Transmitting K-space data for remote reconstruction of magnetic resonance images

Abstract

A medical system (100) includes a local memory (138) storing local machine executable instructions (140) and pulse sequence commands (142), a magnetic resonance imaging system (103), and a local computing system (132). Execution of the local machine executable instructions causes the local computing system to transmit (200) metadata describing a magnetic resonance imaging protocol to a remote computing system (132 ') via a network connection (137). Execution of the local machine executable instructions causes the local computing system to repeatedly control (202) the magnetic resonance imaging system with pulse sequence commands to acquire one (146) of a series of discrete k-space acquisitions. Constructing (204) a compressed k-space acquisition (148) by compressing one of the series of k-space acquisition using a compression module; and transmitting (206) the compressed k-space acquisition to the remote computing system via the network connection.

Description

Technical Field

[0001] The present invention relates to magnetic resonance imaging, and more particularly to the reconstruction of magnetic resonance images. Background Art

[0002] As part of the process for generating images within a patient, a magnetic resonance imaging (MRI) scanner uses a large static magnetic field to align the nuclear spins of atoms. This large static magnetic field is referred to as the B0 field or the main magnetic field. A combination of radiofrequency signals and magnetic gradients is used to encode the nuclear spins. The nuclear spins will emit radiofrequency signals in response to this encoding. These radiofrequency signals can be sampled and stored digitally. These samples are samples in k-space and are referred to herein as k-space data. The k-space data can be transformed into a magnetic resonance image through a Fourier transform.

[0003] U.S. Patent Application Publication US2021003651A1 discloses a medical data processing device including a processing circuit. The processing circuit acquires a first data segment obtained through sparse sampling. The processing circuit generates a first compressed data segment having a quantity lower than the first data segment by multiplying each of the first data segments by each of a set of weight coefficients and adding the multiplied first data segments. The processing circuit performs a first process of outputting a second compressed data segment by applying a trained model to the first compressed data segment, and the trained model is trained by receiving at least one of the first compressed data segments based on sparse sampling and outputting the second compressed data segment based on full sampling. Summary of the Invention

[0004] The present invention provides a medical system, a computer program, and a method in the independent claims. Embodiments are given in the dependent claims.

[0005] The amount of data that may be acquired during a magnetic resonance imaging scan can be very large, and the raw data size of the acquired k-space data can reach hundreds of megabytes or even tens of gigabytes. Once the measured k-space data has been acquired, different methods can be used to reconstruct the magnetic resonance image according to the magnetic resonance imaging protocol. Some of these reconstruction methods can be computationally intensive, especially those that rely on deep learning or optimization processes to perform the reconstruction. Therefore, there is interest in performing image reconstruction on remote or cloud-based platforms where computing power can be provided on demand.

[0006] The difficulty is that transferring a large amount of data to a remote system can cause delays in receiving the reconstructed magnetic resonance images. Embodiments can provide a means to accelerate this. For example, the measured k-space data is acquired as a series of discrete k-space acquisitions. When the discrete k-space data is acquired, it is compressed and then transferred to a remote computing system, where the k-space data is decompressed and then used to reconstruct the magnetic resonance image. This can have several advantages. One potential advantage is that the measured k-space data can reach the remote computing system more quickly and may reduce or alleviate data bottlenecks. Another potential advantage is that for some reconstruction techniques, partial data (such as the blades of propeller k-space data or self-navigating data) can be used immediately to perform partial reconstruction.

[0007] In one aspect, the present invention provides a medical system that includes a local memory storing local machine-executable instructions and pulse sequence commands. The pulse sequence commands are commands or data that can be converted into commands for controlling a magnetic resonance imaging system to acquire k-space data. Generally, the pulse sequence commands are provided in the form of events that occur at different time periods during an acquisition process. The medical system also includes a magnetic resonance imaging system configured to acquire measured k-space data describing an object within an imaging region. The pulse sequence commands are configured to control the magnetic resonance imaging system to acquire the measured k-space data as a series of discrete k-space acquisitions according to a magnetic resonance imaging protocol. The k-space data that can be used to reconstruct a complete image is the measured k-space data. The series of discrete k-space acquisitions can also be referred to as a series of discrete k-space data lines or a series of excitations.

[0008] The medical system also includes a local computing system. The operation of the local machine executable instructions causes the local computing system to transmit metadata describing a magnetic resonance imaging protocol to a remote computing system via a network connection. The operation of the local machine executable instructions causes the local computing system to repeatedly control the magnetic resonance imaging system using pulse sequence commands to acquire one in a series of discrete k-space acquisitions. For example, this can acquire a line of k-space data. The metadata describes various aspects of the acquisition and / or sampling of the measured k-space data. Reconstructing an image from the measured k-space data can be adapted to the acquisition and / or sampling aspects of the measured k-space data. To this end, the metadata is made available during the reconstruction process. This can be achieved by transmitting the metadata together with the compressed k-space data, or the metadata can be made available for the reconstruction process by transmitting it separately or otherwise providing it. Some examples of acquisition and / or sampling aspects may be simple Cartesian k-space sampling, radial or spiral k-space sampling, blade-by-blade Cartesian sampling where the blades have different orientations in k-space ("multi-blade"). Redundant sampling of the central region of k-space can be employed to correct for motion, or separate motion data can be used to correct for the motion of the measured k-space data. In the case of parallel imaging using undersampling of k-space, the metadata may represent different sets of measured k-space data acquired by a particular receive antenna (receive coil). The metadata may represent aspects of the undersampling pattern, the coil sensitivity curve, or the auto-calibration data representing the coil sensitivity curve within the measured k-space data. These aspects may help to adjust the reconstruction strategy for the way the k-space data is acquired. The metadata represents control parameters for selecting and / or adjusting reconstruction aspects. For example, the sampling pattern (or dynamic (k,t)-space, where t represents time) of k-space that applies spatial and / or contrast encoding to the magnetic resonance signal. When reconstructing a magnetic resonance image from the sampled magnetic resonance signal, the reconstruction needs to take into account these encoding aspects of the magnetic resonance signal. In addition, the metadata can represent motion information related to the phase shift of the magnetic resonance signal due to motion. In the reconstruction, this motion information needs to be taken into account to correct for motion. The metadata can also represent aspects of parallel imaging, such as the parallel imaging acceleration factor and the spatial coil sensitivity curve that need to be considered when unfolding the aliasing of the magnetic resonance signal due to k-space undersampling.

[0009] The operation of the local machine executable instructions causes the local computing system to repeatedly build a compressed k-space acquisition as one in a series of compressed k-space acquisitions by compressing one in the series of k-space acquisitions using a compression module. The operation of the local machine executable instructions also causes the local computing system to repeatedly transmit the compressed k-space acquisition to a remote computing system via a network connection. In these features, the magnetic resonance imaging system is controlled to acquire a single discrete k-space acquisition, and then compress it into a compressed k-space acquisition. After it is compressed using the compression module, it is then transmitted to the remote computing system.

[0010] The running of the local machine-executable instructions also causes the local computing system to transmit, via a network connection, at least a portion of a series of compressed k-space acquisitions before completion of the acquisition of the measured k-space data. This embodiment may be beneficial because it provides an extremely efficient and fast way to transmit the measured k-space data to a remote computing system. For example, compared to, for example, collecting, compressing, and then sending all of the measured k-space data to a remote computing system, this may enable more rapid reconstruction of magnetic resonance images.

[0011] In another embodiment, the medical system further includes a remote memory storing remote machine-executable instructions. The medical system further includes a remote computing system. The running of the remote machine-executable instructions causes the remote computing system to receive, via a network connection, metadata describing a magnetic resonance imaging protocol. The running of the remote machine-executable instructions also causes the remote computing system to repeatedly receive compressed k-space acquisitions via the network connection. The execution of the remote machine-executable instructions also causes the remote computing system to repeatedly obtain one of a series of discrete k-space acquisitions by decompressing the compressed k-space acquisitions using a decompression module.

[0012] The execution of the machine-executable instructions also causes the remote computing system to reconstruct a magnetic resonance image from at least a portion of the measured k-space data according to the magnetic resonance imaging protocol specified in the metadata. In different examples, this can take different forms. In some cases, the system may wait for all of the compressed k-space acquisitions to arrive so that a complete measured k-space data can be reconstructed. In such a case, the system will receive the measured data faster than collecting and then transmitting the measured k-space data. In other cases, such as in parallel imaging or in the presence of self-navigation, such as in propeller magnetic resonance imaging techniques, it may be beneficial to perform individual discrete acquisitions earlier. For example, if the discrete acquisitions of k-space are the blades of propeller k-space data, then the system can even start motion compensation before receiving the complete measured k-space data. Thus, this enables more rapid reconstruction of magnetic resonance images.

[0013] In another embodiment, the medical system further includes a motion detection system for acquiring motion data describing the movement of an object. In some examples, the motion detection system can be a sensor system, such as a camera, a respiratory belt, or other sensor systems. In other examples, the motion detection system can work using motion detection performed by the magnetic resonance imaging system itself. For example, fiducial markers can be placed on the surface of the object, self-navigation can be performed, or navigation measurements can be performed for acquiring motion data.

[0014] The execution of the local machine executable instructions causes the local computing system to control the motion detection system to collect motion data during the acquisition of measured k-space data. If the motion data indicates that the object motion exceeds a predetermined motion threshold, the execution of the local machine executable instructions also causes the local computing system to trigger a motion alert. For example, if the object moves too frequently, or the object relocates its body during the acquisition of measured k-space data, the trigger may indicate that motion correction may be required.

[0015] If the motion alert is triggered, the execution of the local machine executable instructions also causes the local computing system to initiate the construction of a compressed k-space acquisition and transmit the compressed k-space acquisition to a remote computing system via a network connection. If the motion alert is triggered, the metadata includes a motion compensation reconstruction request for the measured k-space data.

[0016] The motion alert may indicate object motion that is occurring or has occurred during a portion of the acquisition of the measured k-space data. The motion alert then requests that the remote computing system perform a reconstruction that performs motion compensation.

[0017] As described above, motion detection using a motion detection system can be performed not only using explicit potential hardware such as motion sensors (such as cameras), but also detected in the acquired raw k-space data or even in the image itself. This embodiment is beneficial because motion compensation can be extremely computationally intensive. The ability to detect the motion of an object and trigger remote reconstruction can enable the use of a more powerful computing system to perform the reconstruction.

[0018] In another embodiment, the compression module is implemented as a neural network. Implementing the compression module as a neural network can be beneficial because neural networks process data very quickly and have low computational requirements. Therefore, using a neural network can compress discrete k-space acquisitions faster and may reduce computational requirements and computing power.

[0019] This method can have one or more of the following benefits:

[0020] 1. Data compression in k-space does not require lossy data preprocessing, thus enabling full utilization of the encoding capabilities of the compression neural network employed.

[0021] 2. Compression of MR one-dimensional phase encoding (k-space data) allows independent compression and transmission of data from the scanner. This in turn enables multi-threaded compression and data transmission, thereby improving throughput.

[0022] 3. Compression of 1D signals (k-space data) allows for the use of a smaller encoding network, thereby reducing the encoding and decoding run times. Using a smaller network also improves the rate-distortion tradeoff in cases where it may be beneficial or necessary to transmit the compressed neural network weights together with the compressed data.

[0023] In another embodiment, the compression module is formed by the encoder portion of an autoencoder. The compressed k-space data acquisition is the latent space vector of the autoencoder. Then, the decoder can be formed by the decoder portion of the autoencoder. This can be beneficial because the latent space vector can be used to accurately transmit the k-space data.

[0024] The phase-encoded MR profile (k-space data) can be arbitrarily defined in 2D or 3D k-space, subject to specific physiological and scan parameter constraints. Common choices for the sampling trajectory are along Cartesian, radial, or spiral coordinates. Regardless of the trajectory design, the phase encoding can be linearly projected from 2D or 3D space to one-dimensional space. Then, this one-dimensional signal is processed by an encoding neural network structure and nonlinearly projected to a lower-dimensional latent one-dimensional space for compression. The resulting signal can then be transmitted to the receiver and decoded by a decoding neural network.

[0025] The encoding and decoding networks can be trained offline by solving an autoencoding optimization task. Given a training dataset that is assumed to represent the data to be compressed, the autoencoder optimizes the neural network parameters such that for all samples in the training dataset, the output of the autoencoder is as close as possible to the input of the network. Stochastic gradient descent or its extensions can be used to efficiently solve this task, and the metric for evaluating the reconstruction fidelity can be defined sample-by-sample in the L2 or L1 sense.

[0026] Different embodiments of the present invention can vary the encoding-decoding pipeline. For example:

[0027] 1. A single autoencoder is independently applied to all phase encodings.

[0028] 2. Different autoencoders are used for individual phase encodings.

[0029] 3. A single autoencoder is used for all phase encodings, but is adjusted for individual phase encodings, and additional information captured during the encoding of each phase encoding is transmitted.

[0030] In another embodiment, the autoencoders used as the encoder part and the decoder part are from a vector quantization variational autoencoder (VQ-VAE). Vector quantization variational autoencoders have been shown to be useful for compressing images, as demonstrated by Razavi et al. in "Generating diverse high-fidelity images with vq-vae-2." In Advances in neural information processing systems, pp. 14866-14876. 2019. These same techniques can be used to compress k-space data. Applying VQ-VAE may be beneficial because the data can be accurately recovered after decompression, and VQ-VAE may be simpler and use fewer computational resources required for compressing images.

[0031] Previously acquired k-space data can be used as training data for the autoencoder to train VQ-VAE and other autoencoders to compress data.

[0032] In another embodiment, a magnetic resonance imaging system includes a magnetic resonance imaging coil, and the resonance imaging coil includes digitization circuitry for measuring a series of discrete k-space acquisitions. A local computing system contains the digitizer circuitry. The digitizer circuitry is configured to construct a compressed k-space acquisition when measuring one of the series of k-space acquisitions. This embodiment may be beneficial because it can be used to compress discrete k-space acquisitions immediately after they are acquired. Many magnetic resonance imaging techniques employ so-called parallel imaging. This is where multiple receive coils are used. If the digitization circuitry for each of these receive coils compresses the discrete k-space acquisitions separately, the compression and transmission will be performed in parallel. This can greatly speed up the transmission of the measured k-space data to a remote computing system.

[0033] Notably, when using the digitizer circuitry to construct a compressed k-space acquisition, using a neural network can be very beneficial. This is because neural networks available for compression (such as variational autoencoders) have a very low computational overhead. This means that the digitizer circuitry requires fewer computational requirements and also uses less power. This can be a concern because the digitizer circuitry may be placed in a magnetic resonance imaging system, and it is usually not possible to provide it with a large amount of power. This is because the lines entering the magnetic resonance imaging system may pick up some radio frequency and energy used to perform the magnetic resonance imaging protocol. Therefore, reducing the power consumption of the digitizer circuitry is very beneficial.

[0034] In another embodiment, the running of the machine-executable instructions further causes the computing system to determine a predicted correlation in the discrete k-space acquisitions based on previously acquired k-space data. The metadata includes the predicted correlation. The running of the machine-executable instructions further causes the computing system to decorrelate one of the series of discrete k-space acquisitions prior to its being compressed into the compressed k-space acquisition using the predicted correlation. Performing such decorrelation essentially reduces the amount of data that needs to be compressed. This can reduce the size of the data transmitted to the remote computing system.

[0035] In the remote computing system, the receiver can use the metadata regarding the correlation between the k-space data to reconstruct it after decompression.

[0036] In another embodiment, the previously acquired k-space data is derived from a survey scan of an object or from a previously compressed one of the series of discrete k-space acquisitions. This can be beneficial because in both cases, acquisitions such as the survey scan or the previously acquired acquisition may contain information on how to effectively decorrelate the k-space data.

[0037] In one example, the previously acquired data can be an entire local scout image, a localization map image, or a survey image. Based on this, the correlation between the k-space data can be derived, and a transform can be calculated to decorrelate the data prior to using compression techniques. The method is similar to the imaging technique GRAPPA, where the actual acquisition is altered to avoid measuring redundant data. For example, samples in the k-space are skipped, and these samples can be inferred from adjacent measurements.

[0038] In another example, information from a patient's previous measurement is used to advantageously improve the compression rate of the next data portion to be transmitted.

[0039] In another example, previously transmitted k-space measurements that are already present at the receiver are used to decorrelate the next data to be compressed and transmitted. To this end, the compression method is additionally fed the previously transmitted data while compressing the k-space data currently being compressed. The decoder at the receiving end operates in a corresponding or complementary manner and uses the history of the decompressed blocks to decompress the current k-space data.

[0040] In another aspect, the present invention provides a medical system that includes a remote memory storing remote machine-executable instructions. The medical system also includes a remote computing system. The running of the remote machine-executable instructions causes the computing system to receive metadata via a network connection, the metadata describing the acquisition of k-space data to be measured as a series of discrete k-space acquisitions according to a magnetic resonance imaging protocol. The running of the remote machine-executable instructions also causes the remote computing system to repeatedly receive compressed k-space data acquisitions via the network connection and obtain one of the series of discrete k-space data acquisitions by decompressing the compressed k-space data acquisitions using a decompression module. The aforementioned compression module is complementary and works together with the decompression module.

[0041] The running of the machine-executable instructions also causes the computing system to reconstruct a magnetic resonance image from at least a portion of the series of discrete k-space acquisitions according to the magnetic resonance imaging protocol specified in the metadata. As described above, this portion or the medical system forms a receiver that works complementarily with the transmitter portion detailed previously. This receiver portion is capable of receiving k-space data more quickly, such that computationally intensive magnetic resonance imaging reconstruction can begin while still receiving, or complete measured k-space data can be received more quickly and then reconstruction can begin more quickly.

[0042] In another embodiment, the running of the remote machine-executable instructions causes the remote computing system to begin reconstructing a magnetic resonance image before receiving complete measured k-space data. As described above, it is highly beneficial to begin reconstruction as early as possible so that the magnetic resonance image is available more quickly.

[0043] In another embodiment, the magnetic resonance imaging protocol is a parallel imaging magnetic resonance imaging protocol. This embodiment may be beneficial because the discrete k-space acquisitions can represent images from individual coils. This enables the remote computing system to begin reconstructing a magnetic resonance image before receiving complete measured k-space data.

[0044] In another embodiment, the running of the remote machine-executable instructions causes the remote computing system to begin reconstructing a magnetic resonance image before receiving complete measured k-space data. The magnetic resonance imaging protocol is a PROPELLER magnetic resonance imaging protocol. In PROPELLER magnetic resonance imaging, blades of k-space data are acquired. The discrete k-space acquisitions can be the blades of k-space data. The so-called blades are used to reconstruct undersampled blade images, which can be used for motion correction. This embodiment may be beneficial because it enables motion correction to begin before the complete measured k-space data arrives.

[0045] In another embodiment, the execution of the remote machine executable instructions causes the remote computing system to begin reconstructing a magnetic resonance image before receiving complete measured k-space data. The magnetic resonance imaging protocol is a self-navigated magnetic resonance imaging protocol. This embodiment may be beneficial because the individual discrete acquisitions of k-space data may include self-navigated data. This embodiment may also enable motion correction to be calculated before receiving complete measured k-space data.

[0046] In another embodiment, the execution of the remote machine executable instructions causes the remote computing system to begin reconstructing a magnetic resonance image after receiving complete measured k-space data. For example, this may be useful for a magnetic resonance imaging protocol that is extremely computationally intensive, where the reconstruction has a large computational overhead.

[0047] In another embodiment, the magnetic resonance imaging protocol is a magnetic resonance fingerprint magnetic resonance imaging protocol. This is an example of a protocol with a large computational overhead, and although the system may wait until complete measured k-space data is received, it may receive the complete measured k-space data more quickly, so this may be beneficial.

[0048] In another embodiment, the magnetic resonance imaging protocol is a motion compensated magnetic resonance imaging protocol. For example, this may be a magnetic resonance imaging protocol that uses the initially measured k-space data to check the self-consistency of the motion correction. This may also result in a large computational overhead. Therefore, even if the reconstruction does not begin until complete measured k-space data is received, it is still advantageous because the remote computing system is able to begin such a computationally intensive reconstruction more quickly.

[0049] In another aspect, the present invention provides a computer program that includes local machine executable instructions that are run by a local computing system configured to control a magnetic resonance imaging system. The computer program may also include pulse sequence commands. The magnetic resonance imaging system is configured to acquire measured k-space data that describes an object within an imaging region. The pulse sequence commands are configured to control the magnetic resonance imaging system to acquire the measured k-space data as a series of discrete k-space acquisitions according to a magnetic resonance imaging protocol.

[0050] The running of the local machine-executable instructions causes the local computing system to transmit metadata describing a magnetic resonance imaging protocol to a remote computing system via a network connection. The running of the local machine-executable instructions causes the local computing system to repeatedly control a magnetic resonance imaging system using a pulse sequence command to acquire one of a series of discrete k-space acquisitions. The running of the local machine-executable instructions also causes the local computing system to repeatedly build a compressed k-space acquisition into one of a series of compressed k-space acquisitions by compressing one of the series of k-space acquisitions using a compression module. The running of the local machine-executable instructions also causes the local computing system to repeatedly transmit the compressed k-space acquisition to the remote computing system via the network connection. The running of the local machine-executable instructions also causes the local computing system to transmit at least a portion of the series of compressed k-space acquisitions via the network connection before completion of the acquisition of the measured k-space data.

[0051] In another aspect, the present invention provides a method of operating a medical system. The medical system includes a local memory storing machine-executable instructions and pulse sequence commands. The method also includes a remote memory storing remote machine-executable instructions. The medical system also includes a magnetic resonance imaging system configured to acquire measured k-space data describing an object within an imaging region. The pulse sequence commands are configured to control the magnetic resonance imaging system to acquire the measured k-space data as a series of discrete k-space acquisitions according to a magnetic resonance imaging protocol. The medical system also includes a local computing system. The medical system also includes a remote computing system.

[0052] The running of the local machine-executable instructions causes the local computing system to transmit metadata describing a magnetic resonance imaging protocol to a remote computing system via a network connection.

[0053] The running of the remote machine-executable instructions causes the remote computing system to receive the metadata via the network connection.

[0054] The running of the local machine-executable instructions causes the local computing system to repeatedly control a magnetic resonance imaging system using a pulse sequence command to acquire one of a series of discrete k-space acquisitions. The running of the local machine-executable instructions also causes the local computing system to repeatedly build a compressed k-space acquisition into one of a series of compressed k-space acquisitions by compressing one of the series of k-space acquisitions using a compression module. The running of the local machine-executable instructions also causes the local computing system to repeatedly transmit the compressed k-space acquisition to the remote computing system via the network connection.

[0055] The running of the local machine-executable instructions also causes the local computing system to transmit at least a portion of the series of compressed k-space acquisitions via the network connection before completion of the acquisition of the measured k-space data.

[0056] The running of the remote machine-executable instructions also causes the remote computing system to repeatedly receive compressed k-space acquisitions via a network connection. The execution of the remote machine-executable instructions also causes the remote computing system to repeatedly obtain one in a series of discrete k-space acquisitions by decompressing the compressed k-space acquisitions using a decompression module. The running of the machine-executable instructions also causes the remote computing system to reconstruct a magnetic resonance image from at least a portion of the series of discrete k-space acquisitions according to a magnetic resonance imaging protocol specified in the metadata.

[0057] It should be understood that one or more of the foregoing embodiments of the invention may be combined as long as the combined embodiments are not mutually exclusive.

[0058] As will be appreciated by one skilled in the art, several aspects of the present invention may be embodied as an apparatus, method or computer program product. Accordingly, aspects of the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment (including firmware, resident software, microcode, etc.) or an embodiment combining software and hardware aspects, which may be collectively referred to herein as a "circuit", "module" or "system". Furthermore, aspects of the present invention may take the form of a computer program product embodied in one or more computer-readable media having computer-executable code embodied thereon.

[0059] Any combination of one or more computer-readable media may be used. The computer-readable media may be a computer-readable signal medium or a computer-readable storage medium. As used herein, "computer-readable storage medium" includes any tangible storage medium that can store instructions executable by a processor of a computing device or a computing system. The computer-readable storage medium may be referred to as "computer-readable non-transitory storage medium". The computer-readable storage medium may also be referred to as a tangible computer-readable medium. In some embodiments, the computer-readable storage medium may also be capable of storing data that can be accessed by the computing system of the computing device. Examples of computer-readable storage media include, but are not limited to: floppy disks, magnetic hard disk drives, solid state drives, flash memories, USB thumb drives, random access memory (RAM), read only memory (ROM), optical disks, magneto-optical disks, and register files of a computing system. Examples of optical disks include compact discs (CDs) and digital versatile discs (DVDs), such as CD-ROM, CD-RW, CD-R, DVD-ROM, DVD-RW, or DVD-R discs. The term computer-readable storage medium also refers to various types of recording media that can be accessed by the computer device via a network or a communication link. For example, data can be retrieved via a modem, via the Internet, or via a local area network. The computer-executable code embodied on the computer-readable medium can be transmitted using any suitable medium, including but not limited to wireless, wired, optical fiber, RF, etc., or any suitable combination of the foregoing.

[0060] A computer-readable signal medium may include a propagated data signal having computer-executable code, for example, implemented therein either in baseband or as part of a carrier wave. Such a propagated signal may take any of a variety of forms, including but not limited to, electromagnetic, optical, or any suitable combination thereof. A computer-readable signal medium may be any computer-readable medium that is not a computer-readable storage medium and that can convey, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device.

[0061] "Computer memory" or "memory" is an example of a computer-readable storage medium. Computer memory is any memory directly accessible by a computing system. "Computer storage device" or "storage device" is another example of a computer-readable storage medium. A computer storage device is any non-volatile computer-readable storage medium. In some embodiments, a computer storage device may also be a computer memory, or vice versa.

[0062] As used herein, a "computing system" encompasses electronic components capable of executing programs or machine-executable instructions or computer-executable code. References to a computing system that include examples of a "computing system" should be construed as potentially including more than one computing system or processing core. A computing system can be, for example, a multi-core processor. A computing system can also refer to a collection of computing systems within a single computer system or distributed across multiple computer systems. The term computing system should also be construed as potentially referring to a collection or network of computing devices, each of which includes a processor or multiple computing systems. Machine-executable code or instructions can be executed by multiple computing systems or processors, which can be within the same computing device or can even be distributed across multiple computing devices.

[0063] Machine-executable instructions or computer-executable code can include instructions or programs that cause a processor or other computing system to perform an aspect of the present invention. The computer-executable code for performing operations for aspects of the present invention can be written in any combination of one or more programming languages, including object-oriented programming languages such as Java, Smalltalk, C++, etc., and conventional procedural programming languages such as the "C" programming language or similar programming languages, and compiled into machine-executable instructions. In some cases, the computer-executable code can be in the form of a high-level language or in a pre-compiled form and used in conjunction with an interpreter that generates machine-executable instructions on the fly. In other cases, machine-executable instructions or computer-executable code can be in the form of programming for a programmable logic gate array.

[0064] The computer-executable code can run entirely on the user's computer as a stand-alone software package, partially on the user's computer, partially on the user's computer and partially on a remote computer, or entirely on the remote computer or server. In the latter scenario, the remote computer can be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or a connection that can be made to an external computer (e.g., using an Internet service provider over the Internet).

[0065] Aspects of the present invention are described with reference to the flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It should be understood that each block or portion of blocks in the flowchart illustrations and / or block diagrams can be implemented by computer program instructions in the form of computer-executable code, when applicable. It should also be understood that, when not mutually exclusive, combinations of blocks in different flowchart illustrations and / or block diagrams can be combined. These computer program instructions can be provided to the memory of a general purpose computer, a special purpose computer, or other programmable data processing apparatus to produce a machine, such that the instructions executed via the memory of the computer or other programmable data processing apparatus create means for implementing the functions / actions specified in one or more blocks of the flowchart and / or block diagram.

[0066] These machine-executable instructions or computer program instructions can also be stored in a computer-readable medium that can direct a computer, other programmable data processing apparatus, or other devices to function in a particular manner, such that the instructions stored in the computer-readable medium produce an article of manufacture including instructions for implementing the functions / actions specified in the flowchart and / or one or more block diagram blocks.

[0067] The machine-executable instructions or computer program instructions can also be loaded onto a computer, other programmable data processing apparatus, or other devices to cause a series of operational steps to be performed on the computer, other programmable apparatus, or other devices to produce a computer-implemented process, such that the instructions executed on the computer or other programmable apparatus provide a process for implementing the functions / actions specified in the flowchart and / or one or more block diagram blocks.

[0068] As used herein, a "user interface" is an interface that allows a user or operator to interact with a computer or computer system. A "user interface" can also be referred to as a "human-machine interface device". A user interface can provide information or data to an operator and / or receive information or data from the operator. A user interface can enable input from an operator to be received by a computer and can provide output from the computer to the user. In other words, a user interface can allow an operator to control or manipulate a computer, and the interface can allow the computer to indicate the effects of the operator's control or manipulation. The display of data or information on a display or graphical user interface is an example of providing information to an operator. Receiving data via a keyboard, mouse, trackball, touchpad, pointing stick, graphics tablet, joystick, gamepad, webcam, helmet, pedal, wired glove, remote control, and accelerometer are examples of user interface components for receiving information or data from an operator.

[0069] As used herein, "hardware interface" encompasses interfaces that enable a computing system of a computer system to interact with or control external computing devices and / or apparatuses. A hardware interface may allow the computing system to send control signals or instructions to an external computing device and / or apparatus. A hardware interface may also enable the computing system to exchange data with an external computing device and / or apparatus. Examples of hardware interfaces include, but are not limited to: Universal Serial Bus, IEEE 1394 port, parallel port, IEEE 1284 port, serial port, RS232 port, IEEE488 port, Bluetooth connection, wireless local area network connection, TCP / IP connection, Ethernet connection, control voltage interface, MIDI interface, analog input interface, and digital input interface.

[0070] As used herein, "display" or "display device" encompasses output devices or user interfaces suitable for displaying images or data. A display may output visual, audio, and tactile data. Examples of displays include, but are not limited to: computer monitors, television screens, touchscreens, tactile electronic displays, Braille screens,

[0071] cathode ray tubes (CRTs), storage tubes, bistable displays, electronic paper, vector displays, flat panel displays, vacuum fluorescent displays (VF), light emitting diode (LED) displays, electro-luminescent displays (ELD), plasma display panels (PDP), liquid crystal displays (LCD), organic light emitting diode displays (OLED), projectors, and head-mounted displays.

[0072] k-space data is defined herein as the recorded measurements of radio frequency signals emitted by atomic spins using an antenna of a magnetic resonance apparatus during a magnetic resonance imaging scan. Magnetic resonance data is an example of tomographic medical image data.

[0073] Magnetic resonance imaging or MR image is defined herein as a reconstructed two-dimensional or three-dimensional visualization of anatomical data contained within magnetic resonance imaging data. Such visualization may be performed using a computer. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] Hereinafter, preferred embodiments of the present invention will be described only by way of example and with reference to the accompanying drawings, in which:

[0075] Figure 1 An example of a medical system is illustrated;

[0076] Figure 2 is a flowchart of a method of using a Figure 1 medical instrument;

[0077] Figure 3 An example of a radio frequency coil having a plurality of coil elements is shown;

[0078] Figure 4 An example of an autoencoder for k-space data that can be used for compression is illustrated;

[0079] Figure 5 Examples of image domain and k-space domain compression are illustrated; and

[0080] Figure 6 is shown Figure 5 a timing diagram of the example illustrated in

[0081] List of Reference Numerals

[0082] 100 Medical system

[0083] 101 Transmitter section

[0084] 102 Receiver section

[0085] 103 Magnetic resonance imaging system

[0086] 104 Magnet

[0087] 106 Bore of the magnet

[0088] 108 Imaging region

[0089] 109 Field of view

[0090] 110 Magnetic field gradient coil

[0091] 112 Magnetic field gradient coil power supply

[0092] 114 Radio frequency coil

[0093] 116 Transceiver

[0094] 118 Object

[0095] 120 Object support

[0096] 130 Local computer

[0097] 130’ Remote computer

[0098] 132 Local computing system

[0099] 132’ Remote computing system

[0100] 134 Local hardware interface

[0101] 136 Local network connection

[0102] 136’ Remote network connection

[0103] 137 Network connection

[0104] 138 Local memory

[0105] 138’ Remote memory

[0106] 140 Local machine-executable instructions

[0107] 142 Pulse sequence command

[0108] 144 Metadata

[0109] 146 One of a series of discrete k-space acquisitions

[0110] 148 Compressed k-space acquisition

[0111] 150 Compression module

[0112] 160 Remote machine-executable instructions

[0113] 162 Decompression module

[0114] 164 Series of discrete k-space acquisitions (measured k-space data)

[0115] 166 Magnetic resonance image

[0116] 200 Local computing system: Transmit metadata describing an MR imaging protocol to a remote computing system via a network connection

[0117] 202 Remote computing system: Receive metadata describing the acquisition of measured k-space data as a series of discrete k-space acquisitions according to an MR imaging protocol via a network connection

[0118] 204 Local computing system: Control an MR imaging system using a pulse sequence command to acquire one of a series of discrete k-space acquisitions

[0119] 206 Local computing system: Construct a compressed k-space acquisition as one of a series of compressed k-space acquisitions by compressing one of a series of k-space acquisitions using a compression module

[0120] 208 Local computing system: Transmit the compressed k-space acquisition to a remote computing system via a network connection

[0121] 210 Remote computing system: Receive the compressed k-space acquisition via a network connection

[0122] 212 Remote computing system: Obtain one of a series of discrete k-space acquisitions by decompressing the compressed k-space acquisition using a decompression module

[0123] 214 Remote computing system: Reconstruct a magnetic resonance image from at least a portion of the series of discrete k-space acquisitions

[0124] 300 Coil element

[0125] 302 Optical fiber connection

[0126] 304 Digitizer (DSP)

[0127] 400 Autoencoder

[0128] 500 Image domain compression

[0129] 502 k-space domain compression

[0130] 504 Acquisition

[0131] 506 Fourier transform

[0132] 508 Magnetic resonance image

[0133] 510 NxN Encoder

[0134] 512 Transmission

[0135] 514 NxN Decoder

[0136] 516 Reconstruction

[0137] 518 Parallel encoding

[0138] 520 Transmission

[0139] 522 Parallel decoding Detailed implementation manner

[0140] In these figures, similarly numbered elements are equivalent elements or perform the same function. If the functions are equivalent, the elements that have been previously discussed will not necessarily be discussed again in later figures.

[0141] Figure 1 An example of a medical system 100 is illustrated, which includes a transmitter part 101 and a receiver part 102. The transmitter part 101 is shown to be formed by a magnetic resonance imaging system 103 and a local computer 130. The receiver part is shown to be formed by a remote computer 130'.

[0142] The magnetic resonance imaging system 103 includes a magnet 104. The magnet 104 is a superconducting cylindrical magnet having a bore 106 therethrough. It is also possible to use different types of magnets; for example, split cylindrical magnets and so-called open magnets can also be used. The split cylindrical magnet is similar to a standard cylindrical magnet, except that the cryostat has been split into two parts to allow access to the isoplanes of the magnet, so that the magnet can be used, for example, in combination with charged particle beam therapy. The open magnet has two magnet segments, one on top of the other, with a space therebetween large enough to accommodate an object: the arrangement of the two segment regions is similar to the arrangement of Helmholtz coils. The open magnet is popular because the object is less restricted. Inside the cryostat of the cylindrical magnet is a set of superconducting coils.

[0143] Within the bore 106 of the cylindrical magnet 104, there is an imaging region 108 where the magnetic field is strong enough and uniform enough to perform magnetic resonance imaging. A field of view 109 within the imaging region 108 is shown. The magnetic resonance data acquired is typically acquired for the field of view 109. The object 118 is shown being supported by an object support 120 such that at least a portion of the object 118 is within the imaging region 108 and a predetermined region of interest 109.

[0144] There is also a set of magnetic field gradient coils 110 within the bore 106 of the magnet for acquiring primary magnetic resonance data to spatially encode the magnetic spins within the imaging region 108 of the magnet 104. The magnetic field gradient coils 110 are connected to a magnetic field gradient coil power supply 112. The magnetic field gradient coils 110 are intended to be representative. Generally, the magnetic field gradient coils 110 comprise a set of three discrete coils for spatially encoding in three orthogonal spatial directions. The magnetic field gradient power supply supplies current to the magnetic field gradient coils. The current supplied to the magnetic field gradient coils 110 is controlled as a function of time and can be ramped or pulsed.

[0145] Adjacent to the imaging region 108 is a radio frequency coil 114, which is used to manipulate the orientation of magnetic spins within the imaging region 108 and to receive radio frequency emissions from spins also within the imaging region 108. The radio frequency antenna may comprise a plurality of coil elements. The radio frequency antenna may also be referred to as a channel or an antenna. The radio frequency coil 114 is connected to a radio frequency transceiver 116. The radio frequency coil 114 and the radio frequency transceiver 116 may be replaced by separate transmit coils and receive coils and separate transmitters and receivers. It is to be understood that the radio frequency coil 114 and the radio frequency transceiver 116 are representative. The radio frequency coil 114 is also intended to represent a dedicated transmit antenna and a dedicated receive antenna. Similarly, the transceiver 116 may also represent separate transmitters and receivers. The radio frequency coil 114 may also have a plurality of receive / transmit elements, and the radio frequency transceiver 116 may have a plurality of receive / transmit channels. For example, if a parallel imaging technique such as SENSE is performed, the radio frequency coil 114 will be able to have a plurality of coil elements.

[0146] The transmitter section 101 is also shown as including a local computer 130. The transceiver 116 and the gradient controller 112 are shown as being connected to a hardware interface 134 of the local computer system 130. The local computer 130 is intended to represent one or more computers located near the magnetic resonance imaging system 103. The local computer 130 includes a local computing system 132. The local computing system 132 is intended to represent one or more arithmetic or computing cores. The local computing system 132 is shown as communicating with the hardware interface 134, which enables the local computing system 132 to communicate with the magnetic resonance imaging system 103 to control it and receive k-space data. The local computing system 132 is also shown as communicating with a local network connection 136 and a local memory 138. The local network connection 136 enables the local computer 130 to communicate with a remote computer 130' via a network interface 137.

[0147] The local memory 138 is intended to represent the various types of memory accessible to the local computing system 132. The local memory 138 is shown as storing local machine-executable instructions 140. The local machine-executable instructions 140 enable the local computing system 132 to perform various computing and data processing as well as image processing tasks. The local memory 138 is also shown as containing pulse sequence commands 142. The pulse sequence commands are commands or data that can be converted into commands that enable the local computing system 132 to control and operate the magnetic resonance imaging system 103 via the hardware interface 134.

[0148] Memory 138 is also shown as including metadata 144 that describes pulse sequence commands 142. Remote computer 130' uses the metadata to determine which magnetic resonance imaging protocol to use when reconstructing magnetic resonance images. Local memory 138 is also shown as including one of a series of discrete k-space acquisitions 146 acquired when controlling magnetic resonance imaging system 103 using pulse sequence command 142. Memory 138 is also shown as including compressed k-space acquisition 148, which is the compression result of one of a series of discrete k-space acquisitions 146 compressed using compression module 150. Compression module 150 is shown as being stored by local memory 138.

[0149] Receiver portion 102 is shown as including remote computer 130'. Remote computer 130’ is shown as including remote computing system 132'. Remote computer 130' can be, for example, an implementation of a cloud-based reconstruction service for magnetic resonance imaging system 103. Remote computing system 132’ is shown as communicating with remote memory 138' and remote network interface 136'. Remote network interface 136' is used to form a network connection 137 with local network interface 136. Network connection 137 can be, for example, a LAN, an internet connection, or a wireless data communication. Remote memory 138 is intended to represent the various types of memory accessible to remote computing system 132'.

[0150] Remote memory 138’ is shown as including remote machine executable instructions 160. These include commands that enable remote computing system 132' to perform basic data processing and digital and image processing tasks. Remote memory 138’ is also shown as including the compressed k-space acquisitions received via network connection 137. Remote memory 138' is also shown as including one of a series of discrete k-space acquisitions 146, the said series of discrete k-space acquisitions 146 being obtained by decompressing compressed k-space acquisition 148 using decompression module 162. Remote memory 138’ is also shown as including a series 164 of discrete k-space acquisitions, which in this case is equivalent to the measured k-space data. Memory 138' is also shown as including magnetic resonance image 166 reconstructed from series 164 of discrete k-space acquisitions or the measured k-space data.

[0151] Figure 2 The illustrated operations are shown Figure 1Flowchart of a method of a medical system. First, in step 200, the local computing system 132 transmits metadata 144 to the remote computing system 132' via the network connection 137. Next, in step 202, the remote computing system 132' receives the metadata 144 via the network connection 137. Next, in step 204, the local computing system 132 controls the magnetic resonance imaging system 103 with a pulse sequence command 142 to acquire a series of discrete k-space acquisitions 146. The magnetic resonance imaging system does this continuously. Steps 204, 206, and 208 can be performed in parallel when they can be executed. Next, in step 206, a compressed k-space acquisition 148 is constructed by compressing one of the series of discrete k-space acquisitions 146 using the compression module 150. After constructing the compressed k-space acquisition 148, step 208 is executed.

[0152] In step 208, the local computing system 132 transmits the compressed k-space acquisition 148 to the remote computing system 132' via the network interface 137. After executing step 208, the magnetic resonance imaging system can continuously acquire k-space data. This is indicated by the arrow returning to step 204. When this happens, the method can also continue to step 210. Step 210 represents what happens after one compressed k-space acquisition 148 is transmitted. Next, in step 210, the remote computing system 132' receives the compressed k-space acquisition 148 via the network connection 137. Steps 210 and 212 can be performed in parallel because in step 212, by decompressing the compressed k-space acquisition 148 using the decompression module 162, one of the series of discrete k-space acquisitions 148 can be obtained.

[0153] Then the method returns to step 210 and step 214. The step returning to step 210 indicates how the system continuously receives the compressed k-space data and decompresses it. In some instances, the system will wait to receive the complete series of discrete k-space acquisitions 164 or the measured k-space data and then execute step 214, which is to reconstruct the magnetic resonance image 166 from at least a portion of the series of discrete k-space?? data 164. In other examples, there may be techniques such as motion compensation or parallel imaging where the remote computing system 132' can start reconstructing the magnetic resonance image 166 before receiving all of the series of discrete k-space acquisitions 164. In this case, step 214 can continue as steps 210 and 212 are continuously executed.

[0154] Figure 3Illustrates an embodiment of the radio frequency coil 114. In this example, there are multiple coil elements 300 that independently receive k-space data. The coil elements include an optical connection 302 to the hardware interface 134 and a digitizer 304 or DSP circuit for the measured k-space data. It can be seen that the digitizer 304 has received the measured k-space data and then compresses it into a compressed k-space acquisition 148. These are sent directly to the local computing system 132 via the optical connection 302, and the local computing system 132 forwards it to the remote computing system 132' via the network connection 137. Figure 3 One possible advantage of the system shown is that the k-space data is compressed immediately and done in a parallel manner. It should be noted that in Figure 3 a possible implementation of the compression module 150 is to use a neural network. This can be particularly advantageous as it requires less computational overhead and can reduce the power requirements of the digitizer 304.

[0155] Figure 4 Illustrates an example of how an autoencoder 400 can be used to form the compression module 150 and the decompression module 162. In this example, the compression module 150 is the encoder part of the autoencoder 400, and the decompression module 162 is the decoder part of the autoencoder. The encoder part or compression module 150 receives the discrete k-space acquisition 146 as input and then outputs a latent space vector or a compressed k-space acquisition 148. The compressed k-space acquisition 148 can be transmitted via the network interface 137, where it is then input into the decompression module 162, which is the decoder part of the autoencoder 400. In response to receiving the compressed k-space acquisition 148, the decompression module 162 outputs the discrete k-space acquisition 146.

[0156] Figure 5 Illustrates an image domain compression method 500 and a k-space domain compression method 502 for transmitting magnetic resonance imaging data (such as k-space data) to a remote computing system. Figure 5 Provides a high-level comparison of MR compression in the image domain 500 and the proposed framework in the k-space domain 502 for a single-slice brain scan of NxN dimensions. Compression in the image domain requires all phase-encoding profiles to be acquired before applying preprocessing, domain transformation, and encoding. In contrast, encoding individual k-space phase encodings allows for multi-threaded encoding, transmission (Tx / Rx), and decoding of the raw data.

[0157] Image domain compression 500 starts with acquisition 504 and performs a Fourier transform 506 after all the data has been acquired, resulting in a magnetic resonance image 508. Then an NxN encoder or compressor 510 is performed. It is noted that all data is acquired before the image 508 is reconstructed. This is then transmitted 512, where it is received by a remote computing system and passed through an NxN decoder 514. The NxN decoder is a decompression algorithm. It is then fed into reconstruction 516.

[0158] k-space domain compression 502 works in a different way. Acquisition 504 occurs, but as the individual bits of k-space data become available, they are encoded in parallel 518 and then immediately transmitted 520 via network interface 137. When they are received, they are decoded or decompressed in parallel or asynchronously 522. When the data is received in whole or in part, it goes into reconstruction 516.

[0159] Figure 6 For illustrating the advantages of k-space domain compression 502 over image domain compression 500. Figure 6 A thread diagram of encoding, transmission, and decoding in the image domain 500 and k-space domain 502 is shown. Compression in the k-space domain enables the raw data in phase encoding to be processed independently. Thus, data can be encoded, transmitted (Tx / Rx), and decoded before acquisition is complete, and cloud-based reconstruction can benefit from higher data transmission throughput. Figure 6 A timing diagram of image domain compression 500 and k-space domain compression 502 is shown. In image domain compression 500, acquisition 504 is performed, and preprocessing or a Fourier transform 506 is carried out after the acquisition is completely finished. After that, encoding 510 is performed, whereupon it is then transmitted 512 and then decoded 514. Each process is performed serially in parallel.

[0160] In contrast, k-space domain compression performs many tasks in parallel or simultaneously. When acquisition 504 occurs, encoding 518 is carried out as the individual bits of k-space data become available. When encoding occurs, they are transmitted 520 as soon as they are available. Decoding 522 also occurs partially while the k-space data packets are still being transmitted. It can be seen that for k-space domain compression, decoding 522 is completed faster than image domain compression 500. This is illustrated by the time saved 600.

[0161] Although the present invention has been illustrated and described in detail in the drawings and the foregoing description, such illustration and description shall be regarded as illustrative or exemplary, rather than restrictive. The present invention is not limited to the disclosed embodiments.

[0162] Those skilled in the art can understand and implement other variations of the disclosed embodiments when studying the accompanying drawings, the disclosure content, and the claims when practicing the present invention claimed. In the claims, the word "comprising" does not exclude other elements or steps, and the words "a" or "an" do not exclude a plurality. A single processor or other unit can implement the functions of several items recited in the claims. Although specific measures are recited in mutually different dependent claims, this does not indicate that the combination of these measures cannot be used advantageously. A computer program can be stored / distributed on a suitable medium such as an optical storage medium or a solid-state medium provided together with other hardware or as part of other hardware, but can also be distributed in other forms such as via the Internet or other wired or wireless telecommunication systems. Any reference signs in the claims should not be construed as limiting the scope.

Claims

1. A medical system (100), comprising: a local memory (38) that stores local machine-executable instructions (140) and pulse sequence commands (142); a magnetic resonance imaging system (103) configured to acquire measured k-space data describing an object (118) within an imaging region (108), wherein the pulse sequence commands are configured to control the magnetic resonance imaging system to acquire the measured k-space data as a series of discrete k-space acquisitions (164) according to a magnetic resonance imaging protocol; a local computing system (132), wherein the running of the local machine-executable instructions causes the local computing system to transmit (200) metadata describing the magnetic resonance imaging protocol and representing control parameters for selecting and / or adjusting aspects of reconstruction to a remote computing system (132') via a network connection (137), wherein the running of the local machine-executable instructions causes the computing system to repeatedly: control (202) the magnetic resonance imaging system using the pulse sequence commands to acquire one (146) of the series of discrete k-space acquisitions; construct (204) a compressed k-space acquisition (148) as one of a series of compressed k-space acquisitions by compressing the one of the series of k-space acquisitions using a compression module (150); and transmit (206) the compressed k-space acquisition to the remote computing system via the network connection; and wherein the running of the local machine-executable instructions causes the local computing system to transmit at least a portion of the series of compressed k-space acquisitions via the network connection before completion of the acquisition of the measured k-space data.

2. The medical system according to claim 1, wherein the medical system further comprises: a remote memory (138') that stores remote machine-executable instructions (160); a remote computing system (132'), wherein the running of the remote machine-executable instructions causes the remote computing system to receive (202) the metadata describing the magnetic resonance imaging protocol via the network connection, wherein the running of the remote machine-executable instructions further causes the remote computing system to repeatedly: receive (210) the compressed k-space acquisition via the network connection; and obtain (212) the one of the series of discrete k-space acquisitions by decompressing the compressed k-space acquisition using a decompression module; and wherein the running of the machine-executable instructions further causes the remote computing system to reconstruct (214) a magnetic resonance image (166) from at least a portion of the measured k-space data according to the magnetic resonance imaging protocol specified in the metadata.

3. The medical system according to claim 1 or 2, wherein the medical system further comprises a motion detection system for acquiring motion data describing the motion of the object, wherein the running of the local machine-executable instructions causes the local computing system to: control the motion detection system to acquire the motion data during the acquisition of the measured k-space data; If the motion data indicates that the object moves beyond a predetermined motion threshold, a motion alert is triggered; If the motion alert is triggered, construction of the compressed k-space acquisition is initiated and the compressed k-space acquisition is transmitted via the network connection to the remote computing system, wherein, if the motion alert is triggered, the metadata includes a motion compensation reconstruction request for the measured k-space data.

4. The medical system according to claim 1, 2, or 3, wherein, the compression module is implemented as a neural network.

5. The medical system according to claim 4, wherein, the compression module is formed by an encoder portion of an autoencoder, wherein the compressed k-space acquisition is a latent space vector of the autoencoder.

6. The medical system according to any one of the preceding claims, wherein, the magnetic resonance imaging system includes a magnetic resonance imaging coil, the magnetic resonance imaging coil includes digitization circuitry (304) for measuring a series of discrete k-space acquisitions, wherein the local computing system includes the digitization circuitry, and wherein the digitization circuitry is configured to construct the compressed k-space acquisition when measuring one of the series of k-space acquisitions.

7. The medical system according to any one of the preceding claims, wherein, the running of the machine-executable instructions further causes the computing system to: determine a predicted correlation in the discrete k-space acquisitions based on previously acquired k-space data, wherein the metadata includes the predicted correlation; decorrelate one of the series of discrete k-space acquisitions using the predicted correlation before compressing to the compressed k-space acquisition.

8. The medical system according to claim 7, wherein, the previously acquired k-space data is derived from a scout scan of the object or from a previously compressed one of the series of discrete k-space acquisitions.

9. A medical system (100), comprising: a remote memory (138’) that stores remote machine-executable instructions (160); a remote computing system (132’), wherein the running of the remote machine-executable instructions causes the computing system to receive (202) metadata (144) via a network connection (137), the metadata describing the acquisition of k-space data to be measured as a series of discrete k-space acquisitions (164) according to a magnetic resonance imaging protocol, and wherein the running of the remote machine-executable instructions further causes the remote computing system to repeatedly: receive (210) the compressed k-space acquisition via the network connection; and obtain (212) one of the series of discrete k-space acquisitions by decompressing the compressed k-space acquisition using a decompression module; and wherein the running of the machine-executable instructions further causes the computing system to reconstruct (214) magnetic resonance images (166) from at least a portion of the series of discrete k-space acquisitions according to the magnetic resonance imaging protocol specified in the metadata.

10. The medical system according to claim 9, wherein, The running of the remotely machine-executable instructions causes the remote computing system to begin reconstructing the magnetic resonance image before receiving complete measured k-space data.

11. The medical system according to claim 10, wherein, any one of the following: the magnetic resonance imaging protocol is a parallel imaging magnetic resonance imaging protocol; the magnetic resonance imaging protocol is a PROPELLER magnetic resonance imaging protocol; and the magnetic resonance imaging protocol is a self-navigated magnetic resonance imaging protocol.

12. The medical system according to claim 9, wherein, the running of the remotely machine-executable instructions causes the remote computing system to begin reconstructing the magnetic resonance image after receiving complete measured k-space data.

13. The medical system according to claim 12, wherein, any one of the following: the magnetic resonance imaging protocol is a magnetic resonance fingerprinting magnetic resonance imaging protocol; and the magnetic resonance imaging protocol is a motion compensated magnetic resonance imaging protocol.

14. A computer program comprising locally machine-executable instructions (140) for running by a local computing system (132) configured to control a magnetic resonance imaging system (103), wherein, the magnetic resonance imaging system is configured to acquire measured k-space data describing an object within an imaging region, wherein the pulse sequence command is configured to control the magnetic resonance imaging system to acquire the measured k-space data as a series of discrete k-space acquisitions (164) according to a magnetic resonance imaging protocol; wherein the running of the locally machine-executable instructions causes the local computing system to transmit (200) metadata (144) describing the magnetic resonance imaging protocol and representing control parameters for selecting and / or adjusting aspects of the reconstruction via a network connection (137) to a remote computing system (132'), wherein the running of the locally machine-executable instructions causes the local computing system to repeatedly: control (204) the magnetic resonance imaging system using the pulse sequence command to acquire one (146) of the series of discrete k-space acquisitions; construct (206) a compressed k-space acquisition (148) as one of a series of compressed k-space acquisitions by compressing the one of the series of k-space acquisitions using a compression module; and transmit (208) the compressed k-space acquisition via the network connection to the remote computing system; and wherein the running of the locally machine-executable instructions causes the local computing system to transmit at least a portion of the series of compressed k-space acquisitions via the network connection before completion of the acquisition of the measured k-space data.

15. A method of operating a medical system (100), wherein, the medical system comprises: a local memory (138) storing locally machine-executable instructions (140) and pulse sequence commands (142); a remote memory (138') storing remotely machine-executable instructions (160); A magnetic resonance imaging system (103) configured to acquire measured k-space data describing an object (118) within an imaging region (108), wherein the pulse sequence commands are configured to control the magnetic resonance imaging system to acquire the measured k-space data as a series of discrete k-space acquisitions (164) according to a magnetic resonance imaging protocol; A local computing system (138); and A remote computing system (138'); wherein the running of the local machine-executable instructions causes the local computing system to transmit (200) metadata (144) describing the magnetic resonance imaging protocol and representing control parameters for selecting and / or adjusting aspects of the reconstruction to the remote computing system via a network connection (137); wherein the running of the remote machine-executable instructions causes the remote computing system to receive (202) the metadata via the network connection; wherein the running of the local machine-executable instructions causes the local computing system to repeatedly: Control (204) the magnetic resonance imaging system using the pulse sequence commands to acquire one of the series of discrete k-space acquisitions; Construct (206) a compressed k-space acquisition (148) as one of a series of compressed k-space acquisitions by compressing the one of the series of k-space acquisitions using a compression module; and Transmit (208) the compressed k-space acquisition to the remote computing system via the network connection; wherein the running of the local machine-executable instructions causes the local computing system to transmit at least a portion of the series of compressed k-space acquisitions via the network connection before completion of the acquisition of the measured k-space data; wherein the running of the remote machine-executable instructions further causes the remote computing system to repeatedly: Receive (210) the compressed k-space acquisition via the network connection; and Obtain (212) one of the series of discrete k-space acquisitions by decompressing the compressed k-space acquisition using a decompression module; and wherein the running of the machine-executable instructions further causes the remote computing system to reconstruct (214) a magnetic resonance image (166) from at least a portion of the series of discrete k-space acquisitions according to the magnetic resonance imaging protocol specified in the metadata.

Images