阅读说明：本技术 磁共振指纹方法 (Magnetic resonance fingerprint method ) 是由 格雷戈尔·克尔德费 马蒂亚斯·尼特卡 庞佳宁 彼得·施派尔 于 2019-06-21 设计创作，主要内容包括：一种磁共振指纹(MRF)方法,包括：建立的体素时间序列的信号特性与第一比较信号特性(字典)的第一信号比较以及基于第一比较信号特性和第一信号比较中确定的值生成另外(合成的)比较信号特性。使用所生成的另外比较信号特性来执行通过其可以确定至少第一另外参数和第二另外参数的值的另一信号比较。通过生成另外比较信号特性,可以确定在第一信号比较中不一定已经确定的至少一个另外参数的值。在此期间,所需的处理能力保持较低。通过在另一信号比较的过程中确定第一另外参数和第二另外参数,该方法使得能够整体地确定另外参数的值,因此可以避免由其他另外参数引起的效应造成的另外参数的值的伪造。(A Magnetic Resonance Fingerprinting (MRF) method, comprising: the signal property of the established time series of voxels is compared with a first signal of a first comparison signal property (dictionary) and a further (synthesized) comparison signal property is generated based on the first comparison signal property and a value determined in the first signal comparison. The generated further comparison signal characteristic is used to perform a further signal comparison by which a value of at least the first further parameter and the second further parameter can be determined. By generating the further comparison signal characteristic, a value of the at least one further parameter may be determined which is not necessarily already determined in the first signal comparison. During this time, the required processing power remains low. By determining the first further parameter and the second further parameter during the comparison of the further signal, the method enables the value of the further parameter to be determined as a whole, so that falsification of the value of the further parameter by effects caused by the other further parameter can be avoided.)

1. A method for determining parameter values in voxels of an examination object using a magnetic resonance fingerprinting, MRF, technique, comprising the steps of:

establishing at least one voxel time series (BZS) by means of an MRF acquisition method, wherein a respective value (P, P') of at least one parameter at a position in the examination object represented by a respective voxel is to be determined from the at least one voxel time series,

performing a first signal comparison (103) of at least one segment of a respective signal characteristic of the established voxel time series (BZS) with a corresponding segment of a first comparison signal characteristic (D) to determine a respective value (P) of at least a first parameter of the parameters to be determined,

generating a further comparison signal characteristic (D') based on the first comparison signal characteristic and a value determined in the first signal comparison,

performing a further signal comparison of at least one segment of the respective signal characteristic of the established voxel time series (BZS) with a corresponding segment of the generated further comparison signal characteristic (D') to determine a value of at least a first and a second further parameter, respectively, of the parameters to be determined,

outputting the values of the parameters to be determined for the respective voxels.

2. The method according to claim 1, wherein the voxel time series (BZS) is established such that the signal characteristics of the voxel time series show a dependency on the parameter to be determined.

3. Method according to one of the preceding claims, wherein the second further parameter has an influence on the first further parameter.

4. Method according to one of the preceding claims, wherein the second further parameter has a dependency on the first parameter, wherein the determined value of the first parameter forms a basis for generating the further comparison signal characteristic (D').

5. Method according to one of the preceding claims, wherein the value determined during the first signal comparison showing the parameter independent of the second further parameter is no longer determined during the further signal comparison.

6. The method of one of the preceding claims, wherein the generation of the further comparison signal characteristic (D') comprises: the first comparison signal characteristic is summed over a respective number of predetermined assumed possible variations (V) of the value of the first parameter in the respective voxels determined in the first signal comparison, in each case for the value of the first further parameter.

7. The method according to claim 6, wherein the predetermined assumption is predetermined to be variable based on a priori knowledge about the object under examination.

8. The method according to one of claims 6 or 7, wherein the predetermined hypothetical potential variations correspond to a statistical distribution, wherein the number of predetermined hypothetical potential variations corresponds to the number of different hypothetical widths.

9. Method according to one of the preceding claims, wherein during the first signal comparison at least one, in particular all, parameters from a group of parameters comprising main magnetic field, emission field, transverse relaxation and longitudinal relaxation are determined.

10. Method according to one of the preceding claims, wherein during the first signal comparison the value of the local main magnetic field is determined and the first further parameter is the transverse relaxation time T2and the second further parameter is a parameter dependent on the prevailing phase dispersion in the respective voxel.

11. Method according to one of the preceding claims, wherein during the first signal comparison a value of the longitudinal relaxation time T1 or a value of the transverse relaxation time T2 is determined and the first further parameter is the longitudinal relaxation time T1 or the transverse relaxation time T2and the second further parameter is a ratio between the longitudinal relaxation and the transverse relaxation.

12. A method according to any of the preceding claims, wherein a second further parameter whose value is determined in the further signal comparison is not determined in the first signal comparison.

13. A magnetic resonance system (1) comprising: a magnetic unit (3), a gradient unit (5), a high-frequency unit (7), and a control device (9) having a high-frequency transmit/receive controller (7') and a parameter value determination unit (15), wherein the control device (9) is intended to perform the method according to one of claims 1 to 12 on the magnetic resonance system (1).

14. A computer program which can be loaded directly into a memory of a control device (9) of a magnetic resonance system (1), the computer program having programming means to carry out the steps of the method according to one of claims 1 to 12 when the program is executed in the control device (9) of the magnetic resonance system (1).

15. An electronically readable data medium storing electronically readable control information, wherein the control information comprises at least one computer program according to claim 14 and is in the form of: the control information performs the method according to one of claims 1 to 11 when the data medium is used in a control device (9) of a magnetic resonance system (1).

Technical Field

The invention relates to a magnetic resonance fingerprinting method for improving the determination of local parameter values of an examination object.

Background

Magnetic resonance technology (MR stands for magnetic resonance in the following) is a known technique by which images of the interior of an examination subject can be generated. In short, for this purpose the examination subject is located in a magnetic resonance apparatus in a relatively strong, stationary, homogeneous main magnetic field (also referred to as B)₀Field) with a field strength of 0.2 tesla to 7 tesla and above, so that its nuclear spins are obtained in the direction of the main magnetic field. For triggering nuclear spin resonances, the examination subject is irradiated with high-frequency excitation pulses (RF pulses), the triggered nuclear spin resonances are measured as so-called k-space data, and on the basis thereof the MR images are reconstructed or the spectral data are identified. For spatial encoding of the measurement data, rapidly switched magnetic gradient fields are superimposed on the main magnetic fields and these magnetic fields lie along the track, along which measurement data in k-space are read out. The captured measurement data is digitized and stored as complex values in a k-space matrix. The associated MR image can be reconstructed from the k-space matrix populated with values, for example by means of a multidimensional fourier transform. A series of RF pulses to be transmitted, the gradients to be activated and the readout process used for this purpose and ordered in a specific way are specified as a sequence.

Various sequence types are known which have different sensitivity levels for parameters describing the material in the examination object under examination, such as longitudinal relaxation time T1, transverse relaxation time T2and proton density. A map of the examination object is represented in a sensitivity-weighted manner according to the type of sequence used from an MR image reconstructed from measurement data acquired using a specific type of sequence.

Magnetic resonance imaging produced using a magnetic resonance system can be used to determine the presence and/or distribution of materials in an examination object. The material may be, for example, a (possibly pathological) tissue of the examination object, a contrast agent, a marker substance or a metabolite.

Here, information about the material present can be obtained from the captured measurement data in various ways. A relatively simple source of information is, for example, image data reconstructed from the measurement data. However, there are also more complex methods which identify information about the object under examination, for example from a time series of voxels of image data reconstructed from a set of continuously measured measurement data.

By means of quantitative MR imaging techniques, absolute properties of the measured object, such as tissue-specific T1 and T2 relaxation times in the case of a patient, can be determined. In contrast, conventional sequences, which are typically used in clinical routine, only generate relative signal intensities (so-called weights) for different tissue types, with the following results: diagnostic interpretation is highly dependent on subjective assessment by the radiologist. Thus, quantitative techniques offer the obvious advantage of being objectively comparable, but are difficult to use in the current day-to-day setting due to their long measurement times.

Relatively new quantitative measurement methods, such as the magnetic resonance fingerprinting method (MRF method), can reduce the disadvantages of said long measurement times to acceptable levels. In the MRF method, for each voxel (voxel) or at least a voxel of interest in the image data, the signal properties of the image data reconstructed from measurement data captured successively with different imaging parameters are considered to be a voxel time series, wherein the signal properties of the voxel time series are considered to be a "fingerprint" of the parameters prevailing in the voxel position of the examination object during the measurement. These signal properties are compared with signal properties in a previously identified database of signal properties which are properties of specific materials (so-called "dictionary") using a pattern recognition method in order to identify in the rendered examination object the material represented in the image data reconstructed from the measurement data, or the spatial distribution of tissue-specific parameters (e.g. transverse relaxation time T2 or longitudinal relaxation time T1 — so-called T1 and T2 maps). In this case, the signal characteristics in such a dictionary may also be generated by simulation.

The principle of the method is therefore to compare the measured signal characteristic with a plurality of previously known signal characteristics. Here, signal characteristics of different combinations of T1 and T2 relaxation times, and other parameters, may be identified for the dictionary. For each parameter to be determined there is a corresponding "dimension" in a so-called dictionary, in which different parameter values of the respective parameter are included in order to make different comparison values available. Then, parameter values, e.g. T1 and T2, of voxels (which may be pixels or voxels) in the image are determined, in particular by comparing the measured signal characteristics with all or some of the analog signal characteristics. This process is called "matching". The signal properties in the dictionary that are closest to the measured signal properties determine the parameters of the corresponding voxel, e.g. the relaxation parameters T1 and T2.

In addition to the transverse relaxation time T2 (also called spin-spin relaxation), an effective transverse relaxation time T2, which is shorter than the transverse relaxation time T2, is also known. The effective transverse relaxation time T2 additionally takes into account local variations in the main magnetic field B0 and thus off-resonance effects, which result in a faster decay of the transverse magnetization. The relationship between the two transverse relaxation parameters T2and T2 may be expressed as follows:

where T2' is the decay of the transverse magnetization caused by local variations in the main magnetic field B0, and thereby by phase dispersion.

In principle, in addition to the above-mentioned tissue-specific parameters of the object under examination, measurement-specific parameters can also be identified here, for example the field strength of the applied magnetic field or the local distribution of the strength of the actually irradiated high-frequency field B1+, since the signals captured by MR techniques may depend on the tissue-specific parameters in the object under investigation and the measurement-specific parameters describing the conditions prevailing during the measurement. The imaging parameters used in this case are selected such that the captured measurement data show a dependency on the desired parameter to be determined. For example, a sequence type that is sensitive to the desired parameter may be used for the MRF method. Due to the dependencies and variations in the acquisition parameters and the consideration thereof in comparing the signal characteristics, the desired parameters can be determined from the time series of voxels that have been acquired in this way.

For the MRF method, in principle any echo technique can be used in combination with any method for k-space scanning (e.g. cartesian, helical, radial).

The MRF method, which takes into account the tissue-specific parameters T1 and T2 in the dictionary used and determines them in the measured voxel time series, is described, for example, in the article "Magnetic Resonance imaging", Nature, 495: page 187 and 192 (2013). There, sequences based on TrueFISP (true fast imaging with steady state free precession) are used in conjunction with helical k-space scanning. US 20160061922a1 describes another MRF method.

In principle, any effects not taken into account in comparing signal characteristics in the dictionary, for example because they are not in the signal model from which the dictionary was generated, but have an effect on the signals captured when generating the voxel time series, may falsify the results of the MRF experiment, that is to say the determined parameter values. This effect may be more or less pronounced depending on the sequence used when generating the temporal sequence of voxels.

Such an effect which may have an effect on the signals measured in the voxel time series in MR experiments (MR measurements) is, for example, the signal attenuation due to partial resonance based and thus to intra-voxel phase dispersion. As described above, the effect of this effect is usually quantified by another relaxation time, the effective transverse relaxation time T2. To identify differences in phase dispersion for each voxel within the object, a sequence in which the signal response depends on T2 may be used, for example, to generate a temporal sequence of voxels. Since T2 is assumed to be exponentially decaying (e.g., another assumption that exploits the gaussian distribution of off-resonance within voxels), the signal decay can be scanned, for example, using a T2-dependent multi-echo FLASH sequence. Using the voxel time series captured in this way, and the corresponding comparison signal characteristics depending on T2, the parameters of T2 time can be determined by conventional signal comparisons in the MRF method.

Conversely, if possible effects of the parameters, such as the phase dispersion (also called intra-voxel phase shift), are not taken into account, as is the case, for example, in the cited article by Ma et al, this may also falsify the results of determining the values of the other parameters. For example, in Chiu et al, the article "Effects of RF pulse profile and intra-voxel phase dispersion on MRfinger printing with balanced SSFP readout", Magnetic Resonance Imaging 41, pages 80-86, 2017, the dependence of MRF results on intra-voxel dispersion is described based on the MRF implementation in the cited article by Ma et al.

In order to reduce this dependency it is necessary to reduce this dependency,the article "Pseudo step-State Free preprocessing for MR phasing", Magnetic Resonance in Medicine 77, page 1151-1161, 2017 proposes an MRF implementation aimed at reducing said sensitivity to phase shifts within voxels by changing the sequences used in the generation of the voxel time sequence.

To generate a temporal sequence of voxels, other MRF implementations use such a sequence: wherein despite this effect of intra-voxel phase shifts occur, they are relatively small. Examples are MRF implementations Using FISP (fast imaging with Steady State Precession) sequences, or MRF implementations Using FLASH (fast small angle excitation) sequences, described by Jiang et al in the article "MR imaging Using fast imaging with Steady State preprocessing" (FISP) with Spiral Readout ", Magnetic Resonance in Medicine 74, page 1621 1631, 2015. With these sequence types, in each repetition time TR, the transverse magnetization is completely destroyed ("destroyed"). Thus, the effect of phase dispersion is limited to faster signal attenuation. In this way, the influence of phase dispersion can also be kept constant, for example by keeping the echo time TE constant during the sequence, so that MRF results are at best subject to low-level falsifications.

An article by Wang et al, "In vivo Simultaneous Measurement of delta f, T1, T2and T2 by Magnetic Resonance imaging with a generalized RF Phase", Proc. int. Soc. Mag. Reson. Med.25, p. 3960, 2017Another MRF method is described in which each voxel is assigned a phase dispersion in one matching type. There, the T1, T2and B0 (delta) were determined_f) After the parameter values of (a), the phase dispersion is determined in a further matching step. For another matching step, signals in the B0 dimension from the dictionary are weighted and summed using different widths of lorentz distributions centered around the value determined for B0. Another matching step and the signal obtained therefrom are used to identify the "best fit" width and use it as a measure for phase dispersion. In a further matching step, the determined parameter values of T1, T2and B0 are retained, wherein only phase dispersion (over the entire width) is determined.

Disclosure of Invention

It is an object of the invention to enable an improved determination of parameter values using MRF methods.

This object is achieved by a method for determining parameter values in voxels of an examination object using a Magnetic Resonance Fingerprinting (MRF) technique according to claim 1, a magnetic resonance system according to claim 13, a computer program according to claim 14 and an electronically readable data medium according to claim 15.

A method according to the invention for determining parameter values in voxels of an examination object using a Magnetic Resonance Fingerprinting (MRF) technique comprises the following steps:

establishing at least one voxel time series (BZS) by means of an MRF imaging method, wherein a respective value (P, P') of at least one parameter at a position in the examination object represented by a respective voxel is to be determined from the at least one voxel time series,

performing a first signal comparison (103) of at least one segment of a respective signal characteristic of the established voxel time series (BZS) with a corresponding segment of a first comparison signal characteristic (D) to determine a respective value (P) of a first parameter of the parameters to be determined,

generating a further comparison signal characteristic (D') based on the first comparison signal characteristic and a value determined in the first signal comparison,

performing a further signal comparison of at least one segment of the respective signal characteristic of the established voxel time series (BZS) with a corresponding segment of the generated further comparison signal characteristic (D') to determine respectively (independently) the values of at least a first and a second further parameter of the parameters to be determined,

the values of the parameters to be determined for the respective voxels are output.

By creatively generating the further comparison signal characteristic from the provided set of comparison signal characteristics using the parameter value determined in the first signal comparison, the value of the at least one further parameter, which does not necessarily have to be determined in the first signal comparison, may be determined. Thus, in particular, it may not be possible to determine in a first signal comparison a second further parameter whose value is determined in another signal comparison. This not only increases the number of determinable parameters, but at the same time keeps the processing power in the further signal comparison small, since the further comparison of the signal properties does not require the "actual" further dimension for determining the further parameters, but, as it were, synthetically converts the dimension of the parameter determined in the first signal comparison into the dimension of the further parameter. Additional (simulated) comparison signal characteristics generated from a class of sub-dictionaries that avoid the "real" extra parameter dimension (for another parameter) and thus significantly reduce the required processing power compared to using such an "real" other parameter dimension.

By determining the first further parameter and the second further parameter during the comparison of the further signal, the method enables the values of at least these two further parameters to be determined in their entirety, so that falsification of the value of one parameter due to effects caused by the values of the other parameters can be avoided. This is particularly useful first if the second further parameter has an influence or a dependency on the first further parameter, and vice versa as the case may be.

During the first signal comparison, the value of the local main magnetic field may be determined and as the first and second further parameters, a parameter dependent on the prevailing phase dispersion in the respective voxel, for example the phase dispersion itself, of the transverse relaxation time T2 may be selected, so that falsification of the determined value of the transverse relaxation time T2 by the effect of the phase dispersion may be avoided. Since the value of the facies dispersion in the voxels so determined may be associated with the susceptibility of the tissue represented in the voxels, the determination of the facies dispersion value may also be used to provide an answer to a diagnostic question.

During the first signal comparison, a value of the longitudinal relaxation time T1 or a value of the transverse relaxation time T2 may also be determined, and the first further parameter may be the longitudinal relaxation time T1 or the transverse relaxation time T2, and the second further parameter may be a ratio between the longitudinal relaxation and the transverse relaxation. By using another signal comparison to determine the prevailing ratio of longitudinal to transverse relaxations, the so-called "partial volume" effect can be avoided in particular.

In this way, the described method enables the use of different sequence types with different dependencies on different parameters as freely as possible. This makes it possible to determine the values of as many different parameters as possible with the MRF method as accurately as possible without having to accept falsification of certain parameter values due to the effect of other parameters that have an effect on the signal captured when generating the voxel time series because of the use of the sequence type, and without unduly increasing the processing capacity.

The values of the parameters determined during the first signal comparison, which parameters exhibit no dependency on the second further parameter, can be "held" here during the further signal comparison, i.e. they are no longer determined. Since the absence of said dependency means that no influence on the value determined in the first signal comparison need be expected, in this way the processing capacity can be kept economically low again.

The magnetic resonance system according to the invention comprises a magnet unit, a gradient unit, a high-frequency unit and a control device which is intended to carry out the method according to the invention with a parameter value determination unit.

When the method according to the invention is executed on a control device, the computer program according to the invention implements it on the control device.

Here, the computer program may also be provided in the form of a computer program product which can be loaded directly into the memory of the control device and which has program coding means for carrying out the method according to the invention when the computer program product is executed in the processor unit of the processing system.

The electronically readable data medium according to the invention comprises electronically readable control information stored thereon, wherein the control information comprises at least one computer program according to the invention and takes the form of: such that the control information performs the method according to the invention when the data medium is used in a control device of a magnetic resonance system.

The specified advantages and embodiments relating to the method apply analogously also to the magnetic resonance system, the computer program and the electronically readable data medium.

Drawings

Further advantages and details of the invention will become apparent from the exemplary embodiments described below and the accompanying drawings. The examples given do not represent any limitation of the invention. In the drawings:

figure 1 shows a schematic flow diagram of a method according to the invention,

FIG. 2 shows a possible acquisition scheme of a time series of voxels, an

Figure 3 shows a schematically shown magnetic resonance system according to the invention.

Detailed Description

Fig. 1 is a schematic flow diagram of a method according to the invention for determining parameter values in voxels of an examination object using a Magnetic Resonance Fingerprinting (MRF) technique.

At least one voxel time series (BZS) is created by means of the MRF method, from which a respective value (P, P') of at least one parameter at the position of the examination subject represented in the respective voxel is determined (block 101). Here, the voxel time series BZS is in particular established such that the measured signal characteristics of the voxel time series show a dependency on the parameter to be determined, that is to say a change of the value of the parameter to be determined will result in a change of at least some of the measured signals in the characteristics.

A first signal comparison of at least one segment of the respective signal characteristic of the established voxel time series BZS with a corresponding segment of a first comparison signal characteristic D is performed in order to determine a respective value P of at least a first parameter of the parameters to be determined (block 103).

During the first signal comparison, for example, at least one parameter, in particular all parameters, from the group of parameters comprising the main magnetic field B0, the transmit field B1, the transverse relaxation time T2and the longitudinal relaxation time T1 may be determined.

For this purpose, the method may be used in the context of, for example, an MRF method for determining T1, T2, B1 and B0 simultaneously, wherein the values of the parameters B1+, B0, T1 and T2 are identified in a single continuous measurement at each voxel. An example of a possible acquisition scheme for such an MRF method will be explained below with reference to fig. 2. In the capture scheme shown, TrueFISP type sequence components are used, since these components are characterized by good coding and high signal efficiency of the B0 parameter.

A further comparison signal characteristic is generated based on the first comparison signal characteristic D and the value P determined in the first signal comparison (block 105).

Here, the generation of the further comparison signal characteristic D' may comprise: the first signal characteristic is summed over a respective number of predetermined assumed possible variations (V) of the value of the first parameter in the respective voxels determined in the first signal comparison, in each case for the value of the first further parameter.

Here, the predetermined assumed possible change may be predetermined based on a priori knowledge of the object under investigation.

For example, in the first signal comparison, a value of the longitudinal relaxation time T1 or a value of the transverse relaxation time T2 may be determined, and the first further parameter may be the longitudinal relaxation time T1 or the transverse relaxation time T2, and the second further parameter is a ratio between the longitudinal relaxation and the transverse relaxation. Here, the ratio T1/T2 of possible tissue types may be predetermined to assume possible changes based on a priori knowledge of the tissue types in the examination object and typical ratios of its parameters, such as longitudinal relaxation time to transverse relaxation time T1/T2. Thus, the generated further comparison signal characteristic may serve as a synthetic dimension to change the possible ratio T1/T2, where the value of T1 or T2 has been determined from the first signal comparison. If in a further signal comparison the value of T1 or T2 is simultaneously determined as the first further parameter, the result can also be iteratively improved, where appropriate. Once another signal comparison has determined a "best match" and thus the value of the ratio T1/T2 and the value of T1 or T2, the value of another parameter, T2 or T1, is also determined.

Thus, for example during the first signal comparison, the value of the longitudinal relaxation time T1 may be determined and for the first and second further parameters the transverse relaxation time T2and the ratio between the longitudinal relaxation and the transverse relaxation may be determined.

For example, during the first signal comparison, the value of the transverse relaxation time T2 may also be determined, and for the first and second further parameters the longitudinal relaxation time T1 and the ratio between the longitudinal relaxation and the transverse relaxation.

Furthermore, for example during the first signal comparison, a value of the transverse relaxation time T2 may also be determined, and for the first and second further parameters may be the transverse relaxation time T2and the ratio between longitudinal relaxation and transverse relaxation. In this case, the value of the longitudinal relaxation time T1 can also be found from the determined value of T2and the ratio between the longitudinal relaxation and the transversal relaxation.

Furthermore, it is here also possible to determine the value of the longitudinal relaxation time T1, for example during the first signal comparison, and for the first and second further parameters the longitudinal relaxation time T1 and the ratio between the longitudinal relaxation and the transverse relaxation. In this case, the value of the transverse relaxation time T2 can also be found from the determined value of T1 and the ratio between longitudinal relaxation and transverse relaxation.

Additionally or alternatively, the predetermined hypothetical possible variations may correspond to a statistical distribution, wherein the number corresponds to a number of different possible hypothetical widths of the statistical distribution. This is mainly useful when the second further parameter depends on the first parameter, wherein the determined value forms the basis for generating the further comparison signal characteristic D', since the effect of the first parameter on the second further parameter resulting from said dependency in this way can be investigated.

The further comparison signal characteristic D' may be identified, for example, by summing the first comparison signal characteristic D. This can be done in particular in each case for the comparison signal characteristic D of these comparison signal characteristics D of the fixed value of the first further parameter, which corresponds to the fixed value of the first further parameter, for example T2, of which those values lying within a value range predetermined by the determined value of the first parameter and the respective width of the statistical distribution are summed. In this case, the value of the first parameter determined in the first signal comparison may be located, for example, in the center of the respective width. The corresponding summation may also include weighting according to a statistical distribution. The comparison signal properties D of the desired sum are thus summed so that weighting is performed according to the statistical distribution.

Here, the statistical distribution may be particularly a gaussian distribution or a lorentzian distribution or a true uniform distribution or the like. The appropriate distribution may be selected, for example, experimentally or based on an evaluation based on a priori knowledge of the relationship between the different parameters.

During the first signal comparison, for example, the value of the local main magnetic field B0 may be determined, and the first further parameter may be a transverse relaxation time T2, and the second further parameter may be a parameter dependent on the prevailing phase dispersion in the voxel, in particular the phase dispersion itself, the transverse relaxation time T2' caused by the change in the main magnetic field, or the effective transverse relaxation time T2, or one of its inverse.

For example, the predetermined assumption may vary and thus may be a different width of the statistical distribution centered about the value determined in the first signal comparison, e.g., the main magnetic field B0. The values of such widths prevailing in the voxels, determined by another signal comparison, correspond to the phase dispersion (off-resonance effect) in the voxels.

The local value of the attenuation of the transverse magnetization T2' caused by the local change of the main magnetic field can be described by the value of such width prevailing in the voxel determined by another signal comparison, since the attenuation of the transverse magnetization caused by the local change of the main magnetic fieldIs proportional to said width, designated for example as full width at half maximum FWHM:

in each case, a further signal comparison is carried out of at least one segment of the respective signal property of the identified voxel time series BZS with the corresponding segment of the further comparison signal property D 'generated in block 105, in order to determine the value of at least a first and a second further parameter P', respectively, of the parameters to be determined (block 107).

Since the values of the first further parameter and the second further parameter are determined simultaneously by using a further signal comparison of the comparison signal characteristics generated forming to some extent a sub-dictionary, the values of the first further parameter and the second further parameter are determined in their entirety, so that a forgery of the value of one further parameter by the value of a (previously unknown) further parameter can be avoided.

The values P and P' of the parameters to be determined for the respective voxels may be output, for example in the form of a parameter map, or for example stored for later use.

Thus, the described method may be used, for example, to minimize the falsification of MRF parameter maps caused by, for example, (T2) phase dispersion. Thus, in generating the voxel time series, it is also possible to use a sequence type in which the signal response is influenced by the T2-effect, as is the case, for example, in TrueFISP sequences, which generate a signal that significantly depends on such partial resonance effects within the stop band (or dark band), which may lead, for example, to erroneous T2 values. Thus, the method may be used with any MRF scheme, where the sequence type used shows the time-varying T2-dependency of the signal.

In particular, to avoid falsification of the parameter map due to the phase dispersion effect, the method may determine the value of the main magnetic field B0, and if appropriate also the values of the parameters T1, T2and B1, for example in a first signal comparison (block 103), wherein the first comparison signal characteristics D each comprise the dimensions of the parameters whose values were determined in the first signal comparison, but not the dimensions of the phase dispersed parameters. Since it is known that phase dispersion has little influence on the determination of the values of the parameters T1, B0, and B1 from MRF signal comparison (see the cited article by Chiu et al), the values determined for these parameters can be retained later. In a further signal comparison (block 107), using a further comparison signal characteristic (block 105) generated specifically for this purpose, the value of the parameter T2 can be determined (more precisely) and the phase dispersion can be determined (for the first time) as an independent parameter. The further comparison signal characteristics D' are generated (block 105) such that they (as the previous first comparison signal characteristic D) comprise the individual dimensions of the first further parameter in case of a dispersion of the changed phases, at T2, and taking into account the possible change of the signal by the changed second further parameter. For this purpose, in generating the further comparison signal characteristic D', a sub-dictionary representing the signal variation caused by phase dispersion may be calculated for each voxel, wherein for each value of the T2 dimension, for example, the first comparison signal characteristic is summed, weighted in the B0 dimension, which gives a comparison signal characteristic approximately corresponding to the comparison signal characteristic resulting from spin precession at different frequencies. The weighting is caused by, for example, an assumed distribution of off-resonance within the voxel (e.g., gaussian, lorentzian, uniform distribution, etc.). Here, the distribution is particularly centered around the center frequency of the corresponding B0 value determined in the first signal comparison. The different distribution widths result in signal characteristics of different (intra-voxel) facies dispersion values, which can therefore be taken into account when determining the parameter values.

The method described here can also be used as part of a step in an MRF method with a plurality of signal comparisons between the established time series of voxels and the comparison signal characteristic for the stepwise determination of the parameter values.

Fig. 2 schematically shows a capturing method for capturing at least one temporal sequence of voxels as it may be used in the method according to the invention. The illustrated example shows an acquisition method for establishing a time sequence of voxels, wherein three different sequence types from the group of sequence types TrueFISP (true fast imaging with steady-state free precession), FISP (fast imaging with steady-state free precession) and FLASH (fast small angle excitation) are used. The nature of the sequence types refers in particular to the sensitivity of the individual sequence types to tissue-specific and/or measurement-specific changes in the parameter. For example, the FISP sequence has little sensitivity to changes in the main magnetic field B0, whereas the truEFIP sequence is more sensitive to changes in the main magnetic field B0. FLASH sequences and FISP sequences are sensitive to local variations in the irradiated high frequency field B1 +.

In the example shown in fig. 2, the respective serial numbers of the image datasets captured in the time series are indicated on axis 26 and the different variables are respectively indicated on axis 27. As a first variable, a flip angle in ° from 0 ° at the origin to 90 ° at the axis point 28 is indicated. In the example shown, the axis 26 proceeds from the image dataset 1 to the image dataset 3000.

The 3000 image datasets are distributed over twelve segments 29, 30, 31, 32, 33, 34, 35, 36, 37, 38 and 39.

In the first segment 29, the flip angle used during the capture is indicated by a curve 40 of 200 image data sets, wherein a FISP sequence can be used for the capture in the segment 29. As described in relation to fig. 1, once an RF excitation pulse with a certain flip angle is applied, a complete image data set is captured, and then the next RF excitation pulse with the next flip angle is applied and another image data set is captured. In segment 29, FIG. 2 shows the sum sin²The half curves are distributed with corresponding flip angles. The maximum flip angle may be, for example, 24 ° and a constant phase may be used.

For example only, line 41 is drawn for the 100 th image dataset. The corresponding flip angle is the maximum flip angle of the curve 40.

In the second segment 30, in the example shown, 400 image datasets are acquired using different sequence types, such as TrueFISP sequences. In this case, the flip angles according to the curves 42 and 43 are used. In the case of curve 42, these reach 45 °, and in the case of curve 43 reach 72 °.

For example only, for segment 30, line 44 is again plotted for the flip angle for the fourth hundred image data sets. In this case, the flip angle is 1 °.

A particular feature in segment 30 is the use of two different phase cycles. When passing through the flip angle of curve 42, a phase period of 00 or no phase period is used, and when passing through curve 43, a phase period of 180 ° is used. The 00 phase cycle represents the fixed phase.

In the next segment 31, in curve 45, the capture of 450 image data sets using another sequence type, such as a FLASH sequence, indicates the flip angle. These are smaller than in the FISP or TrueFISP sequences and reach 6 deg.. Their distribution is also sin²And (4) distribution.

In addition to changing the flip angle, when the FLASH sequence is repeated, a phase cycle is applied to produce RF corruption. This increases the phase by a factor of 117 deg., as described above.

Together, the sequences of the different sequence types used in fragments 29, 30 and 31 form box 55. This is used a total of three times in figure 2. Here, only the type of sequence is referenced and not the number of image data sets or the flip angle curve.

In the segment 32, 200 image data sets are again captured using the first sequence type of box 55, i.e. for example a FISP sequence. As in segment 29, the phase is constant, but the maximum flip angle is 45 °. The flip angle used lies on the curve 46.

This is followed by 200 image datasets acquired in segment 33 using a second sequence type of box 55, such as the TrueFISP sequence. Here, a 90 ° phase cycle is used, and the maximum flip angle is 50 °. The flip angle is indicated on the curve 47.

As in segment 31, in segment 34, the next approximately 450 image data sets are captured using the third sequence type of block 55, e.g., a FLASH sequence. Curve 48 shows sin with a maximum of 14 deg.²And (4) distribution.

When the first sequence type of the box 55, for example a FISP sequence, is used for the third time, the curve 49 in the segment 35 reaches 72 ° and shows the flip angle of the high-frequency pulse 19. In this process, the phase is also constant.

In acquiring another 200 image data sets using a second sequence type of box 55, e.g., TrueFISP sequence, a 270 phase cycle is used. The flip angle plotted in the curve 50 in the segment 36 amounts to 65 °.

The next approximately 450 image datasets in the fragment 37 are captured using a third sequence type, e.g., FLASH sequence, of box 55. Curve 51 shows a flip angle curve extending up to 20 deg., again in sin²And (4) distribution.

In the last segment 38, there are two curves 52 and 53 for capturing an image data set with a first sequence type of box 55, e.g. a FISP sequence. These again represent the flip angle curves. As in the previous section above, a constant phase is used, for example, in capturing measurement data using a FISP sequence.

The example shown allows a plurality of tissue-specific and measurement-specific parameters to be determined in the matching step by their presentation in segments of different sequence types with different sensitivities; in particular, the parameters T1, T2and B0 and B1+ may be determined. However, the examples shown should not be considered as limiting. In principle, it is also possible to create a voxel time series with only one type of series. However, the use of a plurality of different sequence types of different features, in particular with reference to a respective sensitivity to tissue-specific and/or measurement-specific parameters, increases the number of parameters that can be determined from the time series of voxels captured in this way and/or the quality of the values of the parameters determined in a manner corresponding to the features of the sequence type used.

The number of illustrative data sets shown captured with one sequence type and the flip angle curve shown should also be considered as examples only.

For the purpose of illustrating the compensation process, the spatial distribution of the particular main magnetic field value B0 is shown in the examination object as an example. During the recording of the measurement data from which image data have been reconstructed, the signal characteristic of each voxel should be compared as a time series of voxels with the comparison signal characteristic to determine at least a local value of the main magnetic field B0. A linear characteristic of the basic magnetic field B0 "from top to bottom" is generated in the examination object to clarify the effect.

The spatially resolved distribution of the main magnetic field values (B0 diagram) in the examination object shown on the left corresponds to the main magnetic field values determined voxel by signal comparison of a time series of voxels with comparison signal characteristics of a dictionary, which comparison signal characteristics cover a smaller range of B0 values than the B0 values present in the examination object. The B0 diagram shown on the left therefore covers B0 values, for example in the range-40 to +40, which are predetermined in any desired unit by the type of recording of the measurement data below the voxel time series and the corresponding comparison signal characteristics (dictionary).

The difference B0 map shown in the center corresponds to the difference image of the B0 map shown on the left, where the B0 map has a coarse spatial resolution, which consists of B0 values derived from a specific B0 sensitive portion of the voxel time series, and thus covers the B0 values in infinitely small steps. It is possible to determine a coarsely resolved B0 map from parts of the voxel time sequence from which measurement data has been recorded with a sequence type that is a property of the FISP sequence, using a smoothing operation, by determining phase differences of the signals in the parts of the voxel time sequence that have been recorded at different echo times.

If the difference B0 plot of the coarse spatially resolved B0 plot shown on the left and the B0 plot resulting from the signal comparison are rounded to a multiple of 1/TR, this results in segments that correspond to multiples of 1.TR (in the example shown, five such segments can be identified). If the rounded result is subtracted from the artifact-experienced B0 map (left), a compensated, deconvolved B0 map is obtained, as shown on the right. The B0 diagram shown on the right is a good representation of the linear behavior of the main magnetic field B0 artificially generated during the measurement even in the example shown in the range from-200 to +200 in the same units as in the B0 diagram shown on the left, by utilizing the processing described in the coarse B0 diagram over a larger range of values.

Here, the fact that: compensation data can already be obtained from the signals of the time series of voxels used without signal comparison, for example by an at least roughly resolved determination of the local arrival of the parameter to be determined, in particular for the measurement-specific parameter, for example the main magnetic field B0, which compensation data can be used for compensation, for example deconvolution, of the parameter values determined by the signal comparison. Although the resolution of such a coarsely resolved parameter map is not directly sufficient due to the local distribution of the parameter values, the parameter values of the coarsely resolved parameter map can be used as compensation data in order to compensate for the higher resolved parameters that have been determined by the signal comparison, e.g. deconvolution as in the example described above.

Thus, in a high resolution of the created voxel time series, a compensation value of the parameter determined for the voxel time series is determined, which may improve the value of the corresponding parameter previously determined and thus compensate for artifacts, wherein the range of values that may be achieved when determining the compensated parameter value may even be enlarged compared to the range of values that may be achieved with a pure signal comparison.

With the MRF recording method, which can determine the parameters T1, T2, B0 and B1+, an exemplary possible sequence of the method according to the invention for determining local parameter values for the parameters T1, T2, B0 and B1 with a multi-stage determination of local parameters can occur as follows.

After a first signal comparison between the established voxel time series and the comparison signal characteristic, wherein in particular the entire signal characteristic of the voxel time series is compared with the corresponding entire characteristic of the comparison signal characteristic, local values (maps) of all parameters T1, T2, B0 and B1+ can be determined.

Figure 3 diagrammatically shows a magnetic resonance system 1 according to the invention. This comprises a magnet unit 3 for generating a main magnetic field, a gradient unit 5 for generating gradient fields, a high-frequency unit 7 for transmitting and receiving high-frequency signals, and a control device 9 intended to perform the method according to the invention. In fig. 3, these individual elements of the magnetic resonance system 1 are shown only in a roughly schematic manner. In particular, the high frequency unit 7 may comprise a plurality of subunits, e.g. a plurality of coils, e.g. schematically shown coils 7.1 and 7.2 or more, which may be intended for transmitting high frequency signals only or for receiving high frequency signals only, or for both.

For the investigation of an examination object U, for example a patient or indeed a phantom, the object can be introduced into the measurement chamber of the magnetic resonance system 1 on a table L. The slice S represents an exemplary target volume of the examination object from which measurement data are captured.

The control device 9 serves for controlling the magnetic resonance system and may in particular control the gradient unit 5 by means of a gradient controller 5 'and the high-frequency unit 7 by means of a high-frequency transmit/receive controller 7'. In this case, the high frequency unit 7 may include a plurality of channels on which signals can be transmitted or received.

The high-frequency unit 7 together with its high-frequency transmit/receive controller 7' is responsible for generating and irradiating (transmitting) a high-frequency alternating field in order to manipulate the spins in the region of the examination subject U to be manipulated (for example in the slice S to be measured). For this purpose, the center frequency of the high-frequency alternating field (also designated as B1 field) is set to the greatest possible extent such that it is close to the resonance frequency of the spins to be steered. The deviation of the center frequency from the resonant frequency is called off-resonance. To generate the B1 field, a current controlled using the high-frequency transmission/reception controller 7' is applied to the HF coil in the high-frequency unit 7.

Furthermore, the control device 9 comprises a parameter value determination unit 15, by means of which parameter value determination unit 15 a signal comparison according to the invention can be performed to determine the parameter value. The control device 9 as a whole is intended to perform the method according to the invention.

The processor unit 13 comprised in the control device 9 is intended to perform all processing operations required for the required measurements and determinations. The results required for this purpose or recognized in the interim can be stored in the memory unit S of the control device 9. Herein, illustrated elements are not necessarily to be understood as physically separate elements, but merely to indicate a subdivision into meaningful elements. However, this may also be implemented, for example, in fewer or even only one physical unit.

Control commands can be sent to the magnetic resonance system and/or the results, for example image data, can be displayed on the control device 9, for example by a user via the input/output device E/a of the magnetic resonance system 1.

The methods described herein may also be provided in the form of a computer program product comprising a program and implementing the described methods on the control device 9 if it is executed on the control device 9. An electronically readable data medium 26 may also be provided on which electronically readable control information is stored, wherein the control information comprises at least a computer program product of the type just described and takes the form of: when the data medium 26 is used in the control device 9 of the magnetic resonance system 1, it carries out the described method.