MR Mammography

Abstract

The invention relates to a method for MR imaging of at least a part of an object, i.e. a female breast, comprising the steps of: (a) subjecting the object 10 to an imaging sequence including RF excitation pulses and switched magnetic field gradients, where a number of echo signals are generated after each RF excitation pulse at different echo times TE1, TE2, TE3, (b) acquiring echo signals along a set of radial or spiral k-space trajectories covering a given k-space region, where each of the echo signals generated after the RF excitation pulses is assigned a different direction of the radial or spiral trajectory in k-space, (c) reconstructing single-echo images for each echo time TE1, TE2, TE3 from the acquired echo signals, (d) deriving high-resolution water, fat, B0, and / or apparent transverse relaxation time T2* maps from the single-echo images using a deep learning algorithm, and (d) synthesizing a high-resolution mammography image from the water, fat, B0, and / or T2* maps. Furthermore, the present invention relates to an MR apparatus 1 and a computer program for the MR apparatus 1.

Description

[Technical field]

[0001] The present invention relates to the field of magnetic resonance (MR) imaging. The present invention relates to a method for MR imaging of an object placed in an examination volume of an MR device. The present invention also relates to an MR device and a computer program executed on the MR device. [Background technology]

[0002] MR imaging, which utilizes the interaction between a magnetic field and nuclear spins to form two- or three-dimensional images, is now widely used, especially in the field of medical diagnostics, because it has many advantages over other imaging techniques for imaging soft tissue, does not require ionizing radiation, and is usually non-invasive.

[0003] According to the common MR method, the body of the patient to be examined is placed in a strong, homogeneous magnetic field B0, the direction of which at the same time defines the axis of the coordinate system to which the measurements relate (usually the z-axis). The magnetic field B0 generates different energy levels in the individual nuclear spins depending on the strength of the magnetic field, which can be excited (spin resonance) by applying an electromagnetic alternating magnetic field (RF magnetic field) of a defined frequency (the so-called Larmor frequency, or MR frequency). From a macroscopic point of view, the distribution of the individual nuclear spins gives rise to a global magnetization, which can be deflected from the equilibrium state by applying an electromagnetic pulse (RF pulse) of a suitable frequency, the corresponding magnetic field B1 of which extends perpendicular to the z-axis, so that the magnetization undergoes a precession about the z-axis. The precession describes the surface of a cone, the opening angle of which is called the flip angle. The magnitude of the flip angle depends on the strength and the time width of the applied electromagnetic pulse. In the case of a so-called 90° pulse, the magnetization is deflected from the z-axis into a transverse plane (90° flip angle).

[0004] After the RF pulse ends, the magnetization relaxes back to its original equilibrium state. The magnetization in the z - direction is rebuilt with a first time - constant T1 (spin - lattice relaxation time or longitudinal relaxation time), and the magnetization in the direction perpendicular to the z - direction relaxes with a second, shorter time - constant T2 (spin - spin relaxation time or transverse relaxation time). The transverse magnetization and its fluctuations can be detected by a receive RF coil placed and oriented within the examination volume of the MR device in such a way that the fluctuations of the magnetization are measured in a direction perpendicular to the z - axis.

[0005] To achieve spatial resolution within the body, time - varying magnetic field gradients extending along three principal axes are superimposed on the uniform magnetic field B0, resulting in a linear spatial dependence of the spin - resonance frequency. Thus, the signal picked up by the receive coil contains components of different frequencies that can be associated with different locations within the body. The signal data obtained via the receive coil corresponds to the spatial - frequency domain and is called k - space data. The k - space data usually contains multiple lines acquired with different phase encodings. Each line is digitized by collecting a number of samples. A set of k - space data is converted into an MR image by Fourier transformation.

[0006] Breast cancer is one of the most common cancers in women. Recent studies have shown that MR imaging can localize small breast lesions that may be missed by mammography. MR imaging is useful for detecting breast cancer in women with breast implants and in relatively young women who tend to have dense breast tissue. In these cases, conventional (X - ray - based) mammography may not be as effective as MR imaging. Since MR imaging does not use ionizing radiation, it can be used, in particular, for screening examinations of women under 40 years of age and also to increase the frequency of annual screening examinations for women at high risk of breast cancer.

[0007] Breast MRI has distinct advantages over conventional mammography but also has potential limitations. For example, it is not always possible to distinguish between healthy tissue and cancerous abnormalities, which can lead to unnecessary breast biopsies. This is sometimes referred to as a "false positive" test result. Another drawback of breast MRI is that it has not been possible until now to identify minute calcium deposits (microcalcifications) that may indicate breast cancer.

[0008] Because of its higher resolution and generally higher specificity, conventional mammography remains the first choice as a breast screening test tool, despite the fact that MRI has high sensitivity and can provide persuasive results, especially in dense breast tissue.

[0009] Winkelmann et al. (Journal of Magnetic Resonance Imaging, 24:939 - 944, 2006) described performing MRI using a radial multi - gradient echo sequence and T2* mapping techniques simultaneously. Some of the undersampled images are reconstructed for different echo times. These images are used simultaneously to calculate high - resolution images and T2* maps.

[0010] Khuli et al. (Proceedings of the International Society for Magnetic Resonance in Medicine, 18th Annual Meeting, p.2489, 2010) described a method for the detection of breast microcalcifications by 3 - tesla MRI. A 3D SPGRE ultra - short echo time sequence with radial reconstruction is used in combination with susceptibility - enhanced imaging.

[0011] In EP3531154A1, an MR imaging method is disclosed in which water / fat separation, as well as B0 mapping and T2* mapping, are performed using a multi-gradient echo sequence by radial acquisition or spiral acquisition. An image with a specified contrast is synthesized from the acquired echo signal data, B0 map, and / or T2* map.

Summary of the Invention

Problems to be Solved by the Invention

[0012] From the above, it can be easily understood that an improved method for MR-based mammography is needed.

Means for Solving the Problems

[0013] According to the present invention, an MR imaging method for an object placed within an examination volume of an MR apparatus is disclosed. This method comprises (a) applying an imaging sequence including an RF excitation pulse and a switched magnetic field gradient to the object, wherein a plurality of echo signals are generated at different echo times after each RF excitation pulse; (b) acquiring echo signals along a set of radial or spiral k-space trajectories that cover a given k-space region, wherein different directions of the radial or spiral trajectories in k-space are assigned to each of the echo signals generated after the RF excitation pulse; (c) reconstructing a single-echo image for each echo time from the acquired echo signals; (d) using a deep learning algorithm to derive maps of high-resolution water, fat, B0, and / or apparent transverse relaxation time (T2*) from the single-echo images; (d) synthesizing a high-resolution mammography image from the maps of water, fat, B0, and / or T2*, A high-resolution intermediate image is reconstructed from a combination of echo signals acquired at different echo times (TE1, TE2, TE3) from the entire k-space region covered by a set of radial or helical k-space trajectories, and the steps of deriving high-resolution water, fat, B0, and / or T2* maps are further based on the high-resolution intermediate image.

[0014] According to the present invention, a plurality of echo signals are acquired at different echo times. The echo signals are acquired along radial or helical k-space trajectories (including known acquisition methods such as the KOOSH-ball, Archimedes spiral, spiral phyllotaxis, etc.). Radial or helical k-space trajectories are preferred over conventional Cartesian k-space trajectories due to their inherent robustness to body motion. When using radial or helical k-space sampling, the center of k-space is oversampled and continuously updated. This redundancy can be advantageously utilized to detect and correct the effects of body motion, B0, and T2*.

[0015] Each of the echo signals generated after the RF excitation pulse is assigned a different direction of the radial or helical trajectory in k-space. This means that "echo time distribution type" sampling of k-space is performed. The radial or helical k-space trajectory is uniquely associated with a specific echo time. To achieve a very high acquisition speed, it should be avoided to acquire the same radial or helical k-space trajectory multiple times at different echo times.

[0016] Single-echo images are reconstructed for each echo time from the echo signals acquired from the k-space regions to be covered. A k-space weighted image contrast (KWIC) filter can be advantageously used for the reconstruction of single-echo images (see Song et al., Magn. Reson. Med., 44, 825-832, 2000). For example, each of the single-echo images can be reconstructed from the undersampled k-space data belonging to their respective echo times. The acquired echo signals belonging to a particular echo time will generally be undersampled, at least at the periphery of k-space. Thus, compressive sensing (CS) is advantageously used to reconstruct single-echo images from the undersampled signal data. The (KWIC-filtered) echo signals belonging to a particular echo time are irregularly distributed within k-space. In CS theory, it is known that the signal data can be significantly reduced. In CS theory, a signal data set having a sparse representation in the transform domain can be recovered from the undersampled measurements by application of an appropriate regularization algorithm. As a mathematical framework for signal sampling and reconstruction, CS defines the conditions under which, even when the k-space sampling density is much lower than the Nyquist criterion, the signal data set can be accurately or at least with high quality reconstructed, and also provides a method for such reconstruction.

[0017] Furthermore, a single high-resolution intermediate image is reconstructed from the entire k-space region covered by a set of radial or spiral k-space trajectories from the entire echo signal, i.e., the echo signals acquired for each different echo time.

[0018] As a next step, a deep learning algorithm is used to derive maps of high-resolution water, fat, B0, and / or apparent transverse relaxation time (T2*) from single-echo images. The calculation of “super-resolution” MR images from low-resolution images using deep learning methods is known as such in the art. Using appropriate training data including single-echo images as input, and using the associated high-resolution (super-resolution) maps of water, fat, B0, and T2* as the output of an artificial neural network, the Dixon method super-resolution deep learning reconstruction algorithm according to the present invention is implemented. To improve the quality of the high-resolution maps of water, fat, B0, and T2*, the deep learning-based reconstruction is further based on high-resolution intermediate images reconstructed from combinations of echo signals acquired for all different echo times, as described above.

[0019] High-resolution tissue classification maps and / or calcification maps can be derived from maps of water, fat, B0, and / or T2*. Further, additional maps are derived, such as, for example, fat classification maps, fat content maps, etc. T2* mainly results from the inhomogeneity of the main magnetic field B0. These inhomogeneities are the result of inherent inhomogeneities in the main magnetic field itself and magnetic field distortions induced by the magnetic susceptibility of tissues. As a result, a high-resolution T2* map derived from a single-echo image (ideally, in combination with maps of water, fat, and B0) can be used according to the present invention to derive a (high-resolution) tissue classification map. In particular, microcalcifications that may be associated with cancerous tumors in breast tissue cause magnetic field distortions induced by their specific magnetic susceptibility in the local environment. This is utilized in the present invention, particularly by using the T2* map, to derive high-resolution calcification maps that are particularly diagnostically valuable. In combination with the super-resolution Dixon method deep learning reconstruction, microcalcifications with a diameter of less than 1 mm can be identified.

[0020] Finally, a high-resolution mammographic image is synthesized from the water, fat, B0, and / or T2* maps. The synthesized mammographic image is calculated by assigning a Hounsfield unit value to each pixel or voxel. The Hounsfield unit value assigned to each image location provides the x-ray radiation attenuation characteristics of the tissues, which allows the calculation of a real mammographic image with exactly the same contrast characteristics as the radiologist is accustomed to in the field of conventional mammographic screening examinations. The assignment of Hounsfield units to the image locations is performed based on the high-resolution water, fat, B0, and / or T2* maps. It is known in the art that Dixon imaging, which provides water / fat separation, is known to be effective for the purpose of generating a synthetic x-ray image. In order to correctly determine the Hounsfield units for each image location, additional contrast information of the same regions is useful. Therefore, the synthesis of the mammographic image is ideally also based on the derived tissue classification and / or calcification maps.

[0021] In order to obtain a uniform k-space distribution of the echo signal, (k x plane / k y The rotation angle of the radial or spiral k-space trajectory (placed in the x-direction plane) can be incremented during the acquisition by the golden angle, which corresponds to 180° multiplied by the golden ratio. It is preferable that the increments of the golden angle are small, avoiding large gradient increments. Another possibility is to acquire echo signals along the radial or spiral k-space trajectory in consecutive order in k-space. This has the advantage that system-induced phase errors resulting from even and odd echo sampling are corrected. A further option is to increment the k-space signal during the acquisition by the golden angle, which corresponds to 180° multiplied by the golden ratio. z The k-space trajectory of each echo is then incremented in the k direction. zIt is to distribute in a direction (i.e., a direction perpendicular to the plane in which the radial or spiral k-space trajectory is rotated). The uniform distribution (as opposed to the distribution on a regular grid of k-space) aids in the derivation of high-resolution maps of water, fat, B0, and / or apparent transverse relaxation time (T2*) from single-echo images using deep learning algorithms. To further aid in the deep learning-based derivation of the high-resolution maps, a variable shift of the echoes along the radial or spiral k-space trajectory is applied. Further, the acquisition bandwidth (i.e., the time width and / or sampling frequency of each echo acquisition) varies between different shots of the imaging sequence and / or between acquisitions of different echo signals.

[0022] In a preferred embodiment, the first echo signal after each RF excitation is generated at an ultra-short echo time (UTE) to extend the range of accessible echo time values. Known partial echo techniques are also applied for this purpose. Echo shifting is alternatively or additionally utilized to improve the coverage of echo times and optimize T2* mapping. Ultra-short echo times are well-suited for the identification of calcifications, which are characterized by very short apparent transverse relaxation times. As a result, fitting and T2* mapping are improved. In combination with the UTE technique, a 3D radial k-space acquisition scheme ("KOOSH-ball") can be utilized to sample a given k-space region. Due to MR hardware limitations (the dead time of the transmit / receive switch), each radial k-space trajectory from the center sampled at ultra-short echo times will have a gap at the center of k-space. To fill these gaps, the known PETRA scheme is utilized. In the PETRA scheme, the central k-space region is filled with gaps by separate (Cartesian) sampling.

[0023] In a further preferred embodiment, the body movement of the object occurring during acquisition is derived from at least one of the echo signals. This can be used to correct the detected body movement in the step of reconstructing a single echo image. For example, one or more of the echo signals representing radial or spiral k-space samples can be used as an endogenous navigator for the detection of body movement or respiratory state. This can be advantageously combined with the method of the present invention to reduce artifacts induced by body movement. In particular, the detected body movement (e.g., the detected displacement of the object being imaged) is due to one of several discrete body movement states experienced by the patient being examined or the body part being examined, for example during breathing. Therefore, a B0 map can be derived for each body movement state from the echo signals assigned to each body movement state. In this way, by incorporating intrinsic body movement and the determination and correction of B0 induced by body movement, more accurate T2* mapping is enabled. The accuracy of tissue mapping in breast imaging can be significantly improved by addressing B0 variations induced by body movement. In particular, B0 variations induced by breathing have a large impact on the mapping of breast tissue. The intermediate information resulting from the detection of body movement and B0 induced by body movement can be provided to a deep learning algorithm to determine an accurate tissue classification map. Such improved fitting also enables the combination of this technique with a susceptibility weighted imaging strategy (SWI), for example, to determine a more accurate phase image.

[0024] As a result, the present invention provides high-resolution MR calcification mammography and tissue characterization techniques. Tissue characterization diagnosis provides additional information to the calcification map. From the same acquisition, the oxygenation level of the tissue, for example, fat classification based on fat content, can be provided within the same scan. This provides tumor-related diagnostic and treatment information well-suited for breast screening examinations and non-contrast breast MR imaging.

[0025] The method of the present invention described so far can be carried out by an MR apparatus including at least one main magnet coil for generating a uniform static magnetic field B0 within an examination volume, several gradient coils for generating magnetic field gradients switched in different spatial directions within the examination volume, at least one body RF coil for generating an RF pulse within the examination volume and / or for receiving an MR signal from a patient's body disposed within the examination volume, a control unit for controlling the temporal evolution of the RF pulse and the switched magnetic field gradients, and a reconstruction unit for reconstructing an MR image from the received MR signals. The method of the present invention can be implemented by corresponding programming of the reconstruction unit and / or the control unit of the MR apparatus.

[0026] The method of the present invention can be advantageously implemented in most MR apparatuses currently used clinically. For this purpose, it is only necessary to utilize a computer program for controlling the MR apparatus so that the MR apparatus executes the method steps described above of the present invention. The computer program may be present on a data carrier or may be present within a data network so as to be downloaded for installation in the control unit of the MR apparatus.

[0027] The accompanying drawings disclose preferred embodiments of the present invention. However, it should be understood that the drawings are for illustrative purposes only and are not intended to define the limitations of the present invention.

Brief Description of the Drawings

[0028]

Figure 1

Figure 2

Figure 3

Figure 4

DETAILED DESCRIPTION OF THE INVENTION

[0029] Referring to FIG. 1, the MR apparatus 1 is shown as a block diagram. The apparatus includes a superconducting or resistive main magnet coil 2 such that a substantially uniform and temporally constant main magnetic field B0 is formed along the z-axis through the examination volume. The apparatus further includes a set of (primary, secondary, and, if appropriate, tertiary) shim coils 2', and the currents through the individual shim coils of the set 2' are controllable for the purpose of minimizing the B0 deviation within the examination volume.

[0030] The magnetic resonance generation and operation system applies a series of RF pulses and switched magnetic field gradients to perform inversion or excitation of nuclear magnetic spins, induction of magnetic resonance, rephasing of magnetic resonance, operation of magnetic resonance, spatial and other encoding of magnetic resonance, saturation of spins, etc., to execute MR imaging.

[0031] More specifically, the gradient amplifier 3 applies current pulses or waveforms to selected ones of the whole-body gradient coils 4, 5, and 6 along the x-axis, y-axis, and z-axis of the examination volume. The digital RF frequency transmitter 7 transmits RF pulses or pulse packets to the body RF coil 9 via the transmit / receive switch 8 to transmit RF pulses to the examination volume. A typical MR imaging sequence consists of a packet of RF pulse segments of short time width that achieve selected operations of the nuclear magnetic resonance signal, along with any applied magnetic field gradients. The RF pulses are used to perform saturation of resonance, excitation of resonance, inversion of magnetization, rephasing of resonance, or operation of resonance, and to select a portion of the body 10 placed within the examination volume. The MR signal is also picked up by the body RF coil 9.

[0032] To generate an MR image of a limited region of the body 10 or for scan acceleration by parallel imaging, a set of local array RF coils 11, 12, 13 are arranged adjacent to the region selected for imaging. The array coils 11, 12, 13 can be used to receive MR signals induced by body coil RF transmission.

[0033] The resulting MR signals are picked up by the body RF coil 9 and / or the array RF coils 11, 12, 13 and are preferably demodulated by a receiver 14 which 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.

[0034] The host computer 15 controls the shim coil 2’, as well as the gradient pulse amplifier 3 and the transmitter 7 to generate any of a plurality of MR imaging sequences such as echo-planar imaging (EPI), echo-volume imaging, gradient and spin-echo imaging, fast spin-echo imaging, etc. For the selected sequence, the receiver 14 quickly and continuously receives single or multiple MR signals following each RF excitation pulse. The data acquisition system 16 performs analog-to-digital conversion of the received signals and converts each MR data sample into a digital format suitable for further processing. In modern MR devices, the data acquisition system 16 is a separate computer specialized for the acquisition of raw image data.

[0035] Finally, the digital raw image data is reconstructed by a reconstruction processor 17 which applies a Fourier transform or other suitable reconstruction algorithm to become an image display. The MR image is a display such as 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 and this image can be accessed to convert slices, projections, or other portions of the image display into a suitable format for visualization, which is performed, for example, via a video monitor 18 which provides a human-readable display of the resulting MR image.

[0036] The host computer 15 is programmed to execute the methods of the present invention described above and below in this specification.

[0037] In one embodiment of the present invention, an imaging sequence is applied that includes an RF excitation pulse and switched magnetic field gradients in the readout direction and phase encoding directions x and y and the slice selection direction z. To fully cover the required regions of k-space with a radial sampling pattern, multiple sets of echo signals are acquired in a sequence of multiple repetitions (shots) using different gradient waveforms in the x-direction / y-direction and / or z-direction. The timing and amplitude of the readout gradient in the x / y direction are selected such that different echo times TE1, TE2, …, TEN are provided. Preferably, the imaging sequence is a UTE sequence such that the echo signal immediately after each RF excitation is generated with an ultrashort echo time.

[0038] Using its 1 / 2 excitation pulse, a basic UTE pulse sequence using radial imaging from the center of k-space can be utilized. The magnetic field gradient is switched in the z-direction for phase encoding to achieve a Cartesian k-space sampling pattern in this direction (by the known "stack of stars" sampling method that enables 3D image reconstruction). The shift k used for super-resolution imaging (see van Reeth et al., Concepts in Magnetic Resonance Part A, vol. 40A(6), 306 - 325, 2012) zSlice-encoding sampling can be advantageously applied. Using radial k-space sampling, the center of the k-space is oversampled. By shifting or delaying the echo acquisition and / or by successive multiple echo acquisitions, multiple echo signals with varying echo times can be acquired. Reverse echo acquisition may be applied. "Echo-time distribution type" sampling of the k-space is performed. This means that the echo times are distributed over the k-space with oversampling at the center of the k-space. This is shown in FIG. 2. FIG. 2 shows the directions of the radial k-space trajectories for radial imaging for three different echo times TE1, TE2, and TE3 in four radial acquisitions per shot of the UTE sequence used. The acquisition sequence of the radial k-space trajectories is indicated by numbers 1, 2, …, 12. In this way, with an optimal distribution of the acquired echo signal data in the k-space, very fast acquisition is achieved. The acquisition of the echo signal data in the manner described is performed in step 31 of the flowchart of FIG. 3.

[0039] As the next step (step 32 in FIG. 3), single-echo images are reconstructed for each echo time from the echo signals acquired from the central portion of the k-space region to be covered. A k-space weighted image contrast (KWIC) filter can be used to reconstruct the single-echo images.

[0040] Furthermore, in step 33, a single high-resolution intermediate image is reconstructed from the entire echo signal, i.e., the echo signal containing all the echo signals acquired for different echo times, and from the entire k-space region covered by a set of radial or spiral k-space trajectories.

[0041] In step 34, body movements occurring during acquisition are derived from at least one of the echo signals. One or more of the radial k-space samples can be used as an endogenous navigator for detecting body movement or respiratory state. In particular, the detected body movements belong to any of several discrete body movement states. Therefore, a low-resolution B0 map is derived for each body movement state from the echo signals assigned to each body movement state.

[0042] In step 35, using a deep learning algorithm, high-resolution maps of water, fat, B0, and T2* are derived from the single echo image (and shifted k z slice encoding). To improve the quality of the high-resolution maps of water, fat, B0, and T2*, the deep learning-based reconstruction is also based on the high-resolution intermediate image reconstructed in step 33. Further, the body movement in step 34 and the intermediate information resulting from the detection of B0 induced by the body movement are provided to the deep learning algorithm to determine an accurate tissue classification map. The accuracy of T2* mapping is significantly improved by addressing the B0 variations induced by body movement.

[0043] From the maps of water, fat, B0, and / or T2*, a high-resolution tissue classification map and a calcification map are derived in step 36.

[0044] Finally, from the maps of water, fat, B0, and / or T2*, a high-resolution mammography image is synthesized in step 37. The synthesized mammography image is calculated by assigning a Hounsfield unit value to each pixel or voxel and calculating the intensity of the pixel or voxel according to the resulting X-ray radiation attenuation. The assignment of the Hounsfield unit to the image position is performed based on the high-resolution maps of water, fat, B0, and / or T2*. The synthesis of the mammography image is also based on the tissue classification and / or calcification map derived to correctly determine the Hounsfield unit for each image position.

[0045] Additional steps (not shown), such as post-processing steps for removing system imperfections (e.g., blurring removal) and / or using confidence maps for deep learning reconstruction, may optionally be included.

[0046] Possible image outputs for the user are high-resolution water, fat, and T2* maps obtained as a result of deep learning reconstruction, particularly synthesized mammography images as shown in FIG. 4. Also, the derived tissue classification and calcification maps may be presented to the user to assist in diagnosis.

Claims

1. A method for MR imaging of an object placed within the examination volume of an MR apparatus, the method comprising: (a) applying an imaging sequence including an RF excitation pulse and a switched magnetic field gradient to the object, wherein a plurality of echo signals are generated at different echo times after each RF excitation pulse; (b) acquiring the echo signals along a set of radial or spiral k-space trajectories that cover a given k-space region, wherein different directions of the radial or spiral trajectory in k-space are assigned to each of the echo signals generated after an RF excitation pulse; (c) reconstructing a single-echo image for each echo time from the acquired echo signals; (d) using a deep learning algorithm to derive high-resolution maps of water, fat, B ０ , and / or apparent transverse relaxation time (T ２ *) from the single-echo images; (d) synthesizing a high-resolution mammography image from the maps of water, fat, B ０ , and / or T ２ *; and A high-resolution intermediate image is reconstructed from a combination of the echo signals acquired at different echo times from the entire k-space region covered by the set of radial or spiral k-space trajectories, and the step of deriving the high-resolution maps of water, fat, B ０ , and / or T ２ * is performed based on the high-resolution intermediate image.

2. The method according to claim 1, wherein a tissue classification map and / or a calcification map is derived from the maps of water, fat, B ０ , and / or T ２ *.

3. The method according to claim 1 or 2, wherein the first echo signal after each RF excitation is generated at an ultra-short echo time.

4. The method according to any one of claims 1 to 3, wherein an echo signal is separately acquired from one or more missing portions of the k-space region.

5. The method according to any one of claims 1 to 4, wherein the set of the radial or spiral k-space trajectories oversamples a central portion of the k-space.

6. The method according to any one of claims 1 to 5, wherein during acquisition, a rotation angle of the radial or spiral k-space trajectory is incremented so as to obtain a uniform k-space distribution of the echo signal.

7. The k-space trajectories of individual echoes are distributed in the k ｚ direction, the method according to any one of claims 1 to 6.

8. During the acquisition of the echo signal, body movement of the object that occurs is derived from at least one of the echo signals, the detected body movement is corrected in the step of reconstructing the single echo image, the detected body movement belongs to one of several body movement states, and a B ０ map is derived for each of the body movement states from the echo signals assigned to the respective body movement states, The method according to any one of claims 1 to 7.

9. The step of deriving the high-resolution water, fat, B ０ , and / or T ２ * map, the step of synthesizing the high-resolution mammography image, and / or the derivation of the tissue classification map and / or the calcification map are derived based on the B ０ map derived for different said body movement states, the method according to claim 8.

10. The method according to any one of claims 1 to 9, wherein a k-space enhanced image contrast (KWIC) filter is used for reconstructing the single echo image.

11. At least one main magnet coil that generates a uniform static magnetic field B within the inspection volume ０ And several gradient coils that generate magnetic field gradients switched in different spatial directions within the inspection volume, and at least one RF coil that generates an RF pulse within the inspection volume and / or receives an MR signal from an object disposed within the inspection volume, and a control unit that controls the temporal transition of the RF pulse and the switched magnetic field gradient, and a reconstruction unit that reconstructs an MR image from the received MR signal. An MR apparatus, wherein the MR apparatus is configured to execute the method according to any one of claims 1 to 10.

12. A computer program executed on the MR apparatus that controls the MR apparatus to execute the method according to any one of claims 1 to 10 when executed on the MR apparatus.

Images