Automatic optimization of MR examination protocols

Abstract

The present invention relates to a method for optimizing an examination protocol for performing MR image acquisitions from a patient's body. The object of the present invention is to facilitate the efficient implementation of accelerated examination protocols that are true or very close substitutes for examination protocols already existing in clinical practice. The method of the present invention comprises the steps of providing an examination protocol comprising specifications of two or more imaging sequences, executing in a computer at least one algorithm that processes said examination protocol as an input for performing an optimization with respect to the execution speed of the examination protocol, taking into account diagnostic relevance weights assigned to the imaging sequences contained in the examination protocol, making an output representative of said optimized examination protocol available to a user and / or performing an MR image acquisition on an MR scanner based on said optimized examination protocol.

Description

[Technical field]

[0001] The present invention relates to the field of magnetic resonance (MR) and finds particular application in conjunction with MR imaging methods and MR scanners for diagnostic purposes, and will be described with particular reference thereto. [Background technology]

[0002] MR imaging, which utilizes the interaction between a magnetic field and atomic nuclear spins to form two- or three-dimensional images, is widely used today, especially in the field of medical diagnostics, for imaging of soft tissues, since it is superior in many ways to other imaging methods, does not require ionizing radiation, and is usually non-invasive.

[0003] Indeed, MR imaging is often the most sensitive or appropriate technique for clinical diagnosis, but long acquisition times limit its use due to costs and considerations of patient comfort and compliance. Once the image field of view and resolution are selected, the minimum MR image acquisition time is usually determined by the amount of raw data that must be acquired to meet the Nyquist criterion.

[0004] The expected post-COVID surge in patients in radiology clearly demonstrates the need for highly accelerated imaging procedures and reduction to essential features of examination protocols in the presence of very high patient volumes. A similar situation arises when the number of cases to be examined increases when, for some reason, the imaging system has not been in operation for a period of time.

[0005] The main development in accelerated MR acquisition in the past few years has been parallel imaging (PI). In PI techniques, the spatial sensitivity profiles of the multiple receive coils of the used MR scanner provide additional information for the reconstruction, allowing successful reconstruction from undersampled k-space data. Two common PI methods are Generalized Autocalibrated Partially Parallel Acquisition (GRAPPA) and Sensitivity Encoding (SENSE). A particularly successful method for accelerating MR image acquisition that uses prior sparsity and incoherent sampling is Compressive Sensing (CS). CS reconstruction is an iterative reconstruction method that estimates MR images from (strongly) undersampled MR signal data by enforcing data consistency and exploiting the assumption that medical images are compressible (i.e., sparse in some spatial transform domain). The drawback of these known approaches to accelerated acquisition is that the reconstruction problem that PI and CS methods were designed to solve is itself time consuming. Although the reconstruction can be done offline, clinical scenarios require speed not only for the actual MR signal acquisition, but also for the reconstruction of the individual scans.

[0006] Meanwhile, artificial intelligence (AI)-assisted acceleration methods for MR imaging are rapidly emerging (see review by Lin et al. (J. Magn. Reson. Imaging 2021, 53:1015-1028)), which disclose methods to significantly reduce scan time while significantly preserving image quality (by enabling successful reconstruction from strongly undersampled k-space data). AI-based image acquisition and reconstruction techniques, especially those based on machine learning (or deep learning), are useful because they perform optimization across many training images before solving the reconstruction for any particular given image. They can exploit common features of the anatomical structures, as well as the structure of undersampling artifacts present across the training images. Thus, unlike CS reconstructions, which require long computational times for each new scan, deep learning-based reconstruction models can complete the reconstruction in seconds. For these reasons, much research is currently being done on the use of deep learning-based approaches to achieve accelerated, high-quality MR image reconstruction.

[0007] However, these AI methods are difficult to implement and, to be generalizable, usually require large training datasets. Moreover, AI-based image reconstruction can only be used for the image contrasts on which the respective reconstruction models were trained, which generally do not match exactly what radiologists are used to seeing from individual standard examination protocols and imaging sequence definitions. In situations where an examination needs to be instantly accelerated, clinics cannot simply replace standard examination protocols with improved and accelerated AI-based techniques (if any), since the image contrasts generated depend on the trained AI models and generally do not match exactly what the standard protocols used in daily practice look like. Often, each facility has several most preferred scans and procedures that differ between clinics, although standardization is often desired.

[0008] US patent application US2008 / 197841 (D1) relates to the assignment of acquisition parameters for MR imaging with a specific physical quantity, i.e. the acquisition parameters are set corresponding to a specific contrast or tissue parameter mapping. The optimization of the acquisition parameters is targeted to the specific physical quantity subject to constraints related to the desired imaging environment (total imaging time, power deposition (i.e. SAR)) or hardware constraints. Summary of the Invention [Problem to be solved by the invention]

[0009] From the above, it is easily understood that there is a need for improved methods of optimizing MR examination protocols with regard to acquisition speed. It is therefore an object of the present invention to facilitate the efficient implementation of accelerated examination protocols that are true or very close replacements for examination protocols already existing in clinical practice. It should be possible to provide each individual clinic with a specific optimized version of the respective standard examination protocol. [Means for solving the problem]

[0010] According to the present invention, a method for optimizing an examination protocol for performing MR image acquisitions from a patient's body is disclosed, the method comprising: providing an examination protocol including specifications for two or more imaging sequences; - executing, in a computer, at least one algorithm which processes the examination protocol as an input for performing an optimization with respect to the speed of execution of the examination protocol, taking into account the diagnostic relevance weights assigned to the imaging sequences contained in the examination protocol; making available an output representative of said optimized examination protocol to a user and / or performing said MR image acquisition on an MR scanner based on said optimized examination protocol. has.

[0011] The proposed method automatically evaluates and exploits the optimization potential of any examination protocol. Thus, the present invention achieves exactly the acceleration required for the examination protocol individually. For each imaging sequence included in the examination protocol, the clinician (e.g., radiologist) can select a weighting factor (e.g., 0≦w) for use by the algorithm in the optimization method. i The diagnostic relevance is specified as a weighting factor of ≦1. The diagnostic relevance represents the relative importance of obtaining good image quality from each individual imaging sequence for a particular diagnostic purpose. For example, if knowledge of the physiological function of the examined organ of the patient is more important than the anatomical aspects, the clinician may decide that the image quality of an imaging sequence providing diffusion-weighted and T2-weighted contrast in an examination protocol should have higher relevance than the image quality of an imaging sequence providing T1-weighted contrast scans in the same examination protocol. By defining the diagnostic relevance weighting, it is possible to consider not only the individual scans or contrasts that are important, but also their combinations (in the sense that the combination is often more than the sum of the parts).

[0012] The result of the optimization, i.e. the accelerated examination protocol, is finally made available to, for example, a clinician for further use. The user may be informed of the result of the automatic optimization, for example by a graphical representation of the generated output examination protocol on a display monitor. The notification typically takes the form of a suggestion to the user. The final decision as to which examination protocol will be executed is taken by the user. The user may apply modifications to the proposed examination protocol based on his / her individual knowledge and experience before starting the actual MR image acquisition.

[0013] The examination protocol may comprise different acquisition parameters specifying two or more imaging sequences. Such parameters mainly depend on the respective clinical question. The parameters may determine different physical variables, such as the voltage applied to the components, for example the radio frequency antenna of the MR scanner used. The parameters may also be the duration between two radio frequency pulses and / or magnetic field gradients to be activated. For example, by defining the resolution, the parameters may characterize the MR images to be acquired. Further parameters may determine the position, orientation and size of the field of view as well as the contrast weighting (for example T1 weighting, T2 weighting, diffusion weighting, etc.) of the MR image acquisition. In an embodiment of the present invention, the optimization involves the modification of the acquisition parameters associated with at least one of the imaging sequences so as to accelerate the instruction execution of the imaging sequences. Similarly, the optimization may include the modification of the k-space sampling pattern and / or the image reconstruction model associated with at least one of the imaging sequences. For example, the algorithm may replace a given imaging sequence with a PI or CS pendant. This obviously requires not only a modification of the k-space sampling pattern, but also a modification of the image reconstruction model (i.e. a detailed and complete definition of the reconstruction procedure related to a particular optimized imaging sequence) necessary to compute high-quality MR images from undersampled k-space data. To this end, the invention proposes the concept of a model-weighted optimized examination protocol, which not only specifies the accelerated imaging sequence to be performed, but also adds a reconstruction model corresponding to the outputs made available by the algorithm. Once such a model-weighted examination protocol is provided, the MR reconstruction engine used for the command execution of an MR examination based on the optimized examination protocol has all the information necessary to reconstruct an MR image from the acquired MR signal data.

[0014] Similarly, the algorithm can replace a given imaging sequence with an AI pendant, and the optimization can include assigning an associated AI-based image reconstruction model to the respective imaging sequence. The AI-based image reconstruction model can use machine learning methods, and the optimization involves training the AI-based image reconstruction model and incorporating the trained artificial intelligence-based image reconstruction model into the output to provide a corresponding "ready-to-use" model-enhanced examination protocol, as described above. The selection of AI-based image acquisition and reconstruction for a particular imaging protocol depends on the exact type of imaging sequence definition and the desired image contrast, as well as the availability of suitable training data (e.g., exemplary MR images acquired with the original examination protocol). It is an insight of the present invention that individually designed and trained AI reconstruction models work best when applied with the specific imaging sequence for which they were trained. Thus, the trained AI reconstruction model should not be a fixed part of the reconstruction engine used for the MR examination. Instead, the trained AI reconstruction model should be provided as an output of the protocol optimization together with the individual optimized imaging sequence definition. While standard testing protocols include only acquisition instructions for the imaging sequence involved, the model-enhanced testing protocol of the present invention also includes an AI reconstruction model trained for reconstruction using the particular imaging sequence, ideally along with a set of detailed instructions specifying how the reconstruction model should be used.

[0015] In another embodiment, the optimization may involve assigning an image reconstruction model to at least one of the imaging sequences that uses MR signal data acquired by executing at least one of the other imaging sequences. Due to the acceleration of MR image acquisition, it means that the reconstruction of MR images related to some of the imaging sequences included in the input examination protocol can be based on MR signal data acquired by the other imaging sequences. The introduction of such a sharing of MR signal data between different imaging sequences may be part of the optimization according to the invention. Possibly, the command execution of at least one of the imaging sequences is completely omitted in the optimized examination protocol, and an image reconstruction model is added to the optimized examination protocol in order to synthesize MR images related to the omitted imaging sequence from MR signal data acquired by command execution of at least one of the other imaging sequences. If the optimization algorithm finds that it is not necessary to execute all the imaging sequences included in the examination protocol, but that it is necessary to synthesize MR images associated with one or more of the imaging sequences from MR signal data acquired by one or more of the other imaging sequences, the algorithm may decide to omit the execution of the respective imaging sequence, resulting in a corresponding significant increase in the speed of execution of the examination protocol. For example, a suitable reconstruction model may synthesize an estimate of a T2-weighted MR image from a survey image and a T1-weighted MR image. Such a method of optimization of an examination protocol may also require modifying the order in which the imaging sequences are performed, since the reconstruction of MR images from some (optimized) imaging sequences may be based on MR images from previous imaging sequences in the examination protocol.

[0016] In another embodiment, at least one algorithm further takes into account a quality trade-off weighting, which is a user-specified trade-off between the quality of MR images resulting from an optimized examination protocol and the increase in execution speed achieved by the optimization. This can be expressed as a further weighting factor 0≦p≦1, where a larger value of p results in a higher acceleration with more compromise in terms of image quality. If more than one optimized version of the examination protocol is desired (e.g., due to different acceleration stages), a list of several weighting factors can be specified, each weighting factor resulting in the generation of one optimized examination protocol. In a practical implementation, the algorithm can determine several optimization options and select one of the optimization options according to an objective function that assigns a higher weighting to the optimization option, resulting in a higher image quality of MR images associated with imaging sequences having a higher diagnostic relevance weighting and at the same time a lower image quality of MR images associated with imaging sequences having a lower diagnostic relevance weighting.

[0017] The inventive method described so far can be executed by a computer used in a clinical workflow. The computer may be a control computer of an MR scanner comprising at least one main magnet coil for generating a main magnetic field in an examination volume, a plurality of gradient coils for generating switched magnetic field gradients in different spatial directions in the examination volume, and at least one RF coil for generating RF pulses in the examination volume and / or for receiving MR signals from a patient's body arranged in the examination volume, the control computer controlling the temporal succession of the RF pulses and the switched magnetic field gradients according to an examination protocol and a reconstruction unit for reconstructing MR images from the received MR signals. The inventive method can be implemented by corresponding programming / software of the control computer of the MR scanner.

[0018] The method of the present invention can be advantageously implemented in most MR environments in current clinical use. For this purpose, it is simply necessary to utilize a computer program that controls the computer used to carry out the above-mentioned method steps of the present invention. The computer program may be present on a data carrier or in a data network, so as to be downloaded for installation in a computer, for example the control computer of an MR scanner.

[0019] The accompanying drawings disclose preferred embodiments of the invention, but it is to be understood that the drawings are designed for illustrative purposes only and are not intended as a definition of the limits of the invention. [Brief description of the drawings]

[0020] [Figure 1] 1 shows an MR scanner for carrying out the method of the invention; [Diagram 2] 1 illustrates an embodiment of a method of the present invention as a flow chart. [Diagram 3] 1 illustrates a conceptual overview of the enhanced model checking protocol according to the present invention; DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0021] With reference to Figure 1, there is shown an MR scanner 1. The scanner comprises superconducting or resistive main magnet coils 2 such that a substantially uniform and time-constant main magnetic field B0 is generated along the z-axis through the examination volume. The apparatus further comprises a set (first, second and, if applicable, third order) of shimming coils 2', the current through each individual shimming coil of the set 2' being controllable to minimize B0 deviations within the examination volume.

[0022] The magnetic resonance generation and manipulation system applies a series of RF pulses and switched magnetic field gradients to invert or excite nuclear magnetic spins, induce magnetic resonance, refocus magnetic resonance, manipulate magnetic resonance, spatially and otherwise encode magnetic resonance, saturate spins, etc. to perform MR imaging.

[0023] Most specifically, gradient pulse amplifiers 3 apply current pulses to selected ones of whole-body gradient coils 4, 5, and 6 along the x, y, and z axes of the examination volume. A digital RF frequency transmitter 7 transmits RF pulses or pulse packets to body RF coil 9 via transmit / receive switch 8, which transmits RF pulses into the examination volume. A typical MR pulse sequence consists of packets of short duration RF pulse segments taken together with optional applied magnetic field gradients to achieve selected manipulation of nuclear magnetic resonance. The RF pulses are used to saturate resonance, excite resonance, invert magnetization, refocus resonance, or manipulate resonance to select a portion of a body 10 located within the examination volume. The MR signals are also picked up by the body RF coil 9.

[0024] For the generation of MR images of limited regions of the body 10 by parallel imaging, a set of local array RF coils 11, 12, 13 are positioned adjacent to the region selected for imaging. The array coils 11, 12, 13 may be used to receive MR signals induced by the body coil RF transmissions.

[0025] The resulting MR signals are picked up by the body RF coil 9 and / or array RF coils 11, 12, 13 and demodulated by a receiver 14, which preferably includes a preamplifier (not shown). The receiver 14 is connected to the RF coils 9, 11, 12, 13 via a transmit / receive switch 8.

[0026] A control computer 15 controls the currents through the shimming coil 2' and the gradient pulse amplifier 3 and the transmitter 7 to generate any of a number of MR pulse sequences, e.g. echo planar images (EPI), echo volume images, gradient and spin echo images, fast spin echo images, etc., according to a predefined examination protocol. For each examination protocol, the receiver 14 receives one or several MR data lines in rapid succession following each RF excitation pulse. The acquisition system 16 performs an analog-to-digital conversion of the received signals and converts each MR data line into a digital format suitable for further processing. In modern MR scanners, the acquisition system 16 is a separate computer dedicated to the acquisition of raw image data. The examination protocol determines the imaging sequence, i.e. the MR pulse sequence, including the frequency and temporal succession of the radio frequency pulses and / or magnetic field gradients to be activated. The parameters included in the examination protocol thereby characterize the MR images to be acquired, in particular with regard to the position, orientation and size of the field of view, as well as contrast weighting.

[0027] Finally, the digital raw image data is reconstructed into an image representation by a reconstruction processor 17, which applies a Fourier transform or other appropriate reconstruction algorithm such as SENSE or SMASH. The MR image may represent a planar slice through the patient, an array of parallel planar slices, a three-dimensional volume, etc. The image is then stored in an image memory, where it can be accessed to convert slices, projections, or other portions of the image representation into an appropriate format for visualization, for example, via a video monitor 18, which provides a human-readable display of the resulting MR image.

[0028] 2, the method of the present invention begins with the selection of an examination protocol to be optimized. The examination protocol includes specifications and parameters of two or more imaging sequences.

[0029] For each imaging sequence included in the examination protocol, the clinician (e.g., radiologist) assigns a weighting factor 0≦w iWe define diagnostic relevance as ≦1. Diagnostic relevance represents the relative importance of obtaining good image quality from each individual imaging sequence for a particular diagnostic purpose.

[0030] In the next step, the clinician defines a list of quality trade-off weights for the optimization process. Each of these specifies a trade-off between the quality of the MR images resulting from the optimized examination protocol and the increase in execution speed achieved by the optimization. This can be expressed as further weighting factors 0≦p≦1, where larger values ​​of p result in higher acceleration with more compromise in terms of image quality. In the illustrated embodiment, more than one optimized version of the selected examination protocol is desired (e.g., to evaluate different acceleration stages), and a list of several weighting factors is specified, each resulting in the generation of one optimized examination protocol.

[0031] Optionally, a set of exemplary MR images is acquired using the examination protocol to be optimized, or alternatively, previously stored images may be collected that can later be used in the optimization process to train the AI ​​reconstruction model.

[0032] As a next step, the examination protocol with the diagnostic relevance weights, the trade-off weights, and the example image data is provided as input to an algorithm executed on a computer to perform an optimization on the execution speed of the examination protocol. The algorithm evaluates several acceleration options. Several different AI-based acceleration techniques are used to accelerate the examination protocol. The type of acceleration technique feasible for a particular examination protocol depends on the exact type of imaging sequence and image contrast, as well as the availability of training data (e.g., example images acquired in the original protocol). Three main classes of acceleration techniques can be identified as follows:

[0033] AI-based image reconstruction.

[0034] The reconstruction of one or more images from MR signal data acquired by one or more of the imaging sequences included in the examination protocol is supported by a trained AI reconstruction model, for example by computing diagnostic quality images from a heavily undersampled k-space data set.

[0035] AI-based generation of additional contrast.

[0036] The trained AI reconstruction model synthesizes MR images associated with one of the imaging sequences based on the MR signal data or images of other imaging sequences, thereby replacing or reducing the amount of MR signal data to be acquired. These synthetic contrasts can be used directly for diagnosis if there is a high degree of confidence that the synthetic images are correct. Alternatively, they can serve as prior knowledge in the reconstruction of highly undersampled acquisitions of specific contrasts.

[0037] Acquisition parameters are tuned and / or PI or CS based techniques are applied without using AI weighted image reconstruction.

[0038] An acceleration option can be viewed as a particular set of acceleration techniques that can be applied to the examination protocol being optimized. A set of acceleration options is proposed by evaluating all possible combinations of acceleration techniques for the protocol.

[0039] The best choice for M acceleration is selected according to an objective function that includes weightings for acceleration and quality aspects, a preferred embodiment of such an objective function is described below.

[0040] The number of imaging sequences in the examination protocol is N. i Let be the scan duration of sequence i.

[0041] Acceleration factor A for acceleration option k kis defined as the ratio of the total scan duration without optimization to the total scan duration with option k. TIFF2025505492000002.tif98 shows the duration of scan i at acceleration option k. A k teeth, It is represented as TIFF2025505492000003.tif1140.

[0042] Relative quality estimator 0≦q k,i ≦1 represents the expected diagnostic image quality of the accelerated imaging sequence i of the acceleration option k relative to the image quality of the original imaging sequence. There are objective quality criteria that can be attributed to the acquired images relative to the images acquired by the original sequence, so that it is possible to determine the relative quality estimator fully automatically. However, in many practical cases, the quality value may depend to some extent on the subjective impression of the radiologist. Therefore, a possible implementation may be the pre-determination of the relative quality estimator for several acceleration approaches in an empirical way, i.e., by requesting feedback from one or more radiologists. The obtained relative quality estimators are stored in a table, so that for each acceleration approach and each relevant context information (e.g., anatomical structure, clinical question), the corresponding relative quality estimator can be looked up.

[0043] Quality factor Q of acceleration option k k is the relative quality estimator q k,i The weighted sum of the weighting coefficients 0≦w i ≦1 depends on the application of the protocol and represents the relevance of image quality for each sequence i. Q k teeth, It is represented as TIFF2025505492000004.tif1131.

[0044] The acceleration trade-off weighting 0≦p≦1 specifies the overall importance of acceleration relative to image quality. Larger values ​​of p result in higher acceleration accompanied by a loss of image quality.

[0045] To determine the best acceleration option, an overall objective function TIFF2025505492000005.tif838 is k opt Maximized to find TIFF2025505492000006.tif933.

[0046] The optimization is performed for multiple values ​​of p, resulting in multiple versions of the accelerated testing protocol with different acceleration factors. The clinician can then decide which version to select to best fill the available time slots for testing.

[0047] When the accelerated option is selected, the necessary AI-based reconstruction model is calculated by training it either on the supplied example images or on synthetically generated data. Pre-trained AI reconstruction models may be available, which are then only re-trained and adapted to the specific type of images under consideration according to the examination protocol.

[0048] It is an insight of the present invention that individually designed and trained AI-based reconstruction models work best when applied to MR signal data acquired by executing the specific imaging sequence for which they were trained. Therefore, the trained AI-based reconstruction models should be provided as the output of the above-mentioned optimization procedure together with the individual optimized imaging sequence definition. For this purpose, the present invention proposes the concept of a model-weighted imaging protocol definition (or model-weighted "ExamCard") as shown in FIG. 3. While a standard (conventional) examination protocol (Protocol A in FIG. 3) contains only acquisition instructions for different imaging sequences defined by their respective acquisition parameters, a model-weighted examination protocol (Protocol B in FIG. 3) also contains an AI-based reconstruction model trained for the reconstruction of MR images from MR signal data acquired by executing a specific imaging sequence, together with an instruction set (FIG. 2 "Reconstruction Instructions") that specifies how the reconstruction model should be applied and which MR signal data serve as input. Once such a model-weighted examination protocol is provided, the MR reconstruction engine (Reconstruction Processor 17 in FIG. 1) has all the information available to reconstruct the image in a specially designed optimized manner. In the illustrated embodiment, the reconstruction of MR images from MR signal data of some imaging sequences is based on MR signal data and / or image data obtained from other imaging sequences in the embodiment of FIG. 3. As a result, the execution order of the imaging sequences becomes important, and the optimized examination protocol may have an optimized sequence order different from the original examination protocol. In the illustrated embodiment, an AI-based reconstruction model A is used to reconstruct MR images from MR signal data resulting from instruction execution of imaging sequence 1. Image data from imaging sequences 1 and 2 are used by an AI-based reconstruction model B to replace or support imaging sequence 3. For example, AI model B uses image data from sequences 1 and 2 and partial k-space data from sequence 3 to reconstruct / synthesize an MR image of the contrast associated with sequence 3.

[0049] In a possible embodiment, the present invention may be applied as an optimization service offered by a service provider (e.g., a manufacturer of MR scanners). Customers can send their own examination protocols (e.g., "ExamCards"), their diagnostic preferences (diagnostic relevance weightings) and ideally a set of example images to the optimization service. The examination protocols are then optimized by a computer run by the optimization service in order to increase the acquisition speed according to the method of the present invention as described above. The customer receives one or several optimized model-enhanced versions of the specific examination protocols, suitable for direct "out-of-the-box" replacement of the original examination protocols.

Claims

1. 1. A method for optimizing an examination protocol for performing MR image acquisitions from a patient's body, the method comprising: providing an examination protocol including specifications for two or more imaging sequences; - executing, in a computer, at least one algorithm that processes the examination protocol as an input for performing an optimization regarding a trade-off between image quality and speed of execution of the examination protocol, taking into account diagnostic relevance weightings assigned to imaging sequences included in the examination protocol, wherein the diagnostic relevance weightings represent the relevance of image quality for a specific diagnostic purpose; making available an output representative of the optimized examination protocol to a user and / or performing the MR image acquisition on an MR scanner based on the optimized examination protocol. A method comprising:

2. The method of claim 1 , wherein the optimization comprises modifying acquisition parameters associated with at least one of the imaging sequences.

3. The method of claim 1 or 2, wherein the optimization comprises modifying an image reconstruction model associated with at least one of the k-space sampling pattern and / or the imaging sequence.

4. The method of claim 3 , wherein the k-space sampling pattern is modified to result in an undersampling of k-space.

5. The method of claim 1 or 2, wherein the optimization comprises assigning an artificial intelligence-based image reconstruction model to at least one of the imaging sequences.

6. 6. The method of claim 5, wherein the artificial intelligence based image reconstruction model uses a machine learning method, and the optimization includes training the artificial intelligence based image reconstruction model and incorporating the trained artificial intelligence based image reconstruction model into the output.

7. 3. The method of claim 1, wherein the optimization includes a step of assigning an image reconstruction model to at least one of the imaging sequences, the image reconstruction model using MR signal data or image data acquired by performing at least one other imaging sequence.

8. 3. The method of claim 1 or 2, wherein execution of at least one of the imaging sequences is omitted in the optimized examination protocol, and an image reconstruction model is added to the optimized examination protocol to synthesize an MR image related to the omitted imaging sequence from MR signal data acquired by executing at least one of the other imaging sequences.

9. The method according to claim 1 or 2, wherein the optimization comprises modifying the order in which the imaging sequences included in the examination protocol are performed.

10. 3. The method of claim 1, wherein the at least one algorithm further takes into account a quality trade-off weighting, which is a user-specified trade-off between the quality of the MR image resulting from the optimized examination protocol and the increased execution speed achieved by the optimization.

11. 3. The method of claim 1, wherein the algorithm determines several optimization options and selects one of the optimization options according to an objective function that assigns a higher weighting to an optimization option that results in higher image quality of the MR images associated with imaging sequences having a higher diagnostic relevance weighting and simultaneously results in lower image quality of the MR images associated with imaging sequences having a lower diagnostic relevance weighting.

12. A step of reading a digital representation of an examination protocol including specifications for two or more imaging sequences; - executing at least one algorithm that processes the examination protocol as input for performing an optimization regarding the trade-off between image quality and speed of execution of the examination protocol, taking into account diagnostic relevance weights assigned to imaging sequences included in the examination protocol, wherein the diagnostic relevance weights represent the relevance of image quality for a specific diagnostic purpose; making available an output representative of the optimized testing protocol; A computer that is configured to run

13. A step of reading a digital representation of an examination protocol including specifications for two or more imaging sequences; - executing at least one algorithm that processes the examination protocol as input for performing an optimization regarding the trade-off between image quality and speed of execution of the examination protocol, taking into account diagnostic relevance weights assigned to imaging sequences included in the examination protocol, wherein the diagnostic relevance weights represent the relevance of image quality for a specific diagnostic purpose; making available an output representative of the optimized testing protocol; A computer program having instructions for executing a program.

14. 1. An MR scanner comprising: at least one main magnet coil for generating a main magnetic field in an examination volume; several gradient coils for generating switched magnetic field gradients in different spatial directions in the examination volume; at least one RF coil for generating RF pulses in the examination volume and / or for receiving MR signals from a patient's body positioned in the examination volume; a control computer for controlling a time sequence of the switched magnetic field gradients and RF pulses according to an examination protocol; and a reconstruction unit for reconstructing MR images from the received MR signals, wherein the MR scanner comprises: reading a digital representation of the examination protocol, the digital representation including specifications for two or more imaging sequences; - executing, in the control computer, at least one algorithm that processes the examination protocol as an input for performing an optimization regarding the trade-off between image quality and speed of execution of the examination protocol, taking into account diagnostic relevance weightings assigned to imaging sequences included in the examination protocol, wherein the diagnostic relevance weightings represent the relevance of image quality for a particular diagnostic purpose; performing MR image acquisition on the MR scanner based on the optimized examination protocol; 1. An MR scanner configured to perform:

Images