Interpolated imaging

Abstract

A method for the artifact-reduced generation of formatted interpolated image data (FIBD) is described. In this method, image data (BD) is acquired from an image acquisition area. Supplemented image data (EBD) is generated by adding zero values ​​to the image data (BD) in an image edge region. Furthermore, supplemented k-space data (ERD) is generated by transforming the supplemented image data (EBD) into k-space. Subsequently, interpolated k-space data (IRD) is generated by interpolating the supplemented k-space data (ERD). Then, interpolated image data (IBD) is generated by image reconstruction based on the interpolated k-space data (IRD). Finally, formatted interpolated image data (FIBD) is generated by removing edge regions from the interpolated image data (IBD). An image data generation device (70) is also described. In addition, a magnetic resonance imaging system (90) is described.

Description

[0001] The invention relates to a method for generating formatted interpolated image data. The invention also relates to an image data generation device. Furthermore, the invention relates to a magnetic resonance imaging system.

[0002] When image data is processed by interpolation in Fourier space (in magnetic resonance imaging, Fourier space is usually called k-space) with respect to its resolution, image artifacts can occur if the object being imaged does not fit entirely within the image area. Such a scenario is found in Fig. Figure 1 illustrates this. Image data should initially be understood generally as information presented visually, based on measurement data. The image data was thus generated by an image acquisition unit from an image acquisition area through a physical measurement. The term image acquisition unit is intended to include, in particular, image acquisition units that operate using optical methods, but also imaging systems that work with the active excitation of an area to be imaged by electromagnetic waves, especially high-frequency signals or X-rays, which are emitted by the image acquisition unit itself. A particularly important type of imaging, especially in medicine, is magnetic resonance imaging or magnetic resonance measurement.

[0003] Imaging systems based on magnetic resonance measurement, particularly of nuclear spins, known as magnetic resonance tomographs or magnetic resonance imaging systems, have become established and proven their worth through a wide range of applications. In this type of image acquisition, a static base magnetic field B0, which serves to initially align and homogenize the magnetic dipoles under investigation, is typically superimposed with a rapidly switched magnetic field, the so-called gradient field, to achieve spatial resolution of the imaging signal. To determine the material properties of an object being imaged, the dephasing or relaxation time after a displacement of the magnetization from its initial alignment is measured, allowing for the identification of various material-specific relaxation mechanisms and relaxation times.The deflection is usually achieved by a number of RF pulses (the abbreviation RF stands for high frequency), also referred to as excitation pulses, and the spatial resolution is based on a time-defined manipulation of the deflected magnetization using the gradient field in a so-called measurement sequence or control sequence, which defines a precise temporal sequence of RF pulses, the change of the gradient field (by emitting a switching sequence of gradient pulses) and the acquisition of measured values.

[0004] Typically, an intermediate step is used to correlate measured magnetization—from which the aforementioned material properties can be derived—with a spatial coordinate of the measured magnetization in the space where the object under investigation is located. In this intermediate step, acquired raw magnetic resonance data, also known as k-space data, are arranged at readout points in the so-called "k-space," where the coordinates of k-space are encoded as a function of the gradient field. The magnitude of the magnetization (especially the transverse magnetization in a plane perpendicular to the previously described fundamental magnetic field) at a specific location of the object under investigation can be determined from the readout point data using a Fourier transform. This transform calculates the signal strength in space from a signal strength (magnitude of the magnetization) that is assigned to a specific frequency (the spatial frequency) or phase angle.The representation of signal values ​​in spatial space as image data is often also indicated by the terms image space or image data space, in contrast to k-space, in which signals are represented as a function of frequencies and phases. k-space is sometimes also referred to as frequency space or Fourier space. The latter refers to the fact that k-space data can be obtained by a Fourier transform of image data.

[0005] However, to save time, only a partial sampling of k-space is often performed. This reduced sampling of k-space leads to a reduction in the image information reconstructed from the sampled k-space data, particularly a reduction in image resolution. To compensate for this loss of information, methods exist to address undersampling. Such interpolation methods can involve augmenting the k-space data or directly increasing the resolution within the image data space. Furthermore, even with image data that is not undersampled or is even oversampled, there is a need to increase the resolution through interpolation.

[0006] For image interpolation in the frequency domain or k-space, the image to be interpolated is first transformed into k-space and then interpolated by zero-filling or artificial data using an artificial intelligence (AI)-based method, such as a Deep Resolve Sharp technique. For example, the number of pixels in a 2D image with 128x128 pixels is doubled to 256x256 pixels, thus halving the distance between the pixels from 1 mm to 0.5 mm in a field of view of 128x128 mm². This method is frequently used in MRI imaging. However, it can lead to infolding artifacts after image interpolation, as seen in Fig. 1 is displayed when the object being examined is larger than the field of view (FOV).

[0007] Deep Resolve Sharp (abbreviated as "DRS") is a method for interpolating MR images using a neural super-resolution network, which can increase the resolution of image data. Such a method is described in Yulun Zhang et al., "Residual Dense Network for Image Super-Resolution," Proceedings of the IEEE conference on computer vision and pattern recognition 2018.

[0008] Deep Resolve is an advanced medical imaging technology that combines deep learning and artificial intelligence techniques to improve the magnetic resonance imaging (MRI) image reconstruction process, specifically increasing the resolution and sharpness of MRI images. It comprises several components, including Deep Resolve Sharp and Deep Resolve Boost.

[0009] To avoid invagination of large objects in MRI imaging, k-space is generally oversampled during acquisition. In the frequency encoding direction, this is usually possible without additional time expenditure. After the actual image reconstruction from k-space to the image data space or spatial space (e.g., using Fourier transform, GRAPPA (Generalized Autocalibrating Partial Parallel Acquisition), or SENSE (Sensitivity Encoding)), the oversampled area is truncated, leaving only the desired field of view. If subsequent image interpolation is to be performed, the problem described earlier can occur: invagination artifacts appear because the information from the oversampled image area is now missing.

[0010] This problem has so far been solved by interpolating the k-space data before the actual image reconstruction (e.g., using zero-filling). However, this leads to increased reconstruction time, especially when more advanced reconstruction techniques are used, such as those supported by AI (artificial intelligence). Siemens Healthineers MR (Magnetic Resonance Imaging) uses a method called Deep Resolve Boost (DRB) for this purpose.

[0011] Fig. Figure 2 schematically shows the reconstruction chain for a conventional reconstruction and thus a conventionally used solution for interpolating image data.

[0012] Fig. Figure 3 shows the reconstruction chain using an advanced reconstruction technique (e.g., DRB). In this latter reconstruction chain, the reconstruction time is reduced by first truncating the image edge area caused by oversampling, thus generating a smaller amount of data to be processed. However, due to the reduction in data volume, the described inversion artifacts can occur, for which there has been no solution to eliminate or mitigate them until now.

[0013] US 2016 / 0 313 416 A1 describes a method for obtaining interpolated sensitivity data from recording coils during an MR measurement.

[0014] Therefore, one task in magnetic resonance imaging is to avoid or at least reduce artifacts, and in particular inflection artifacts, when increasing the resolution of magnetic resonance image data through interpolation, and / or to reduce the time required compared to a conventional approach where no data reduction takes place.

[0015] This problem is solved by a method for generating formatted interpolated image data according to claim 1, by an image data generation device according to claim 7 and by a magnetic resonance imaging system according to claim 8.

[0016] In the inventive method for preferably artifact-reduced generation of formatted interpolated image data, image data is acquired from an image acquisition area. The image data has a predetermined image format that depicts the image acquisition area. As briefly explained at the outset, "acquisition of image data" is understood to mean a measurement on the basis of which image information regarding a local or spatial distribution of the measured values, or values ​​determined on the basis of the measured values, is generated. Magnetic resonance imaging is mentioned as a particularly preferred method of image acquisition or imaging. However, the inventive method is not limited to magnetic resonance imaging. Generally speaking, the inventive method is aimed at increasing the resolution of image data obtained through measurements.

[0017] After image acquisition, which includes image reconstruction, supplementary image data is generated in a predetermined image border region, preferably by adding zero values ​​to the image data. This predetermined image border region lies outside the predetermined image format and borders it. "Outside the predetermined image format" means that the predetermined image border region lies outside the boundaries of the predetermined image format.

[0018] Preferably, the image edge region selected for augmentation is the one where a depicted object is "cut off" by the image edge, meaning it is only partially depicted, with the object extending to the image edge. In other words, a "dummy" image region is added to a selected image edge, which then forms the predetermined image edge region. This image edge region is designed in such a way that no involution artifacts occur during transformation or interpolation. This added image region preferably comprises a single column or row, or a layer with the thickness of one voxel. This measure avoids involution artifacts caused by objects appearing in the image edge region. Furthermore, augmented k-space data is generated by transforming the augmented image data into k-space. Such a transformation is preferably performed using a Fourier transform.The term "k-space" should be understood to refer specifically to the k-space used in magnetic resonance imaging (MRI). However, it should also be understood to encompass a Fourier space into which image data is transformed from spatial or image data space via a Fourier transform. The latter applies particularly to image data not generated by MRI.

[0019] Subsequently, interpolated k-space data are generated by interpolating the supplemented k-space data, thereby creating unmeasured k-space data based on the supplemented k-space data. In such an interpolation, additional k-space data is preferably added as intermediate values ​​or intermediate data between existing k-space data, for example, based on the existing k-space data. A particularly preferred method includes so-called "zero-filling," in which the k-space is simply filled with zero values.

[0020] Zero-filling is routinely used to expand the image matrix size in the phase-encoded direction. In 2D imaging, this increase in matrix size occurs "within the plane" (e.g., from 256 to 512 pixels). Gains occur with the first doubling of the matrix size; thereafter, no further benefit is realized. Although zero-filling adds no information to the input raw data, it can still improve the apparent spatial resolution of the image due to reduced subvolume artifacts. Zero-filling serves as a method for interpolating the signals of neighboring voxels, giving the image a smoother and less "pixelated" appearance.

[0021] Interpolated image data is then generated by image reconstruction based on the interpolated k-space data. This image reconstruction is preferably performed by a Fourier transform of the interpolated k-space data into the image data space. As a consequence of adding an image border region to the image data and as a result of the transformations associated with the interpolation, the interpolated image data exhibits border regions that lie outside the predetermined image format.

[0022] Finally, formatted interpolated image data is generated by removing the image border areas that lie outside the predetermined image format. Preferably, opposing image border areas in the interpolated image data, corresponding to the image border area already created in the step for generating augmented image data, are eliminated. The removed image border areas are aligned analogously to, or encompass, the image border area added to the image space during the augmentation step in terms of its extent and arrangement. Removing the image border areas restores the original format of the image data used as input. In other words, the augmentation of the image data in the image border area before interpolation is reversed after interpolation, thus preserving the original image format.This advantageously avoids folding artifacts, especially when objects or important image areas border on or extend beyond the edge of the image capture area.

[0023] The image data generation device according to the invention has an image acquisition unit for receiving and / or generating image data from an image acquisition area with a predetermined image format depicting the image acquisition area.

[0024] Part of the image data generation device according to the invention is also a supplementary unit for generating supplementary image data by adding predetermined image data in a predetermined image border area, which lies outside the predetermined image format and borders the predetermined image format. As already explained, the image border area is selected such that an image border is supplemented or extended in which an object borders the image border, for example, because it is not fully depicted.

[0025] The image data generation device according to the invention also includes a transformation unit for generating augmented k-space data by transforming the augmented image data into k-space or Fourier space. A Fourier transform is preferably used for this purpose.

[0026] Part of the image data generation device according to the invention is also an interpolation unit for generating interpolated k-space data by interpolating the supplemented k-space data, whereby unmeasured k-space data are generated on the basis of the supplemented k-space data.

[0027] The image data generation device according to the invention also includes an image generation unit for generating interpolated image data by means of image reconstruction based on the interpolated k-space data, wherein the interpolated image data includes image edge areas which lie outside the predetermined image format.

[0028] The image reconstruction preferably comprises a Fourier transform, which preferably includes an inverse operation to the type of transformation used in the transformation of image data into k-space. Therefore, when referring to a "Fourier transform," the corresponding inverse transformation can also be meant, which, when chained with its associated transformation, generates the identity. Furthermore, the image data generation device according to the invention comprises a formatting unit for generating formatted interpolated image data by removing the image border regions in the interpolated image data that lie outside the predetermined image format. The image border regions correspond in their arrangement to the image border region that was generated during the generation of the augmented image data. The image border regions preferably comprise opposing image border regions that were created due to the transformation into k-space or...Fourier space and the inverse transformation into image space were created. In this sense, the image border areas are arranged and aligned analogously to the predetermined image border area that was added in image space during the augmentation step. Image border areas can also be added in multiple dimensions during the augmentation step, and correspondingly, image border areas can be removed or clipped in multiple dimensions during the formatting step.

[0029] The removal of image edge areas should therefore preferably include the removal of opposing edge areas. The image data generation device according to the invention shares the advantages of the inventive method for the preferably artifact-reduced generation of formatted interpolated image data.

[0030] If the image data generation unit generates and processes image data that includes magnetic resonance imaging data, then the image acquisition unit comprises: - an input interface for receiving k-spatial data from a study area of ​​a study object, - a transformation unit for transforming the received k-space data into image space, thereby generating image data, - a reduction unit for generating reduced image data by removing image edge areas of the image data that lie outside the predetermined image format and border the predetermined image format, - a reverse transformation unit for retransforming the reduced image data into k-space, generating reduced k-space data, - a reconstruction unit for reconstructing magnetic resonance imaging data based on reduced k-space data.

[0031] The transformation unit preferably comprises a reconstruction function, which is preferably different from the reconstruction function of the reconstruction unit. The transformation by the transformation unit is preferably performed by a mathematically relatively simple algorithm, most preferably a Fourier transform. The algorithm used by the reconstruction unit, preferably a DRS algorithm, is preferably geared towards achieving a high quality of image generation and particularly benefits from the data reduction by the reduction unit, especially since the improved image quality is usually achieved through increased mathematical or operational effort, which can be reduced or even minimized by the preceding data reduction.

[0032] In this preferred embodiment of the image data generation device according to the invention, image reconstruction of magnetic resonance image data is performed, wherein the amount of image data in the image space is first reduced by cutting off one or more boundary areas, reduced k-space data are generated by a reverse transformation of the reduced image data, and thus the k-space data used as the basis for image reconstruction are reduced in order to simplify or accelerate subsequent reconstruction processes.

[0033] Advantageously, by reducing the amount of input data, an existing standard algorithm for increasing image resolution, preferably based on the principle of super-resolution, and in particular a DRS algorithm, can be used with considerable time savings, while maintaining image quality. This is because, if necessary, compensatory formatting is performed in the image data space and, if required, in the k-space after the super-resolution algorithm is applied, thus compensating for interpolation-induced artifacts. In particular, the reconstruction time can be made independent of the degree of oversampling, since the image data to be interpolated can be restricted to a predetermined format by the described reduction step, regardless of the degree of oversampling.

[0034] The magnetic resonance imaging system according to the invention comprises a scanning unit, a central control unit for controlling the scanning unit, and an image data generation unit according to the invention for processing k-space data from the scanning unit. The magnetic resonance imaging system according to the invention shares the advantages of the method according to the invention for generating formatted interpolated image data.

[0035] A large proportion of the aforementioned components of the image data generation device according to the invention can be implemented wholly or partially as software modules in a processor of a suitable computing system, e.g., a control unit of a magnetic resonance imaging system or a computer used to control such a system. A largely software-based implementation has the advantage that even previously used computing systems can be easily retrofitted by a software update to operate in the manner of the invention. In this respect, the problem is also solved by a corresponding computer program product with a computer program that can be directly loaded into a computing system, containing program sections to execute the steps of the inventive method for generating formatted interpolated image data when the program is executed in the computing system.In addition to the computer program itself, such a computer program product may include additional components such as documentation and / or additional components, including hardware components such as hardware keys (dongles, etc.) for using the software.

[0036] For transport to the computer system or control unit and / or for storage on or in the computer system or control unit, a computer-readable medium, such as a memory stick, a hard drive, or other portable or permanently installed data carrier, can be used, on which the program sections of the computer program that can be read and executed by a computer system are stored. The computer system may, for example, have one or more cooperating microprocessors or similar components for this purpose.

[0037] The dependent claims and the following description each contain particularly advantageous embodiments and further developments of the invention. In particular, the claims of one claim category may also be further developed analogously to the dependent claims of another claim category. Furthermore, within the scope of the invention, the various features of different embodiments and claims may also be combined to form new embodiments.

[0038] Preferably, the generation of formatted interpolated image data should include the removal of opposing image edge regions. This is because, due to the transformation of the augmented image data into k-space and a subsequent back-transformation of the interpolated k-space data into image space, the image area is "expanded" by an additional region opposite the augmented image edge region added during the generation of the augmented image data.

[0039] In one embodiment of the inventive method for the artifact-reduced generation of formatted interpolated image data, the generation of supplemented image data is carried out by one of the following procedures: - Supplementing the predetermined image border area with constant values, - Adding a border value to the predetermined image border area, - Supplementing the predetermined image border area with zero values, - Supplementing the predetermined image edge area with image data based on a reflection from an opposite image edge area or image edge.

[0040] In this context, the term "boundary value" refers to a range of pixel values ​​or voxel values, particularly grayscale values, present at the edge of the area.

[0041] A zero value is understood to be a constant value corresponding to an intensity of "zero". Such an intensity is reached when nothing is present at the location in question, or when negligible entities are present, which are, for example, represented as blacked-out elements.

[0042] Reflection of the image edge area means that the opposite edge of the image is mirrored onto the other side. This is intended to compensate for any folding of the opposite side.

[0043] In a variant of the inventive method for artifact-reduced generation of formatted interpolated image data, the image data comprise magnetic resonance image data or the image data are obtained based on magnetic resonance image data, and the step of generating image data from an image acquisition area comprises the following steps: - Receiving k-spatial data from a study area of ​​a study object, - Reconstructing magnetic resonance imaging data based on reduced k-space data.

[0044] In this variant, magnetic resonance imaging (MRI) data is used as the image data for the method according to the invention. The reconstruction of the image data, or the MRI data, can be performed using a conventional reconstruction method, such as GRAPPA, SENSE, or a simple Fourier transform. However, the reconstruction can also include more advanced reconstruction methods, such as DRB.

[0045] To accelerate the entire process, it is now preferred to remove opposing image edge areas of the magnetic resonance image data that lie outside and border the predetermined image format in order to obtain the image data. Reducing the image area reduces the amount of image data, thus accelerating subsequent image processing. This approach is particularly advantageous for the application of conventional reconstruction methods, such as GRAPPA or SENSE, or a simple Fourier transform, and contributes to accelerating the subsequent process steps.

[0046] If an advanced reconstruction method, preferably an AI-based reconstruction method, in particular DRB, is to be used, the steps for generating image data from an image acquisition area preferably comprise the following steps: - Generating image data by transforming the received k-space data into image space, - Generating reduced image data by removing image edge areas of the image data that lie outside the predetermined image format and border the predetermined image format, - Generating reduced k-space data by back-transforming the reduced image data into k-space, - Reconstructing the magnetic resonance imaging data based on the reduced k-space data.

[0047] In this particularly advantageous variant, data in the image space are eliminated before the actual advanced image reconstruction to simplify or accelerate subsequent reconstruction. Preferably, but not exclusively, the received k-space data includes oversampled k-space data. Truncating opposing image edge regions allows for a reduction in the data basis underlying the actual reconstruction of magnetic resonance imaging data. In this case, image edge regions caused by oversampling are truncated in the image space.

[0048] As already mentioned in connection with the corresponding device, reducing the amount of input data allows the use of an existing standard algorithm for increasing image resolution based on the principle of super-resolution, in particular a DRS or DRB algorithm, with significant time savings, while maintaining image quality. This is achieved because, if necessary, compensatory formatting is performed in the image data space and k-space after the super-resolution algorithm is applied, thus compensating for interpolation-induced artifacts. In particular, the reconstruction time can be made independent of the degree of oversampling, since the image data to be interpolated can be restricted to a predetermined format by the described reduction step, regardless of the degree of oversampling.

[0049] In general, it can be stated that the method according to the invention can be applied in particular, but not limited to, all types of magnetic resonance image reconstruction methods in which interpolation takes place after the actual image reconstruction.

[0050] Advantageously, the method according to the invention can be used not only for artifact-free accelerated magnetic resonance imaging, but also for the interpolation of measurement-based image data of other types. The method according to the invention can be used whenever interpolation is required to increase the resolution of such image data. For interpolation, image data is transformed into k-space via Fourier transformation, interpolated there, and then back-transformed.

[0051] The image data preferably includes image data generated by medical imaging devices, the resolution of which is to be increased by interpolation in Fourier or k-space. Such medical imaging devices include, in particular, devices of the following type: - X-ray imaging systems, - Computed tomography systems, - Ultrasound imaging systems.

[0052] In the case of using the method for oversampled magnetic resonance imaging, the oversampled k-space data are preferably obtained by oversampling in the readout direction or frequency encoding direction with a predetermined oversampling factor greater than 1. Preferably, in this variant, the removal of the image edge regions in the image space includes the removal of image edge regions caused by the oversampling.

[0053] In this variant, the generation of augmented image data is achieved by adding predetermined image data to a predetermined image border region. This involves supplementing image data in the image border region, preferably by adding zero values, through the addition of a column, preferably of zero values, whose longitudinal direction runs in the phase-encoding direction. The image border region corresponds to the position of one of the removed image border regions. However, the extent of the augmented image border region can now comprise a single column with the thickness of only one pixel or voxel, and thus be significantly smaller perpendicular to the longitudinal direction than the previously removed image border region(s).In the final formatting process for generating formatted interpolated image data, image border areas are removed. These areas were caused by the previously added border area and are positioned in the same direction as the added border area. Typically, adding a border area, followed by a transformation into k-space and then back into image space during interpolation, creates border areas on both sides, which are then removed from both sides.

[0054] If, instead, the oversampled k-space data is obtained by oversampling in the phase-encoding direction with a predetermined oversampling factor greater than 1, then the generation of augmented image data by adding predetermined image data in a predetermined image edge region preferably includes the addition of image data in the image edge region, preferably the zero values, and the addition of a row, preferably of zero values, whose longitudinal direction runs in the readout direction or frequency-encoding direction. The image edge region corresponds to the position of one of the removed image edge regions with respect to its positioning. The extent of the augmented image edge region can now comprise a single row with the thickness of only a single pixel or voxel, and thus be significantly smaller transversely to the longitudinal direction than the removed image edge region(s).In the final formatting process for generating formatted interpolated image data, the previously added image border area is removed in the same direction. Typically, adding an image border area during interpolation, followed by a transformation into k-space and another transformation back into image space, creates border areas on both sides, which are then removed on both sides.

[0055] In general terms, additions, preferably zero values, can be added in all three dimensions or in any dimension where relevant image data is present in the boundary region of an image area to prevent inversion artifacts. In addition to the directions mentioned, image data or magnetic resonance imaging (MRI) data can also be added in the z-direction of a MRI system. In the context of MRI, the z-direction refers to the direction of the rotational symmetry axis of a MRI system.

[0056] Preferably, the reconstruction of the magnetic resonance image data involves the application of an artificial intelligence-based reconstruction method. Often, the amount of input data or training data required for the training of such a method is considerable. Advantageously, according to the invention, the amount of input data is reduced, thus accelerating the training process.

[0057] The artificial intelligence-based reconstruction method preferably incorporates a deep resolve technique (see above), which is frequently used for reconstructing magnetic resonance imaging data. A particular advantage of such a deep resolve technique is that, due to the reduced size of the training data, the effort required to adapt the method to a specific type of image data can be reduced without increasing the occurrence of involution artifacts.

[0058] Preferably, the generation of interpolated k-space data includes filling the boundary regions in k-space with zero values. Filling k-space with zero values ​​increases the resolution of the reconstructed image data. Since, according to the invention, the image data has already been supplemented in the boundary region of the image space, preferably with zero values, the generation of inflection artifacts, which frequently occur after such interpolation, can be advantageously avoided, especially when the object to be imaged is larger than the field of view on which the imaging is based.

[0059] The invention is explained in more detail below with reference to the accompanying figures and exemplary embodiments. The figures show: Fig. 1. A schematic diagram showing a pictorial representation of a group of objects, some of which are not fully depicted. Fig. 2 a schematic diagram illustrating a method for generating magnetic resonance image data of an object under investigation with increased resolution according to the state of the art, Fig. 3 a schematic diagram which also illustrates a method for generating magnetic resonance image data of an object under investigation with increased resolution according to the state of the art, Fig. 4 a schematic diagram illustrating a method for generating formatted interpolated image data, in particular magnetic resonance image data, of an object under investigation according to an embodiment of the invention, Fig. 5 a flowchart illustrating a method for generating formatted interpolated image data of an object of investigation according to an alternative embodiment of the invention, Fig. 6 a flowchart illustrating partial steps of a method for generating formatted interpolated magnetic resonance image data of an object under investigation according to an alternative embodiment of the invention, Fig. 7 a schematic representation of an image data generation device according to an embodiment of the invention, Fig. 8 a schematic representation of a magnetic resonance imaging data generation device according to an embodiment of the invention, Fig. 9 a magnetic resonance imaging system according to an embodiment of the invention.

[0060] In Fig. Figure 1 shows a schematic diagram 10, which depicts a group of objects, some of which are not fully shown. The diagram in Fig. The image shown was created through image interpolation in k-space. For this process, the image to be interpolated was first transformed into k-space and then supplemented by filling in zero values ​​in the boundary region. For example, the number of pixels in a 2D image is quadrupled, thus halving the distance between the pixels.

[0061] However, this method can lead to folding artifacts, as is also the case in Fig. 1 is recognizable. For illustration, see in Fig. In image 1, a narrow vertical line EF is visible on the right side, resulting from folding the left edge of the image onto the right. Since the field of view (the field of view of the imaging system) is smaller than the dimensions of the image in Fig. In the arrangement of bright light fields or objects O shown in 1, the left edge area folds into the right side, especially in MR imaging.

[0062] In Fig. Figure 2 is a schematic diagram 20, which illustrates a method for generating magnetic resonance image data of an object of investigation with increased resolution according to the state of the art.

[0063] Fig. Figure 2 schematically illustrates the reconstruction chain for a conventional reconstruction. In step 2.I, an interpolation in k-space is performed, for example, by "zero-filling." Then, in step 2.II, a conventional reconstruction of the interpolated k-space data is carried out. Finally, in step 2.III, the reconstructed image is cropped to the desired field of view. However, the reconstruction in the Fig. The two methods mentioned above are based on the supplemented k-spatial data. Therefore, the reconstruction requires increased computational effort.

[0064] In Fig. Figure 3 is a schematic diagram 30, which illustrates a method for generating magnetic resonance image data of an object under investigation with increased resolution according to the state of the art.

[0065] In step 3.I, a Fourier transform is performed on the k-space data (RD) generated by oversampling into image space. Within image space, image edge regions attributable to oversampling are clipped, and the resulting reduced image data (RBD) is then transformed back into k-space via Fourier transformation, generating further reduced k-space data (RRD). Subsequently (see step 3.II), the actual image reconstruction (e.g., using GRAPPA, SENSE, DRB, etc.) can be performed based on these reduced k-space data (RRD).

[0066] In step 3.II, an "advanced" reconstruction of image data BD is performed based on the reduced k-space data RRD. The "advanced" reconstruction is based on the in Fig. Figure 3 illustrates a reconstruction method that uses artificial intelligence to reconstruct image data (BD). As already mentioned, a "conventional" reconstruction can also be used as an alternative. However, with the aforementioned "advanced" reconstruction, the effect of reducing the data set by cropping image edge areas is particularly pronounced. Since the data set for reconstruction has been reduced, the time required to reconstruct the image data (BD) is also reduced compared to a reconstruction based on unreduced k-space data (RD).

[0067] In step 3.III, the image data undergoes a Fourier transformation back into k-space, followed by interpolation in k-space by filling the k-space with zero values, a process also known as "zero-filling." This generates augmented k-space data ERD.

[0068] In step 3.IV, the supplemented k-space data ERD generated in step 3.III is transformed again into the spatial space, resulting in interpolated image data IBD.

[0069] In Fig. Figure 4 is a schematic diagram 40 illustrating a method for generating formatted and interpolated magnetic resonance image data of a test object with increased resolution according to an embodiment of the invention. The two steps 4.1 and 4.2 proceed analogously to those described in Figure 4.1. Fig. The three steps shown are 3.I and 3.II. In step 4.I, a Fourier transform is performed on the raw data RD, acquired through oversampling, into image space, thereby generating image data BD. In image space, image edge areas attributable to oversampling are clipped, and the resulting reduced image data is then transformed back into k-space via Fourier transformation, producing reduced k-space data RRD.

[0070] In step 4.II, an "advanced" reconstruction of magnetic resonance imaging (MBD) data is performed based on the reduced k-space data (RRD). This "advanced" reconstruction is based on the... Fig. Figure 3 illustrates a reconstruction method that uses artificial intelligence to reconstruct magnetic resonance imaging (MRI) data. Since the data set for reconstruction has been reduced, the time required to reconstruct the MRI data has also been reduced.

[0071] In step 4.III, zero values ​​are inserted at the edge of the magnetic resonance image data (MBD) in the spatial or image space, resulting in supplemented image data (EBD). The orientation of the edge region corresponds to the orientation of the edge regions that were clipped in step 4.1. Multiple opposing or non-opposite edge regions can also be added if objects at different edges were not fully imaged.

[0072] In step 4.IV, analogous to step 3.III, a Fourier transform of the augmented image data EBD back into k-space is performed, followed by interpolation in k-space by filling k-space with zero values, a process also known as "zero-filling." This generates augmented k-space data ERD.

[0073] In step 4.V, the added k-space data ERD are transformed back into local space, generating extended and interpolated image data IBD.

[0074] Furthermore, in step 4.V, edge areas containing zero values ​​due to the addition of zero values ​​in step 4.III are clipped, generating formatted interpolated FIBD image data. This differs from the procedure in Fig. 3. Due to the addition of zero values ​​in step 4.III, inflection artifacts in the formatted interpolated image data FIBD can be avoided.

[0075] In Fig. Figure 5 shows a flowchart 500 illustrating a method for generating formatted interpolated image data of an object of investigation with increased resolution according to an alternative embodiment of the invention.

[0076] In step 5.I, image data (BD) is generated from an image acquisition area. This image data (BD) can include, for example, photographic image data or magnetic resonance imaging (MBD) data.

[0077] In step 5.II, supplementary image data EBD is generated by adding zero values ​​to the image data BD in an image border region where an object touches the image edge. This measure is crucial to avoid inflection artifacts during subsequent interpolation.

[0078] In step 5.III, augmented k-space data ERD is generated by transforming the augmented image data EBD into k-space or Fourier space via Fourier transformation. The transformation into Fourier space is performed because the actual interpolation to increase image resolution takes place in Fourier space.

[0079] In step 5.IV, interpolated k-space data IRD is generated by interpolating the supplemented k-space data ERD in k-space.

[0080] Subsequently, in step 5.V, interpolated image data IBD is generated through image reconstruction based on the interpolated k-space data IRD.

[0081] In step 5.VI, formatted interpolated image data FIBD is generated by removing edge areas of the interpolated image data IBD.

[0082] In Fig. Figure 6 shows a flowchart illustrating the first steps of a method for generating formatted, interpolated image data of an object under investigation with increased resolution according to an alternative embodiment of the invention, which is particularly useful when applying the method according to the invention to magnetic resonance imaging data. The steps shown in Figure 6 illustrate the first steps of a method for generating formatted, interpolated image data of an object under investigation with increased resolution according to an alternative embodiment of the invention, which is particularly useful when applying the method according to the invention to magnetic resonance imaging data. Fig. The 6 steps shown are from the one in Fig. The 5 shown step 5.I includes.

[0083] At the in Fig. In the illustrated embodiment 6, in step 5.Ia, oversampled raw data RD of an object under investigation are acquired by a scanning unit of a magnetic resonance imaging system.

[0084] In step 5.Ib, the raw data RD from k-space is transformed into image space. In image space, image edge areas attributable to oversampling are clipped, and the resulting reduced image data RBD is then transformed back into k-space via Fourier transformation, producing reduced k-space data RRD. This process reduces the amount of k-space data that will later serve as the basis for image reconstruction.

[0085] In step 5.Ic, magnetic resonance imaging (MBD) data is reconstructed based on the reduced k-space data (RRD). This MBD data is then combined with the data in Fig. The 5 illustrated methods are further processed as image data BD.

[0086] In Fig. Figure 7 illustrates a schematic representation of an image data generation device 70 according to an embodiment of the invention.

[0087] The image data generation unit 70 comprises an image acquisition unit 71 for receiving image data BD from an image acquisition area.

[0088] The image data generation unit 70 includes a supplementary unit 72 for generating supplementary image data EBD by adding zero values ​​to the image data BD in an image border region. The zero values ​​are added to prevent the occurrence of inflection artifacts in the opposite image border region during subsequent interpolation.

[0089] Part of the image data generation device 70 is also a transformation unit 73 for generating supplementary k-space data ERD by transforming the supplementary image data EBD into k-space.

[0090] The image data generation device 70 according to the invention also comprises an interpolation unit 74 for generating interpolated k-space data IRD by interpolating the supplemented k-space data ERD.

[0091] Furthermore, the image data generation device 70 according to the invention has an image generation unit 75 for generating interpolated image data IBD by means of image reconstruction based on the interpolated k-space data IRD.

[0092] Part of the image data generation device 70 according to the invention is also a formatting unit 76 for generating formatted interpolated image data FIBD by removing edge areas of the interpolated image data IBD.

[0093] In Fig. Figure 8 illustrates a magnetic resonance imaging data generation device 80 according to an embodiment of the invention.

[0094] The magnetic resonance imaging data generation unit 80 represents a further development of the one already in Fig. Figure 7 illustrates the image data generation device 70 and includes a special image acquisition unit 71, which has the following subunits 71a, 71b, 71c: The image acquisition unit 71 includes an input interface 71a for receiving oversampled k-space data RD from an investigation area ROI of an investigation object O, which is scanned as part of a magnetic resonance imaging.

[0095] Part of the image acquisition unit 71 is also a reduction unit 71b for generating reduced k-space data RRD by removing one of the areas in the image space associated with oversampling.

[0096] Finally, the image acquisition unit 71 also includes a reconstruction unit 71c for reconstructing magnetic resonance imaging data (MBD) based on the reduced k-space data (RRD).

[0097] The remaining in Fig. The components 72, 73, ..., 76 shown in the diagram correspond to those in Fig. 7 components already illustrated 72, 73, ..., 76 of the in Fig. 7 illustrated image data generation device 70.

[0098] In Fig. Figure 9 shows a magnetic resonance imaging system 90, also referred to as an MR system, according to an embodiment of the invention, which includes a magnetic resonance image data generation device 80 according to an embodiment of the invention, as described in Figure 90. Fig. 8 is illustrated, includes.

[0099] The magnetic resonance imaging system 90 comprises, on the one hand, the actual magnetic resonance scanner or magnetic resonance scanning unit 102 with an examination room 103 or patient tunnel, into which a patient O, or in this case a patient or test subject, in whose body there is, for example, a specific organ to be imaged, can be moved on a couch 108.

[0100] The magnetic resonance scanner 102 is equipped in the usual manner with a base field magnet system 104, a gradient system 106, an RF transmitting antenna system 105, and an RF receiving antenna system 107. In the illustrated embodiment, the RF transmitting antenna system 105 is a whole-body coil permanently installed in the magnetic resonance scanner 102, whereas the RF receiving antenna system 107 consists of local coils to be positioned on the patient or subject (in Fig. 9 (symbolized only by a single local coil). In principle, however, the whole-body coil 105 can also be used as an RF receiving antenna system and the local coils 107 as an RF transmitting antenna system, provided that these coils can each be switched to different operating modes.

[0101] The MRI system 90 also includes a central control unit 113, which is used to control the MRI system 90. This central control unit 113 comprises a sequence control unit 114 for pulse sequence control. This unit controls the temporal sequence of radio frequency pulses (RF pulses) and gradient pulses depending on a selected imaging sequence PS according to a pulse sequence scheme PSS. Such an imaging sequence PS, or the pulse sequence scheme PSS underlying the imaging sequence PS, can be predefined, for example, within a measurement or control protocol P. Typically, various control protocols P for different measurements are stored in a memory 119 and can be selected by an operator (and changed if necessary) and then used to perform the measurement.

[0102] For the output of individual RF pulses, the central control unit 113 includes a high-frequency transmitter 115, which generates and amplifies the RF pulses and feeds them into the RF transmitting antenna system 105 via a suitable interface (not shown in detail). For controlling the gradient coils of the gradient system 106, the control unit 113 includes a gradient system interface 116. The sequence control unit 114 communicates appropriately, e.g., by transmitting sequence control data SD, with the high-frequency transmitter 115 and the gradient system interface 116 for transmitting the pulse sequences PS. The control unit 113 also includes a high-frequency receiver 117 (which also communicates appropriately with the sequence control unit 114) to acquire magnetic resonance signals received by the RF transmitting antenna system 107 in a coordinated manner.

[0103] The central control unit 113 also includes a magnetic resonance imaging data generation unit 80 according to the invention, which contains the Fig. 8 shows a detailed illustration of the structure.

[0104] The magnetic resonance imaging data generation unit 80 is configured to take the acquired data after demodulation and digitization as raw data or k-space data RD and to reconstruct formatted interpolated magnetic resonance imaging data FIBD from it. This magnetic resonance imaging data FIBD can then be stored, for example, in a memory 119.

[0105] The central control unit 113 can be operated via a terminal with an input unit 111 and a display unit 109, thus enabling a single operator to control the entire MRI system 90. MRI image data (FIBD) can also be displayed on the display unit 109, and measurements can be planned and started using the input unit 111, possibly in combination with the display unit 109. In particular, suitable control protocols with appropriate measurement sequences can be selected and, if necessary, modified as described above.

[0106] The MR system 90 according to the invention, and in particular the control unit 113, can also have a large number of other components, not shown in detail here, but which are usually present on such devices, such as a network interface to connect the entire system to a network and to be able to exchange raw data RD and / or image data or parameter maps, but also other data, such as patient-relevant data or control protocols.

[0107] How suitable raw data (RD) can be acquired by irradiating RF pulses and generating gradient fields, and how MR image data can be reconstructed from them, is generally known to those skilled in the art and will not be explained in detail here.

[0108] From the above description, it is clear that the invention effectively provides possibilities to improve a method for generating magnetic resonance image data with regard to the required time duration or the image artifacts that occur.

[0109] It should be noted that the features of all embodiments or further developments disclosed in figures can be used in any combination.

[0110] Finally, it should be noted once again that the detailed methods and setups described above are exemplary embodiments and that the basic principle can be varied in many ways by those skilled in the art without departing from the scope of the invention, insofar as it is defined by the claims. For the sake of completeness, it should also be noted that the use of the indefinite articles "a" or "an" does not preclude the possibility that the features in question may be present multiple times. Likewise, the term "unit" does not preclude the possibility that it consists of several components, which may also be spatially distributed. Regardless of the grammatical gender of a particular term, persons of male, female, or other gender identities are included.

Claims

[1] Method for generating formatted interpolated image data (FIBD), comprising the steps: - Generating image data (BD) from an image acquisition area with a predetermined image format that represents the image acquisition area, - Generating augmented image data (EBD) by adding predetermined image data in a predetermined image border area that lies outside the predetermined image format and borders the predetermined image format, - Generating augmented k-space data (ERD) by transforming augmented image data (EBD) into k-space, - Generating interpolated k-spatial data (IRD) by interpolating the augmented k-spatial data (ERD), which generates unmeasured k-spatial data based on the augmented k-spatial data (ERD), - Generating interpolated image data (IBD) by image reconstruction based on interpolated k-space data (IRD), wherein the interpolated image data (IBD) includes image edge areas that lie outside the predetermined image format, - Generating formatted interpolated image data (FIBD) by removing the image border areas that lie outside the predetermined image format from the interpolated image data (IBD), wherein the image data (BD) includes magnetic resonance imaging (MBD) data or is obtained based on magnetic resonance imaging (MBD) data, and the step of generating image data (BD) from an image acquisition area includes the steps: - Receiving k-space data (RD) from a study area of ​​a study object (O), - Reconstructing magnetic resonance imaging (MBD) data based on k-space data (RRD); and whereby The image data (BD) are obtained by removing predetermined image edge areas of the magnetic resonance image data (MBD) that lie outside the predetermined image format and border the predetermined image format. or whereby The step of generating image data (BD) from an image acquisition area includes the following steps: - Generating reduced image data (RBD) by transforming the received k-space data (RD) into image space and removing image border areas in image space that lie outside the predetermined image format and border the predetermined image format, - Back-transforming the reduced image data (RBD) into k-space, generating reduced k-space data (RRD), - Reconstructing magnetic resonance imaging (MBD) data based on reduced k-space data (RRD). [2] Method according to claim 1, wherein the remote image edge areas comprise image edge areas opposite each other which are arranged in the same direction as the predetermined image edge area. [3] Method according to claim 1 or 2, wherein the generation of augmented image data (EBD) is carried out by one of the following procedures: - Supplementing the predetermined image border area with constant values, - Supplementing the predetermined image border area with the border value, - Supplementing the predetermined image border area with zero values, - Supplementing the predetermined image edge area with image data from an opposite image edge area based on a reflection. [4] Method according to one of the preceding claims, wherein the received k-space data (RD) comprise oversampled k-space data (RD) and the oversampled k-space data (RD) were obtained by oversampling in the readout direction and / or phase-encoding direction with a predetermined oversampling factor which includes a value greater than 1 and the removal of the image edge areas comprises the removal of image edge areas caused by the oversampling. [5] Method according to claim 4, wherein the generation of augmented image data (EBD) is carried out by adding predetermined image data in a predetermined image border area. - in the case of oversampling in the read-out direction, this involves adding a column that runs in the phase-encoding direction, - in the case of oversampling in phase-encoding direction, this includes adding a line that runs in the readout direction. [6] Method according to any of the preceding claims, wherein the reconstruction of the magnetic resonance imaging (MRI) data comprises the application of an artificial intelligence-based reconstruction method. [7] Image data generation device (70, 80), comprising: - an image acquisition unit (71) for receiving image data (BD) from an image acquisition area with a predetermined image format depicting the image acquisition area, - an augmentation unit (72) for generating augmented image data (EBD) by adding predetermined image data in a predetermined image border area which lies outside the predetermined image format and is adjacent to the predetermined image format, - a transformation unit (73) for generating augmented k-space data (ERD) by transforming the augmented image data (EBD) into k-space, - an interpolation unit (74) for generating interpolated k-spatial data (IRD) by interpolating the augmented k-spatial data (ERD), by which unmeasured k-spatial data are generated on the basis of the augmented k-spatial data (ERD), - an image generation unit (75) for generating interpolated image data (IBD) by image reconstruction based on the interpolated k-space data (IRD), wherein the interpolated image data (IBD) include image edge regions which lie outside the predetermined image format, - a formatting unit (76) for generating formatted interpolated image data (FIBD) by removing the image edge areas which lie outside the predetermined image format in the interpolated image data (IBD), wherein the image data (BD) comprise magnetic resonance image data (MBD) and the image acquisition unit (71) includes: - an input interface (71a) for receiving k-space data (RD) from a survey area (ROI) of a survey object (O), - a reduction unit (71b) for generating reduced k-space data (RRD) by transforming the k-space data (RD) into image space and removing image edge areas of the image data which lie outside the predetermined image format and border the predetermined image format, thereby generating reduced image data (RBD), and for back-transforming the reduced image data (RBD) into k-space, thereby generating reduced k-space data (RRD), - a reconstruction unit (71c) for reconstructing magnetic resonance image data (BD) based on the reduced k-space data (RRD), and wherein the image data generation device (70, 80) is configured to generate formatted interpolated image data (FIBD) according to a method according to one of the preceding claims. [8] Magnetic resonance imaging system (90), comprising: - a scanning unit (102), - a central control unit (113) for controlling the scanning unit (102), - an image data generation device (70, 80) according to claim 7 for processing k-space data (RD) of the scanning unit (102). [9] Computer program product comprising instructions which, when the program is executed by a computer, cause it to perform the steps of the method according to any one of claims 1 to 6. [10] Computer-readable storage medium comprising instructions which, when executed by a computer, cause it to perform the steps of the method according to claims 1 to 6.

Images