpositioning of the MR data can take place before or after the Fourier transformation. Before or after the interpolation, in k-space 210 a density compensation is implemented the compensates for denser data sampling in the overlap region of the two segments. A phase correction and movement compensation of the two segments can also be implemented before the superposition.

(59) The coverage of k-space 210 with the two k-space segments 200-1, 200-2 and a non-quadratic field of view is illustrated in FIG. 7. In this example, FoV.sub.x=2FoV.sub.y, 1=1 and 2=2.

(60) An additional exemplary embodiment of the invention in which k-space 210 is scanned by means of three k-space segments 200-1, 200-2, 200-3 is depicted in FIG. 8. The k-space segments 200-1, 200-2, 200-3 are respectively rectangular. Each k-space segment 200-1, 200-2, 200-3 comprises a long side along the phase encoding gradient field direction ky and a short side along the readout gradient field direction kx of the respective k-space segment 200-1, 200-2, 200-3. The k-space lines 220 of each k-space segment 200-1, 200-2, 200-3 are traversed multiple times to detect multiple gradient echoes by means of the multi-echo MR measurement sequence described in the preceding. In contrast to this, the different k-space lines 220 of each k-space segment 200-1, 200-2, 200-3 are acquired given the use of a gradient echo sequence or spin echo sequence after different excitation pulses 401-1. Given use of TGSE hybrid sequence, different lines of a segment are acquired in proximity to different spin echoes.

(61) In contrast to the PROPELLER-based techniques as discussed above (for example with regard to FIGS. 6 and 7), the different k-space segments 200-1, 200-2, 200-3 are not rotated counter to one another around the k-space center 211. Rather, the three k-space segments 200-1, 200-2, 200-3 are shifted counter to one another essentially in parallel with the short side, i.e. in parallel with the readout gradient field direction kx. The shift of the k-space segments 200-1, 200-2, 200-3 along the readout gradient field direction kx is designed such that all k-space segments 200-1, 200-2, 200-3 together cover k-space 210 with the scope necessary to achieve a desired resolution.

(62) In the example of FIG. 8, the different k-space segments 200-1, 200-2, 200-3 partially overlap. However, it is not necessary that the k-space segments 200-1, 200-2, 200-3 partially overlap. A larger (smaller) degree of overlap can reduce (increase) the efficiency of the multi-echo MR measurement sequence. This is the case since MR data in the overlap region of the different k-space segments 200-1, 200-2, 200-3 are acquired multiple times. The repeatedly acquired MR data can be used to correct or alleviate artifacts of the MR images that could otherwise occur, for example as a result of a movement of the examined person between the detection of the gradient echoes for the different k-space segments 200-1, 200-2, 200-3.

(63) The shift of the k-space segments 200-1, 200-2, 200-3 along the readout gradient field direction kx can be achieved particularly simply via a corresponding selection of the readout pre-phasing gradient 403-1. An additional readout pre-phasing moment can be achieved via the specific dimensioning of the readout pre-phasing gradient 403-1. If the three k-space segments 200-1, 200-2, 200-3 in the example of FIG. 8 are acquired with the sequence from FIG. 3, for example, and if A is the integral of the readout gradient over the readout interval, a shift in k-space by an edge length of the k-space segment opposite to/along the readout direction is then achieved via an additional readout pre-phasing moment of minus/plus A to acquire the first/third k-space segment 200-1, 200-3. This shift would correspond to segments that specifically do not overlap. For overlapping segments, the magnitude of the additional readout pre-phasing gradient is accordingly selected to be smaller.

(64) In FIG. 9, a cuboid k-space segment 200 is depicted. The k-space segment 200 includes two rectangular 2D sub-segments 200aa, 200bb that are shifted counter to one another along the slice selection gradient field direction k-space, this is a 3D k-space segment 200. Furthermore, k-space lines 220 for the two sub-segments 200aa, 200bb are drawn in FIG. 9. The k-space lines 220 are again oriented along the short side of the rectangular sub-segments 200aa, 200bb or, respectively, along the readout gradient field direction kx. In general, techniques described in the preceding with regard to the 2D k-space segments can also be applied to the sub-segments 200aa, 200bb of a 3D k-space segment 200.

(65) In FIG. 10, a sequence scheme is depicted that enables the acquisition of MR data for the readout points 235 as they are depicted in FIG. 9. An additional phase encoding gradient field 402a to select one of the sub-segments 200aa, 200bb is switched along the slice selection direction 301. Depending on the selection of the additional phase encoding gradient field 402a, a greater or smaller number of sub-segments 200aa, 200bb can be realized and the spacing between two adjacent sub-segments 200aa, 200bb along the slice selection gradient field direction kz can be selected in a well-defined manner.

(66) It is possible to use the 3D k-space segments 200as they have been discussed in the preceding with regard to FIGS. 9 and 10by tilting and/or rotating and/or shifting within k-space 210 to cover said k-space 210. For example, the k-space segments 200 can be rotated counter to one another along the plane defined by the readout gradient field direction kx and either the first phase encoding direction ky or the slice selection direction kz. It would also be possible that the 3D k-space segments 200 (as they have been discussed in the preceding with regard to FIGS. 9 and 10) are rotated counter to one another around the k-space center 211 such that they scan a sphere 240 in k-space 210 (see FIG. 11A) or a cylinder 241 (see FIG. 11B).

(67) The detection of the gradient echoes at the different echo points in time 501, 502, 503 for a bipolar gradient echo train with the readout gradients 403-2, 404-1, 405-1 is depicted in FIG. 12. The detection of the gradient echoes takes place within the scope of a spin echo/gradient echo hybrid sequence. The gradient echoes (illustrated by stars in FIG. 12) are respectively detected at a refocusing pulse 401-2a, 401-2b, 401-2c. The portion of the spin echo/gradient echo hybrid sequence that is emphasized by the dot-dash line is repeated for different k-space lines 220. Upon repetition, the moment of the phase encoding gradient 402 is thus varied. Finally, a spoiler gradient is switched to dephase the remaining transverse magnetization (not shown in FIG. 12).

(68) If it is thereby achieved that all phase encoding lines of a k-space segment are acquired in one echo train, problems that result from a patient movement are thus reduced: the duration of one echo train is normally so short that movement that occurs during the acquisition of a k-space segment is frozen. The remaining movement that occurs between the acquisition of different k-space segments can be corrected (as is known from conventional PROPELLER imaging) or reduced by the weighting of the individual k-space segments or, respectively, leads to relative minor image artifacts in comparison to Cartesian imaging.

(69) For example, it is possible that the second echo point in time 402 can be temporally coincident with the spin echo formed by the respective refocusing pulse 401-2a, 401-2b, 401-2c. If the echo spacing TE between successive gradient echoes is thereby selected such that the phase evolution difference between fat and water amounts to 180, three contrasts with phase shift 180, 0, 180 are thus achieved. The respective second gradient echo 502 coincides with the spin echo and is thus in-phase; the two others, 501 and 503, are opposed in phase.

(70) From FIG. 12 it can be seen that the different refocusing pulses 401-2a, 401-2b, 401-2c are part of a series of multiple refocusing pulses that follow the RF excitation pulse 401-1. A pre-phasing readout gradient field 403-1 is applied between the RF excitation pulse 401-1 and the first refocusing pulse 401-2a. Instead of the bipolar gradient echoes 403-2, 404-1, 405-1, corresponding techniques are also possible with monopolar readout gradient fields.

(71) The undersampling of a k-space segment 200-1 is illustrated in FIG. 13. The dash-dot k-space lines 220 are not measured during the data acquisition. Undersampling means that the density of measured sample points 235 that is required according to the Nyquist theorem is not met. Techniques of ppa imaging can be used for substation of the unmeasured k-space lines 220 insofar as more acquisition coils and what are known as coil calibration data are present.

(72) Coil calibration data are normally the data of low-resolution images that are sufficiently sampled and that are acquired with the same acquisition coils. Various ppa techniques can be used to calculate the missing k-space lines 200, for example techniques operating in k-space such as GRAPPA or techniques operating in image space such as SENSE. Given what are known as auto-calibrating techniques, coil calibration data (ACS data) can be acquired via dense sampling, for example near the k-space center 211. This is schematically illustrated in FIG. 13 in that no missing k-space lines 220 are present near the k-space center 211.

(73) For example, it is possible that the ACS data are acquired only for a k-space segment 200-1, and are reconstructed from these ACS data for the additional k-space segments. For these additional k-space segments, the ACS data can be obtained with the aid of a gridding operation or with the aid of a share operation in k-space 210 via rotation of the acquired ACS data. Alternatively, the region near the k-space center 211 can be densely sampled for multiple or all k-space segments to acquire the ACS data.

(74) A flow chart of a method to determine MR images by means of a multi-echo MR measurement sequence is depicted in FIG. 14.

(75) The method begins in Step S1. First, a current k-space segment is determined in Step S2. In Step S3, an RF pulse is then radiated to excite the transverse magnetization. In Step S4, a current k-space line of the current k-space segment is selected, in particular by applying a phase encoding gradient field.

(76) The detection of the first gradient echo at a first echo point in time takes place in Step S5. The detection of a second gradient echo at a second echo point in time subsequently takes place in Step S6. Optionally, additional gradient echoes can subsequently be detected.

(77) In Step S7 a check is made whether an additional k-space line should be scanned for the current k-space segment. If this is the case, Steps S3-S7 are implemented again. Otherwise, in Step S8 a check is made whether an additional k-space segment should be scanned. If this is the case, Steps S2-S7 are implemented again. Otherwise, in Step S9 two MR images are respectively determined for the first and second echo point in time. The method ends in Step S10.

(78) Naturally, the features of the embodiments and aspects of the invention that are described herein can be combined with one another. In particular, the features can be used not only in the described combinations but also in other combinations or individually.

(79) 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.