Magnetic resonance imaging using CEST contrast enhancement

Abstract

Systems and methods for deriving chemical exchange saturation transfer (CEST) contrast images are presented. In one aspect according to the invention, the method comprises: acquiring signals by subjecting a portion of the body (10) to a CEST imaging protocol, the CEST imaging protocol comprising RF saturation at several different saturation frequency shifts; generating a Z-spectrum from the acquired signal for each voxel in a set of voxels from the magnetic resonance image; modeling a Z-spectrum background, the Z-spectrum background comprising contributions from direct water saturation, magnetization transfer and one or more fat signal components, where parameters of the model are determined by fitting the modeled Z-spectrum background to the generated Z-spectrum for each voxel, thereby omitting regions of the Z-spectrum affected by the CEST effect; and deriving a CEST contrast image based on the modeled Z-spectrum background and the Z-spectrum generated for each voxel. The system includes a processor configured to perform the method.

Description

TECHNICAL FIELD

[0001] The present invention relates to the field of magnetic resonance (MR) imaging. The present invention relates to a method of MR imaging of at least a portion of a body placed in a main magnetic field B0 within an examination volume of an MR system. The present invention further relates to an MR system and to a computer program for an MR system. BACKGROUND

[0002] MR methods of image formation utilize the interaction between magnetic fields and nuclear spins in order to form two-dimensional or three-dimensional images, which are widely used nowadays, especially in the field of medical diagnostics, because for soft tissue imaging they are superior to other imaging methods in many respects, do not require ionizing radiation, and are usually not invasive.

[0003] In some medical applications, the difference in MR signal strength from standard MR protocols (i.e. contrast between different tissues) can not be sufficient to obtain satisfactory clinical information. In this case, contrast enhancement techniques are applied. One known method of contrast enhancement and increasing the sensitivity of MR detection (by several orders of magnitude) is a method based on "chemical exchange saturation transfer" (CEST), as originally described by Balaban et al. (see, for example, US 6,962,769 B1). With this CEST technique, image contrast is obtained by changing the strength of the water proton signal in the presence of a contrast agent or endogenous molecules whose pool of protons resonates at a different frequency from the main water resonance. This is achieved by selectively saturating the nuclear magnetization of the exchangeable protons, which resonate at a frequency different from the water proton resonance. Exchangeable protons can be provided by exogenous CEST contrast agents (e.g. DIACEST, PARACEST or LIPOCEST agents), but can also be found in biological tissue (i.e. endogenous amide protons in proteins and peptides, protons in glucose or protons in metabolites such as choline or creatine). For this purpose, a frequency-selective saturation RF pulse is used that matches the MR frequency of the exchangeable protons. Saturation of the MR signal of the exchangeable protons is then transferred by chemical exchange with the water protons in the vicinity of the patient being examined into the MR signal of the water protons, thereby reducing the water proton MR signal. Selective saturation at the MR frequency of the exchangeable protons thus causes the creation of negative contrast in the proton density-weighted MR image.

[0004] Amide proton transfer (APT) MR imaging is a CEST technique based on endogenous exchangeable protons that allows highly sensitive and molecular level specific detection of pathological processes, such as an increase in protein concentration in malignant tumor tissue. APT signals also sensitively report locally altered pH levels (as the exchange rate depends on pH), which can for example be used to characterize ischemic stroke. APT / CEST MR imaging has multiple advantages over traditional MR contrast with or without the use of contrast agents. APT / CEST MR imaging allows highly specific detection and differentiation of exogenous and endogenous contrast, which is much more sensitive than spectral MR / NMR techniques. This high sensitivity (SNR efficiency) can be used to obtain molecular contrast information with comparable resolution to typical MR imaging applications in clinically acceptable examination times.

[0005] The signal attenuation at a particular saturation frequency offset compared to the MR signal obtained without RF saturation not only comes from the CEST effect, but also from direct saturation of water protons, and in addition, also from magnetization transfer from semi-solid macromolecules during in vivo imaging. It can be argued that direct saturation as well as magnetization transfer will cause signal attenuation that is symmetric with respect to the resonance frequency of water protons. Therefore, in conventional CEST MR imaging, the saturation transfer effect of exchangeable protons to water is identified for each image voxel typically by asymmetric analysis of the acquired MR signal amplitudes as a function of saturation frequency offset (the so-called Z-spectrum). This asymmetric analysis is with respect to the MR frequency of water protons, which for convenience is assigned to a saturation frequency offset of 0 ppm.

[0006] The measurement of the amplitudes of the acquired MR signals as a function of saturation frequency offset and the asymmetric analysis are inherently very sensitive to any inhomogeneity of the main magnetic field B0. This is because a small shift in the center frequency (e.g. a shift of 0.1 ppm in the saturation frequency relative to the chemical shift of water) easily causes a change in the asymmetric data of more than 10%. This change leads to a large number of artifacts in the finally reconstructed CEST contrast image. It has been shown (Zhou et al., Magnetic Resonance in Medicine, 60, 842-849, 2008) that B0 inhomogeneity can be corrected on a voxel-by-voxel basis in APT imaging by re-centering the Z-spectrum on the basis of a separately acquired B0 map.

[0007] Another problem in CEST MR imaging is that it is often difficult to robustly remove the signal contribution of fat protons, in particular in the presence of B0 inhomogeneities. However, residual fat signal contributions lead to strong bias asymmetries in the amplitude of the acquired MR signal as a function of the saturation frequency offset around the chemical shift of fat protons at -3.4 ppm with respect to water protons. This is of particular concern in applications where MR images of organs with substantial fat content, such as the liver or the breast, are to be acquired. CEST imaging protocols employing fat suppression pulses have been successfully used to reduce artifacts caused by nearby fat tissue.

[0008] However, fat suppression (alone), e.g. using SPIR / STIR / SPAIR RF pulses, is often not sufficient to correct for partial volume fat effects in CEST MR imaging. In general, only a few percent of residual fat signal, which varies spatially with RF field inhomogeneities, is acceptable in quality to remove fat from MR images, but it still causes severe artifacts in CEST MR imaging.

[0009] Furthermore, standard fat suppression pulses tend to be incompatible with RF saturation for CEST Z-spectrum acquisition in the fat proton frequency range. In fact, combined RF saturation fat suppression (e.g. selective inversion pulses in SPIR) can even increase signal intensity around the fat frequency.

[0010] The reason is that with partial saturation of spectral fat signal components, e.g. close to -3.4 ppm (aliphatic fat component), the signal balance between different fat components changes and can lead to an increase in signal level, as explained below. Considering different characteristics of alkene (e.g. +0.6 ppm) and aliphatic (-3.4 ppm) spectral fat components with fat suppression pulses present improves the modeling of background effects.

[0011] Standard fat suppression, in particular SPIR, SPAIR or STIR, uses selective fat- selective fat signal inversion pulses. The water resonance and positive chemical shift are deliberately left untouched, as fat suppression does not significantly affect the SNR of the acquired water image. When only the aliphatic fat component is inverted, opposite phases are prepared between the aliphatic and olefinic fat signals. Fat suppression is optimized when the contributions of aliphatic and olefinic signals are (close to) equal. This balance is changed when additional RF saturation pulses are introduced for the preparation of the APTw / CEST contrast. When the aliphatic component is saturated by the CEST saturation pulse around -3.4 ppm, the remaining fat component is olefinic (positive phase, non-inverted), resulting in a false-positive asymmetry by a signal level increase. Vice versa, when saturated at the olefinic frequency, the sum of inverted aliphatic components is still present, although greatly reduced in amplitude. Here, in the low signal region close to full water saturation, the negative phase of the inverted aliphatic component can lead to a negative signal amplitude, which appears as a mirror image of the elevated water level in the amplitude Z-spectrum.

[0012] CEST-Dixon techniques have been proposed to correct for fat contributions precisely by basing the CEST analysis on water-only images. However, robustness is often compromised due to the interaction between RF saturation and Dixon water / fat separation. A fixed multi-line fat spectrum is used to calculate water and fat fractions from multi-echo MR signals, but it turns out that CEST RF saturation selectively removes some fat lines, which can lead to false water-fat separation (or even water / fat exchange) at various frequency shifts in the Z-spectrum. SUMMARY

[0013] From the above it is readily understood that there is a need for improved MR imaging techniques. It is therefore an object of the present invention to provide a CEST MR imaging method and MR system that enables precise fat correction, preferably based on a small number of saturation frequency shift acquisitions. This will enable clinical use of 2D and 3D protocols for body applications of CEST MR imaging.

[0014] According to the present invention, a method for MR imaging of at least a portion of a body in a main magnetic field Bo placed within an examination volume of an MR system is disclosed. The method of the present invention comprises:

[0015] acquiring MR signals by subjecting the portion of the body to a CEST imaging protocol comprising RF saturation at several different saturation frequency shifts;

[0016] reconstructing a Z-spectrum for each voxel of an MR image from the acquired MR signals;

[0017] modeling the Z-spectrum background including contributions from direct water saturation, magnetization transfer and one or more fat signal components, wherein parameters of the model are determined by fitting the modeled Z-spectrum background to the generated Z-spectrum for each voxel, omitting regions of the Z-spectrum affected by CEST effects; and

[0018] deriving a CEST contrast image by calculating the difference between the modeled Z-spectrum background and the reconstructed Z-spectrum for each voxel.

[0019] According to the present application, the part of the body is subjected to saturation RF pulses, each having a saturation frequency offset with respect to the MR frequency of water protons. In correspondence with conventional CEST MR imaging, saturation RF pulses are radiated at different saturation frequency offsets around the MR frequency of water protons (0 ppm), for example between + / - 3.5 ppm. After each saturation step, MR signals are acquired, for example by means of a spin-echo type sequence. The steps of RF saturation and MR signal acquisition are repeated for different saturation frequency offsets. In a conventional manner, a Z-spectrum is reconstructed from the MR signals acquired for each voxel of a given MR image slice or volume (2D or 3D).

[0020] The present application proposes to model the background of the Z-spectrum of a CEST acquisition for each voxel based on contributions from direct water saturation, magnetization transfer and one or more fat signal components. This approach yields an accurate and reliable fitting of the Z-spectrum data reconstructed from the acquired MR signals to a signal model of the Z-spectrum background. In more detail, according to the present application, the modeled Z-spectrum background is fitted to the reconstructed Z-spectrum for each voxel, wherein specifically the regions of the Z-spectrum affected by CEST effects, where CEST effects are expected to cause Z-spectrum asymmetry, are omitted. The residuals between the model fit and the measured data then provide an accurate result for the contribution of CEST effects. This means that the desired CEST contrast image is finally obtained by calculating the difference between the modeled Z-spectrum background (fitted to the reconstructed Z-spectrum) and the reconstructed Z-spectrum for each voxel, this time specifically in the regions of the Z-spectrum affected by CEST effects, i.e. the regions during the fitting procedure. Thus, the CEST contrast image can be seen as the residuals of the fitting of the Z-spectrum background for each voxel.

[0021] The present application broadens the field of application of CEST in breast and abdominal examinations where partial volume effects are significant, mainly because fat saturation effects can be accurately corrected.

[0022] In embodiments of the present invention, the CEST imaging protocol further comprises fat suppression. The present invention observes that fat suppression leads to a simplification of the shape of the Z-spectrum and to a modified (partially inverted) residual contribution of the spectral fat signal component. The effect of the partial volume fat component on the Z-spectrum is reduced and depends little on the variation of the echo time of the imaging sequence used (e.g. the in- and out-of-phase contributions of the fat protons to the acquired MR signal). The remaining fat contribution caused by fat suppression and RF saturation interaction is corrected by modeling the Z-spectrum background (including not only the contribution from direct water saturation, magnetization transfer (from semi-solid macromolecules), but also the contribution from the fat signal component, e.g. by a (partially inverted) multi-line fat spectrum).

[0023] In another embodiment, the method of the present invention further comprises the step of obtaining a B0 map and applying a saturation frequency shift correction to the reconstructed Z-spectrum according to the B0 map. The B0 map within the imaged part of the body can be determined with MR signals acquired by means of single-point or multi-point Dixon techniques. Generally, the B0 field map, a water image and a fat image are obtained by means of Dixon techniques. Thus, the embodiment allows for applying the Dixon method for both B0 mapping and water / fat separation at the same time. Alternatively, the B0 map can be determined separately by any B0 mapping technique in the art. The B0 shift for each voxel can also be determined from the reconstructed Z-spectrum itself (as the frequency shift at the global minimum of the Z-spectrum).

[0024] In a possible embodiment, the contribution of the direct saturation of the water protons is reproduced in the Z-spectrum background model by a Lorentzian curve with an amplitude A L and a width w L The contribution from magnetization transfer can be modeled by a Gaussian curve with an amplitude A G and a width w G The fat signal can be modeled by a single-line or multi-line MR spectrum of fat protons with an amplitude A F with a resonance frequency known a priori. Thus, by fitting the modeled Z-spectrum background to the reconstructed (measured) Z-spectrum for each voxel, the following parameters are determined: the amplitude A L and the width w L of the Lorentzian curve modeling the direct water saturation, the amplitude A G and the width w G of the Gaussian curve modeling the magnetization transfer, and the amplitude A of the multi-line spectrum modeling the fat signal.F Thus, the fitting procedure is based on only five unknown parameters.

[0025] In a further embodiment, the fat suppressed pulse is used to reach a component of the (e.g. olefin) spectrum which is not fat suppressed and an additional component (e.g. aliphatic) which is reduced by the fat suppression effect s, with 0.0 < s < 1.0. The effect of fat suppression can be fixed for a given APTw / CEST acquisition protocol or included in the fitting procedure (then using six unknown parameters). N l1 N l2 The different fat components are modeled separately, with 0.0 < s < 1.0, using a component of the (e.g. olefin) spectrum which is not fat suppressed and an additional component (e.g. aliphatic) which is reduced by the fat suppression effect s. The effect of fat suppression can be fixed for a given APTw / CEST acquisition protocol or included in the fitting procedure (then using six unknown parameters).

[0026] In yet another embodiment, the noise level is specified during the fitting procedure. The estimation of the noise level can be performed directly via the acquired APTw / CEST data (e.g. based on a reference acquisition S 0 ) or included in the fitting procedure.

[0027] In yet another embodiment, the number of different saturation frequency offsets, i.e. the number of points of the reconstructed Z-spectrum, is less than 50, preferably less than 20. The insight of the present invention is that a limited number of Z-spectrum points is sufficient to accurately determine the CEST contrast. The proposed fat suppression acquisition combined with the Z-spectrum background model enables a stable fitting procedure even for a small number of acquired saturation frequency offsets. A stable fitting can be obtained, e.g. with only 10-20 (preferably 15-18) acquired frequency offsets (with only five or six fitting parameters, see above). Fat suppression is helpful in this case because even if the fat content in the respective voxel is high, it will lead to much smaller changes in the Z-spectrum.

[0028] The method of the present invention described so far can be performed with the aid of an MR device comprising at least one main magnet coil for generating a uniform, steady magnetic field within an examination volume, a number of gradient coils for generating switched magnetic field gradients in different spatial directions within the examination volume, at least one RF coil for generating RF pulses within the examination volume and / or for receiving MR signals from a body of a patient positioned within the examination volume, a control unit for controlling the temporal succession of RF pulses and switched magnetic field gradients, and a reconstruction unit for reconstructing MR images from the received MR signals. The method of the present invention is preferably implemented by corresponding programming of the control unit and / or the reconstruction unit of the MR system.

[0029] ​Any embodiment of the method according to the invention can be advantageously executed by systems in most clinically used MR environments. For this purpose, only a computer program is needed to control the computer (e.g., PC) included in the system, causing it to perform the method steps explained above. The computer program may reside on a data carrier or in a data network for download for installation, for example, by downloading it to the control unit or reconstruction unit of the MR system. Attached Figure Description

[0030] The accompanying drawings disclose preferred embodiments of the invention. However, it should be understood that the drawings are intended for illustrative and explanatory purposes only and are not intended to limit the scope of this disclosure. In the drawings:

[0031] Figure 1 An MR system according to the present invention is shown;

[0032] Figure 2 A flowchart illustrating the method of the present invention is shown;

[0033] Figure 3 The Z-spectrum reconstructed and fitted according to the present invention is shown;

[0034] Figure 4 The phantom setup for testing the method of the present invention is shown;

[0035] Figure 5 It shows how by Figure 3 The image results obtained by performing the method of the present invention on the phantom setting;

[0036] Figure 6 The Z-spectrum reconstructed and fitted according to the present invention is shown, excluding fatty acid suppression;

[0037] Figure 7 An example image containing three regions of interest is shown;

[0038] Figure 8 The fitting results for a region of interest with a moderate fat content are shown according to an embodiment of the present invention;

[0039] Figure 9 Another embodiment of the invention is shown for use with Figure 8 Fitting results for the same region of interest;

[0040] Figure 10 The following is a fitting result for a region of interest with a high fat content, according to another embodiment of the present invention;

[0041] Figure 11 The fitting results for a region of interest with low fat content are shown according to another embodiment of the present invention;

[0042] Figure 12 Fig. 1 illustrates an in-vivo example according to the present application. DETAILED DESCRIPTION

[0043] For the purposes of promoting and understanding the principles of the present disclosure, reference will now be made to the embodiments illustrated in the drawings, and specific language will be used to describe the same. It will nevertheless be understood that no limitation of the scope of the disclosure is intended. Alterations and further modifications of the described devices, systems, and methods, and any further applications of the principles of the disclosure are fully contemplated and included within the scope of the disclosure as would normally be understood by those skilled in the art to which the disclosure pertains. In particular, it is fully contemplated that features, components, and / or steps described with respect to one embodiment can be combined with features, components, and / or steps described with respect to other embodiments of the disclosure. However, for the sake of brevity, various iterations of these combinations will not be described separately. Features described with respect to a system can be implemented in corresponding fashion in a computer-implemented method and / or a computer program product.

[0044] Reference Figure 1 MR system 1 is shown. The device includes a superconducting or resistive main magnet coil 2 that creates a substantially uniform, spatially constant main magnetic field Bo along the z-axis through an examination volume. The device also includes a set of (first-, second-, and, if applicable, third-order) shim coils 2' in which the current flowing through the individual shim coils of the set 2' is controllable in order to minimize B0 deviations within the examination volume.

[0045] Magnetic resonance generation and manipulation system applies a series of RF pulses and switched magnetic field gradients to invert or excite nuclear magnetic spins, induce magnetic resonance, refocus magnetic resonance, manipulate magnetic resonance, spatially or otherwise encode magnetic resonance, saturate spins, etc. to perform MR imaging.

[0046] More specifically, gradient amplifiers 3 apply current pulses to selected ones of whole-body gradient coils 4, 5, and 6 along the x, y, and z-axes of the examination volume. A digital RF frequency transmitter 7 sends RF pulses or pulse packets via a transmit / receive switch 8 to a body RF coil 9 to send RF pulses to the examination volume. A typical MR imaging sequence includes packets of RF pulse segments that, in conjunction with any applied magnetic field gradients, effect a selected manipulation of nuclear magnetic resonance. The RF pulses are used to saturate, excite resonances, invert magnetization, refocus resonances, or manipulate resonances and select portions of the body 10 located in the examination volume. MR signals are also picked up by the body RF coil 9.

[0047] For generating MR images of a limited region of the body 10, a set of local array RF coils 11, 12, 13 is placed in close proximity to the region selected for imaging by means of parallel imaging. The array coils 11, 12, 13 can be used to receive MR signals induced by body coil RF transmission.

[0048] The resulting MR signals are picked up by the body RF coil 9 and / or by the array RF coils 11, 12, 13 and demodulated by a receiver 14, which preferably includes a preamplifier (not shown). The receiver 14 is connected to the RF coils 9, 11, 12 and 13 via a transmit / receive switch 8.

[0049] A host computer 15 controls the shim coils 2' as well as the gradient pulse amplifiers 3 and the transmitter 7 to generate any one of a number of MR imaging protocols such as echo planar imaging (EPI), echo volume imaging, gradient and spin echo imaging, fast spin echo imaging, etc. For a selected sequence, the receiver 14 receives single or multiple MR data lines in rapid succession after each RF excitation pulse. A data acquisition system 16 performs analog-to-digital conversion of the received signals and converts each MR data line to a digital format for further processing. In a modern MR device, the data acquisition system 16 is a separate computer that is dedicated to acquiring raw image data.

[0050] Ultimately, the digital raw image data is reconstructed into an image representation by a reconstruction processor 17 applying a Fourier transform or other suitable reconstruction algorithm such as SENSE or GRAPPA. The MR image can represent a planar slice through the patient, an array of parallel planar slices, a three-dimensional volume, etc. The image is then stored in an image memory, where it can be accessed for converting slices, projections or other portions of the image representation into a suitable format for visualization, e.g. via a video monitor 18 that provides a human readable display of the resulting MR image.

[0051] Reference is made to the flowchart shown and further to Figure 2 The order of the method steps according to embodiments of the application is explained in the following with reference to Figure 1

[0052] ​In step 20, prior to the acquisition of MR signals by means of a spin echo sequence, which can be a fast spin echo (FSE) or turbo spin echo (TSE) sequence or a related pulse sequence like GRASE, the part of the body 10 is subjected to a CEST imaging protocol comprising saturation RF pulses at different saturation frequency offsets. The saturation RF pulses are emitted via the body RF coil 9 and / or via the array RF coils 11, 12, 13, with appropriate control of the transmitter 7 via the host computer 15 to set the saturation frequency offsets relative to the MR frequency of water protons. Different saturation frequency offsets are applied around the MR frequency of water protons (0 ppm), e.g. between + / - 3.5 ppm. A further reference acquisition can be performed "off-resonance", i.e. with a very large frequency offset, which does not affect the MR signal amplitude of water protons, or with the RF saturation power turned off. The CEST imaging protocol used also includes fat suppression RF pulses, e.g. according to the known SPIR, SPAIR or STIR schemes.

[0053] In step 21, a Z-spectrum is computed for each voxel of a given MR image slice or volume. This means that for each voxel position and each saturation frequency offset used in step 20, the MR signal amplitude S divided by the MR signal amplitude of the reference acquisition S 0 is computed S / S 0 . Figure 3 An example Z-spectrum is shown for one selected voxel.

[0054] In step 22, a B0 map is obtained, e.g. via the Dixon technique, using the (partial) Z-spectrum acquisition in step 20 (with appropriate variation of the echo time TE) or a separate B0 mapping scan.

[0055] In step 23, the computed Z-spectrum is corrected for B0 inhomogeneities by applying a saturation frequency offset correction, i.e. the shift of the Z-spectrum along the saturation frequency axis direction in Figure 3 .

[0056] In step 24, a model of the Z-spectrum background S model （f） is fitted to the computed and B0 corrected Z-spectrum. Here, direct water saturation is modeled by a Lorentzian curve with amplitude A L and width w L , and magnetization transfer is modeled by a Lorentzian curve with amplitude A G and width wG Gaussian curve modeling, fat signal through amplitude A F Modeling of the inverted (due to fatty acid suppression) multiline MR spectrum of fatty acid protons. Therefore, the model can be written as: [1]

[0057] Five variable model parameters are sufficient: A L , A G , A F ,as well as w L , w G All distributions are centered after B0 correction. (Adipose protons) f l , w f The spectral frequencies and widths of ) are known a priori. T E This refers to the correlation echo time of the imaging sequence used for MR signal acquisition in gradient echo or echo plane acquisition scenarios. For fast spin echo acquisition, it can be... T E Set to zero. The model is fitted to the B0-corrected Z-spectrum for each voxel, where data from the reconstructed and B0-corrected Z-spectrums within the spectral range affected by the CEST effect of interest (e.g., for APTw 3.5 ± 1.5 ppm; OH-CEST 1.2 ± 0.5 ppm) are ignored in the fitting process to ensure that the resulting model corresponds only to the background of the Z-spectrum (without asymmetry caused by the CEST effect).

[0058] This model can be extended to represent the different contributions of multiple spectral fatty components, which can be modified differently by fatty suppression: [2]

[0059] Please note that in | S 0 –model-function| Place absolute values ​​on both sides. The fat component with negative phase can reduce the signal to negative values, and then mirror it to positive values ​​by taking the absolute value of the Z spectrum of amplitude only.

[0060] Individual fat range a l Normalization makes: [3]

[0061] Fat inhibition effect sIt can be fitted or fixed (e.g., empirically established) to adapt the model to specific MRI protocol settings (fat suppression parameters).

[0062] Since a near-zero Z-spectral amplitude level may indicate a positive noise bias, it can be used to estimate the noise level (e.g., via...). S 0 (and overall noise level or by fitting method) and exclude values ​​below the noise level for fitting. For fast spin echo type MRI acquisition, the echo time parameter in formulas [1] and [2] is used. T E It can be set to zero because all components are rephased for the refocused echo.

[0063] Finally, in step 25, a CEST contrast image is derived by calculating the difference between the modeled Z-spectral background and the reconstructed Z-spectral in the Z-spectral region affected by the CEST effect (e.g., 3.5 ± 1.5 ppm) for each voxel.

[0064] Figure 3 The data points in the image are formed from the Z-spectrum of image voxels with approximately 50% fat fraction using SPIR fat suppression during MR signal acquisition (acquired from the model). The spectral region around the main fat frequencies is -3.4 ppm, which shows a level higher than that without CEST saturation. S 0 The signal increases (normalized to 1.0). The model of the Z-spectrum background (solid line), including direct water saturation, magnetization transfer effects, and inverted fat spectra, largely approximates the measured background. The phantom shows no CEST signal at around 3.5 ppm. If present, the CEST contrast would be derived as the difference between the measured Z-spectrum data points and the model Z-spectrum curves at the corresponding frequencies. Therefore, asymmetric analysis, as is common in conventional CEST methods, is not performed.

[0065] The method of the present invention was tested using a 3 Tesla model with dual-channel RF emission, a 16-channel head coil, and a second-order B0 shim. Two phantoms were used, topped with sunflower seed oil, and then coagulated (at 60°C for 10 minutes) using either (1) phantom fluid (water plus 0.8 g / L CuSO4, T1 = 450 ms, no APT / CEST effect) or (2) egg white (APT / CEST effect) to obtain the magnetization transfer effect. Tilted MR image slices provided a range of partial volumetric fat effects. Figure 4 A details the phantom setup using sunflower seed oil as the fat component and (1) phantom fluid or (2) coagulated egg white for APT / CEST and magnetization transfer effects. The cross-section of the tilted MR image slices is shown in... Figure 4A is shown as a white box. Accordingly, the fat fraction of the voxels in the entire slice varies greatly. Figure 4 B illustrates a cross section of the phantom according to the oblique image slice. Figure 4 C shows the corresponding fat fraction MR image acquired using a GRE Dixon protocol.

[0066] During the test of the method of the invention, 40 RF pulses of 50 ms duration were alternately transmitted via the RF channel (100% duty cycle), resulting in a saturation time T sat = 2 s, and a radio frequency field strength B 1,rms = 2 μΤ. A single-shot 2D fast spin echo (FSE) sequence (with and without SPIR fat suppression) was used with a field of view of (160 mm) 2 , a voxel size of 1.2 x 1.2 x 8 mm 3 , a repetition time and echo time of T R / T E = 5700 ms / 7.0 ms, a flip angle of 90°, a refocusing angle of 120°, a center k-space ordering, and N = 17 saturation frequency offsets (± 1560 ( S 0 ), ± 7.8, ± 6.3, ± 5.1, ± 3.9, ± 2.9, ± 1.8, ± 0.9, ± 0.3 ppm). The total acquisition time was T acq = 2 min.

[0067] B0 mapping was performed with a 3-point multi-shot GRE Dixon sequence with the same geometry: T R = 14 ms, T E = 0.6 ms, a flip angle = 35°, 8 averages, T acq = 45 s.

[0068] The Z-spectrum was calculated and interpolated to 100 points, including B0 correction as described above. Five parameters were fitted: A L 、 A G 、 A F 、 w L 、 w G . A fixed fat proton linewidth w F = 110 Hz and fat frequency f l （Nl =7 line ） , similar to the common multi-peak Dixon water-fat separation. 11 interpolation points were chosen for the fitting procedure at ±7.0, ±4.5, -3.4, -2.3, ±1.3, -0.9, ±0.3 ppm, specifically excluding the amide (or amine) CEST effect range (in the range +1.3...+4.5 pm). The CEST contrast image (APTw) was then derived as the residual from the Z-spectrum background fit. Another image (NOE) was also derived as the residual from the fit at -3.5 ppm, used as a quality check for the fitting procedure. For comparison with the conventional method, the asymmetry image (MTR asym ) was derived from the measured Z-spectrum (B0 corrected) as S -3.5 ppm - S +3.5 ppm / S 0 。

[0069] Figure 5 APTw, NOE and MTR asym images acquired from the model of Figure 4 and fitted from the description model are shown. Images a-f without APT / CEST and magnetization transfer effects from partially coagulated egg white, images g-i with APT / CEST and magnetization transfer effects from partially coagulated egg white. For images d-f, SPIR fat suppression was applied. In APTw image a a uniform, essentially zero intensity of various fat contents throughout the image slice is observed, while MTR asym image c shows strong artifacts. As shown in APTw image d, some areas of high fat content cannot be modeled sufficiently. This is due to low signal levels. Signal overshoot leads to false high intensities in MRT asym image f. APTw image g has high intensities in areas containing egg white (with strong CEST effects). NOE images b, e, h show overall low values, except for some loss / high intensity signal for SPIR in areas of high fat content (image e).

[0070] Further in-vivo tests were performed on a volunteer (IRB approved protocol, informed consent obtained) with the same imaging protocol at 3T, but with a 16-channel breast RF coil. Figure 7 Image examples, Z-spectrum images at +2.9 ppm, including three regions of interest (ROIs), namely fatl, fat2, parl, for detailed analysis of areas with different fat content, are shown.​​​​Figure 8 The fit results using the basic model (step 24) for fat1 ROI, with moderate fat content, are shown, while Figure 9 The extended model for the same ROI was used, where the fat frequencies f l N l1 N l2 =7 line), one olefin at +0.6 ppm relative to water, and six aliphatic at -3.7, -3.4, -3.0, -2.6, -2.3, -1.8 ppm. Figure 8 The basic fit model in cannot sufficiently capture the olefin fat signal at +0.6 ppm, while Figure 9 The extended model in can. Figure 10 and Figure 11 show further ROI examples with high fat content fat2 ROI and low fat content par1 ROI, respectively. For the extended model, the fit quality is good (fit correlation R > 0.98) in all cases. Table 1 lists example fit parameter results obtained for the three example ROIs in human breast tissue with different fat content. Table 1

[0071] Figure 12 In vivo imaging examples are shown, displaying a range of ±5%. While the standard MTR asym The processing shows low signal artifacts throughout the breast tissue for the basic model, but the extended fit method APTw shows a flat representation of the APTw / CEST signal.

[0072] In summary, the addition of olefin components with different weights related to SPIR fat suppression can improve the Z-spectrum fit and interpretation of in vivo real ROI data. In all cases, the actual "zero-crossing" of the model function is not visible in the data due to positive noise bias. The observed noise level increases with increasing fat content. As expected, the fit residuals at +3.5 ppm (APTw) are very small and slightly positive for ROIs with a large non-fat tissue content (fat1, par1). By comparison of the Lorentz / Gaussian functions and in vivo data, it is evident that the residual olefin fat signal with SPIR suppression can affect the evaluation of APTw for moderate or high fat levels (fat2, par2). Figure 9 , Figure 10 ). At low fat levels (par1), the APTw signal is dominated by the olefin signal, and the APTw / CEST signal is not significantly affected by the SPIR fat suppression. Figure 11 ​​), the remaining fat signal apparently causes the asymmetry in the measured data at ±3.5 ppm and changes the spectral shape around the water frequency. Here, the fit of the SPIR effect yields a higher value (s = 0.09, at the fit boundary) compared to the fat voxel (s = 0.06). In compartments with low fat content, the fit of the SPIR effect can be less reliable and can be modified for a given protocol. The residual B0 offset was found to be below 10 Hz by a fit of the frequency center, which indicates that no center fitting is necessary in this example.

[0073] In any embodiment, due to the very high fat content in regions (e.g. fat fraction > 90%) where the overall signal level can be very small, with fat suppression, these regions can be masked, i.e. not included in the fitting process, because the fitting process can be unstable in these regions. These regions do not provide information about the APT / CEST effect of interest, because these effects are only apparent on the water proton signal.

[0074] In any embodiment, optionally, the B0 mapping / B0 correction step (steps 22, 23 in Figure 2 ) can be skipped and instead another model parameter is introduced, which takes the frequency center of the Z-spectrum into account. This can require more Z-spectrum acquisition. On the other hand, the robustness against frequency variations between the separate B0 mapping and CEST acquisition (e.g. due to patient motion) can be improved. This model fit is able to identify B0 variations in a large number of water / fat fractions.

[0075] While SPIR fat suppression is explicitly described in the embodiments, any type of fat suppression pulse (e.g. STIR or SPAIR) can be adapted to the model according to equations [1] and [2]. In case of the model according to equation [2], the suppression effect (amplitude and phase) can be different for other fat suppression pulses.

[0076] Model according to equation [1] in the context of Z-spectra S model (f) Modifications can be made to fit Z-spectra without fat suppression. In this case, the fat content has a large influence on the shape of the Z-spectrum. Figure 6 Z-spectra for a voxel with a fat fraction of about 50% are shown, which were acquired from a phantom without fat suppression during the acquisition. In contrast to the Figure 3 SPIR acquisition in, the spectrum is more affected by the fat component and can become complex due to the applied echo times T E (gradients echo / echo planar acquisition). Nevertheless, the background of the Z-spectrum is successfully fitted to the model according to the above equations. Smodel (f) Fitting, the plus sign is used in front of the fat spectrum. However, in the case of gradient echo or echo planar acquisition (for fast spin echo TE = 0), it is not possible to ensure a stable fit of all echo times TE To align the phase of the water and fat lines, it can be preferred to have a specific TE value, like "in-phase" or "out-of-phase". When a model according to equation [2] is to be used for Z-spectrum without fat suppression, all fat components need to be modeled with a plus sign (f = 0). If there is no fat suppression, there is no inverted or suppressed fat component. Nl1 Fat frequency and Nl2 = 0). If there is no fat suppression, there is no inverted or suppressed fat component.

[0077] It should be understood that one or more of the foregoing embodiments of the application can be combined, as long as the combined embodiments are not mutually exclusive.

[0078] As will be appreciated by those skilled in the art, the present application can be embodied as a device, a method or a computer program product. Accordingly, aspects of the present application can take the form of an entirely hardware embodiment, an entirely software embodiment (including firmware, resident software, micro-code, etc.) or an embodiment combining software and hardware aspects that can all generally be referred to herein as a "circuit," "module" or "system." Furthermore, aspects of the present application can take the form of a computer program product embodied in one or more computer readable medium(s) having computer executable code embodied thereon.

Claims

1. A method of imaging at least part of a body (10), comprising: acquiring signals by subjecting the part of the body (10) to a CEST imaging protocol comprising RF saturation at several different saturation frequency offsets; generating a Z-spectrum from the acquired signals for each voxel from a set of voxels of a magnetic resonance image; modeling a Z-spectrum background comprising contributions from direct water saturation, magnetization transfer and one or more fat signal components, wherein parameters of the model are determined by fitting the modeled Z-spectrum background to the generated Z-spectrum for each voxel, omitting regions of the Z-spectrum affected by CEST effects; and deriving a CEST contrast image based on the modeled Z-spectrum background and the generated Z-spectrum for each voxel.

2. The method of claim 1, wherein, The CEST contrast image is derived based on a difference between the modeled Z-spectrum background and the generated Z-spectrum for each voxel.

3. The method of claim 1 or 2, wherein, The CEST imaging protocol further comprises fat suppression.

4. The method of claim 3, wherein, The plurality of fat signal components is modeled differently.

5. The method of claim 4, wherein, The modeling of the plurality of fat components is adapted to the applied fat suppression.

6. The method of any one of the preceding claims, further comprising: obtaining a B0 map; and applying saturation frequency offset correction to the generated Z-spectrum from the B0 map.

7. The method according to any of the preceding claims, wherein, The B0 map is obtained from the acquired signals by means of single-point or multi-point Dixon techniques.

8. The method according to any of the preceding claims, wherein, For generating the Z-spectrum, the number of points of the acquired signals is less than 50, preferably less than 20.

9. The method according to any of the preceding claims, wherein, The contribution from direct water saturation is modeled by a Lorentzian curve with amplitude A L and width w L .

10. The method according to any of the preceding claims, wherein, The contribution from the magnetization transfer is modeled by a Gaussian curve with a width of A G and a width of w G a Gaussian curve with a width of 11. The method according to any of the preceding claims, wherein, The fat signal is modeled by a single or multi-line MR spectrum of fat protons with an amplitude A F of the fat protons.

12. The method according to any of the preceding claims, wherein, The parameters determined by fitting the modeled Z-spectrum background to the generated Z-spectrum for each voxel are: the amplitude and width of the Lorentzian curve modeling the direct water saturation, the amplitude and width of the Gaussian curve modeling the magnetization transfer, and the amplitudes of the multi-line spectrum modeling the fat signals.

13. The method of any of the preceding claims, wherein, For spin-echo type acquisition, the echo time in the Z-spectrum background model is set to zero.

14. A system (1) for imaging at least part of a body (10), comprising a processor (17) configured to: receive signals associated with the part of the body (10) from an imaging system, the part of the body subjected to a CEST imaging protocol comprising RF saturation at several different saturation frequency offsets; generate a Z-spectrum from the acquired signals for each voxel from a set of voxels of a magnetic resonance image; model a Z-spectrum background comprising contributions from direct water saturation, magnetization transfer and one or more fat signal components, wherein determine parameters of the model by fitting the modeled Z-spectrum background to the generated Z-spectrum for each voxel, omitting regions of the Z-spectrum affected by CEST effects; and derive a CEST contrast image based on the modeled Z-spectrum background and the generated Z-spectrum for each voxel.

15. The system of claim 11, wherein, The CEST contrast image is derived based on a difference between the modeled Z-spectrum background and the generated Z-spectrum for each voxel.

16. The system of claim 14 or 15, further comprising: at least one main magnet coil (2) for generating a uniform, static magnetic field within an examination volume; a number of gradient coils (4, 5, 6) for generating switched magnetic field gradients in different spatial directions within the examination volume; at least one RF coil (9) for generating RF pulses within the examination volume and / or for receiving MR signals from a body (10) of a patient positioned in the examination volume; a control unit (15) for controlling the temporal succession of RF pulses and switched magnetic field gradients, and wherein the processor is part of a reconstruction unit (17) for reconstructing images from the received MR signals.

17. The system of any one of claims 14 to 16, further comprising a display configured to display the CEST contrast image output by the processor.

18. A computer program comprising computer-executable instructions, wherein, Running of the instructions causes the computer to perform the method of any one of claims 1 to 13.

Images