Automatic field of view alignment for magnetic resonance imaging

Abstract

Disclosed herein is a medical system (100, 300, 500) comprising: a memory (110) storing machine-executable instructions (120) and a predictor algorithm (122), the predictor algorithm configured to output predicted field of view alignment data (128) for a magnetic resonance imaging system (502) in response to inputting one or more localizer magnetic resonance images (124) and subject metadata (126). The predictor algorithm comprises a trainable machine learning algorithm. The medical system further comprises a processor (104) configured to control the medical system. Execution of the machine-executable instructions causes the processor to: receive (200) the one or more localizer magnetic resonance images and the subject metadata; and receive (202) the predicted field of view alignment data from the predictor algorithm in response to inputting the one or more localizer magnetic resonance images into the predictor algorithm and in response to inputting the subject metadata.

Description

Technical Field

[0001] The present invention relates to magnetic resonance imaging, and in particular to planning for magnetic resonance imaging. Background Art

[0002] As part of the process for producing the image in the patient's body, a large static magnetic field is used by a magnetic resonance imaging (MRI) scanner to align the nuclear spin of atoms. This large static magnetic field is referred to as the B0 field or main magnetic field. The various quantities or properties of an object can be measured spatially using MRI. For example, the various anatomical or physiological properties of an MRI survey object can be used. In order to image the appropriate position in the object, first a preliminary magnetic resonance image, commonly referred to as a locator or locator image, is collected. The locator is then used by the operator to appropriately locate the field of view (region of interest) for subsequent scans.

[0003] Conference publication et al. (2010). “Automated Scan Plane Planning for Brain MRIusing 2D Scout Images” (in Proceedings of ISMRM, 2010, p. 3136) discloses the use of an automatic algorithm for scan plane planning. Summary of the Invention

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

[0005] A difficulty with automatically positioning the field of view following a localizer MRI image is that the initial position of the object and the object's anatomy can vary significantly. A trained operator may also have difficulty consistently placing the field of view to obtain acceptable results in subsequent scans. Embodiments of the present invention can address this problem by using a predictor algorithm with a trainable machine learning component to position the field of view. The predictor algorithm uses one or more localizer MRI images and object metadata as input to generate predicted field of view alignment data. This can have several advantages. First, a medical imaging database containing historical localizer MRI images, object data, and position fields of view can be mined for training data.

[0006] In one aspect, the present invention provides a medical system comprising: a memory storing machine-executable instructions configured to control the medical system. The memory further stores a predictor algorithm configured to output predicted field of view alignment data in response to inputting one or more localizer magnetic resonance images and subject metadata. As used herein, predicted field of view alignment data encompasses data that can be used to specify a position of a field of view for a subsequent magnetic resonance imaging scan.

[0007] As used herein, localizer magnetic resonance images encompass magnetic resonance images that can be used to plan subsequent magnetic resonance imaging scans. These are typically low-resolution, very wide-field-of-view magnetic resonance images; however, this need not be the case, as most magnetic resonance images can be used to plan further magnetic resonance imaging scans. As used herein, object metadata encompasses data that describes the object and is relevant to the proper execution of the magnetic resonance imaging protocol. For example, the object metadata may list descriptors of the object, such as the subject's height, weight, gender, suspected medical condition, and other parameters. These object metadata may affect the prediction of the field-of-view alignment data. For example, children have a different, immature brain anatomy compared to the brain anatomy of adults. In addition, obese patients require a larger field of view to visualize the abdominal area in its entirety compared to a slim patient to be examined.

[0008] The predictor algorithm comprises a trainable machine learning algorithm. The medical system further comprises a processor configured to control the medical system. Execution of the machine executable instructions causes the processor to receive the one or more localizer magnetic resonance images and the object metadata. In different examples, both the one or more localizer magnetic resonance images and the object metadata may be received in different ways. A suitable machine learnable algorithm may be based on a deep learning model that uses both images and text (so-called text-image embeddings) for training and inference. A practical example is, for example: https: / / zpascal.net / cvpr2018 / Wang_TieNet_TextImage_Embedding_CVPR_2018_paper.pdf, https: / / arxiv.org / pdf / 1704.03470.pdf , https: / / arxiv.org / pdf / 1711.05535.pdf , http: / / openaccess.thecvf.com / content_ECCV_2018 / papers / Ying_Zhang_Deep_Cross Modal_Projection_ECCV_2018_paper.pdf.

[0009] Remarkably, a neural network employing such a complex variety of inputs (images, text, classification variables) is able to predict the field of view setting from the input localizer image and patient metadata.

[0010] In some cases, receiving may involve retrieving them from a memory or storage device. In other examples, they may be received via a network. In other examples, the processor may control something such as a magnetic resonance imaging system to acquire the one or more localizer magnetic resonance images, and may receive the object metadata, for example, from a storage device used to plan magnetic resonance imaging scans or via a terminal or other user interface.

[0011] Execution of the machine-executable instructions further causes the processor to receive predicted field of view alignment data from the predictor algorithm in response to inputting the one or more localizer magnetic resonance images into the predictor algorithm and in response to inputting the subject metadata. The one or more localizer magnetic resonance images and the subject metadata are input into the predictor algorithm, and in response, the predictor algorithm provides predicted field of view alignment data. This can be beneficial because the predicted field of view alignment data can be used to configure subsequent magnetic resonance imaging scans. This can, for example, enable automated magnetic resonance imaging. It can also be an auxiliary tool or quality control tool used when an operator is manually operating a magnetic resonance imaging system.

[0012] As used herein, a medical system can encompass different types of systems in different paradigms. In one example, the medical system can be a magnetic resonance imaging system and the described embodiments can be integrated into components of the magnetic resonance imaging system. In another example, the medical system can be a workstation for planning and / or analyzing magnetic resonance imaging data and images. In another example, the medical system can be a service provided by a cloud or cloud computing system via a network or other data exchange interface.

[0013] In another embodiment, the memory further stores training data. The training data comprises training entries. Each of the training entries includes one or more training magnetic resonance images, training subject metadata, and training field of view alignment data. The one or more training magnetic resonance images and training subject metadata represent the type of data to be input to the predictor algorithm. The training field of view alignment data represents the type of data output by the predictor algorithm.

[0014] The memory also includes a training algorithm configured to train the predictor algorithm using a comparison between the predicted field of view alignment data and the training field of view alignment data. This comparison can be performed, for example, in various ways. In some examples, the comparison can be made using a trained neural network. In other examples, the geometric positions of the predicted field of view alignment data and the training field of view alignment data can be compared and quantified.

[0015] Execution of the machine-executable instructions further causes the processor to receive predicted field of view alignment data from the predictor algorithm in response to inputting the one or more training magnetic resonance images into the predictor algorithm and in response to inputting the training subject metadata. Execution of the machine-executable instructions further causes the processor to train the predictor algorithm using a comparison between the predicted field of view alignment data and the training field of view alignment data. This may be beneficial because it provides a means for improving the accuracy of the predicted field of view alignment data output by the predictor algorithm.

[0016] In some cases, the comparison between the predicted field of view alignment data and the training field of view alignment data is determined by the predictor algorithm itself. For example, the two pieces of data can be input into the predictor algorithm, and the training proceeds automatically. In other cases, execution of the machine-executable instructions can also cause the processor to use another algorithm or mathematical comparison to determine the comparison between the predicted field of view alignment data and the training field of view alignment data. For example, the coordinates and positions of the two fields of view can be numerically compared.

[0017] In another embodiment, execution of the machine-executable instructions further causes the processor to generate the training data by extracting the one or more training magnetic resonance images, the training subject metadata, and the training field of view alignment data from a medical image database. Typically, when performing an magnetic resonance imaging protocol, data from an examination may be stored in a database. For example, the data from a particular examination may be stored in a so-called DICOM file. In one example, the data may be extracted from a DICOM file. This embodiment and related embodiments may be beneficial because they provide a means for using actual examination data and field of view alignment to generate or improve the predictor algorithm.

[0018] In another embodiment, the predictor algorithm is a convolutional neural network. For example, the convolutional neural network can be trained using deep learning. The predicted field of view alignment data can be numerically compared with the training field of view alignment data.

[0019] In another embodiment, the predictor algorithm includes a feature extractor configured to provide a feature vector using the object metadata and the one or more localizer magnetic resonance images. In different examples, the feature extractor can take different forms. For example, an algorithm that finds anatomical landmarks can include a feature vector that identifies the locations of these anatomical landmarks. Similarly, deformable shape models or even anatomical atlases can also be used in this manner. In other examples, a neural network can be used to provide a feature vector that can then be used by a trainable machine learning algorithm to output the predicted field of view alignment data.

[0020] A trainable machine learning algorithm is configured to output the predicted field of view alignment data in response to inputting the feature vector. Using feature vectors in this manner can be beneficial because they are highly transparent and can be easily understood by a physician or other professional using the medical system. This can reduce the vulnerability of the predictor algorithm's response. This can also enable the use of a trainable machine learning algorithm that is more easily understood by humans.

[0021] In another embodiment, the trainable machine learning algorithm is a decision tree algorithm.

[0022] In another embodiment, the trainable machine learning algorithm is a k-nearest neighbor algorithm.

[0023] The use of decision trees or k-nearest neighbor algorithms is beneficial where a human can manually check the model and see if it is safe and / or does not contain data that is misleading or may cause the predictor algorithm to malfunction or provide false data.

[0024] In another embodiment, the feature extractor is a trained neural network configured to provide the feature vector. This embodiment can be beneficial because the trained neural network is used only to provide the feature vector and essentially classifies the image and locates various landmarks that are then used by the predictor algorithm. This can make the operation of the predictor algorithm more transparent and understandable to humans, for example.

[0025] In another embodiment, the feature extractor is configured to provide the feature vector by fitting a deformable shape model.

[0026] In another embodiment, the feature extractor is configured to provide the feature vector using an anatomical atlas.

[0027] In another embodiment, the medical system is a medical imaging workstation.

[0028] In another embodiment, the medical system is a cloud-based MRI planning system. For example, an MRI system or other medical imaging workstation can contact the medical system via the Internet or other network connection, send the subject metadata and the one or more localizer MRI images to the medical system, and receive the predicted field of view alignment data in return.

[0029] In another embodiment, the medical system further comprises a magnetic resonance imaging system.

[0030] In another embodiment, the memory further comprises localizer pulse sequence commands configured to control the magnetic resonance imaging system to acquire localizer magnetic resonance imaging data. Execution of the machine-executable instructions further causes the processor to acquire the localizer magnetic resonance imaging data by controlling the magnetic resonance imaging system using the localizer pulse sequence commands. Execution of the machine-executable instructions further causes the processor to reconstruct the one or more localizer magnetic resonance images from the localizer magnetic resonance imaging data. This embodiment may be advantageous because the automatic determination of the field of view is directly incorporated into the magnetic resonance imaging system.

[0031] In another embodiment, the memory further includes clinical pulse sequence commands configured to control the magnetic resonance imaging system to acquire clinical magnetic resonance imaging data. Execution of the machine-executable instructions further causes the processor to generate a modified pulse sequence command by modifying the clinical pulse sequence command using the predicted field of view alignment data. For example, the predicted field of view alignment data can be used to modify or change the field of view in the modified pulse sequence command. Execution of the machine-executable instructions further causes the processor to control the magnetic resonance imaging system to acquire the clinical magnetic resonance imaging data using the modified pulse sequence command.

[0032] In some examples, execution of the machine-executable instructions may further cause the processor to reconstruct one or more clinical magnetic resonance images from the clinical magnetic resonance imaging data.

[0033] In another aspect, the present invention provides a computer program product comprising machine executable instructions for execution by a processor controlling the medical system. The computer program product also includes an implementation of the predictor algorithm, which is configured to output predicted field of view alignment data in response to inputting one or more localizer magnetic resonance images and object metadata. The predictor algorithm includes a trainable learning algorithm. Execution of the machine executable instructions causes the processor to receive the one or more localizer magnetic resonance images and the object metadata. Execution of the machine executable instructions also causes the processor to receive predicted field of view alignment data from the predictor algorithm in response to inputting the one or more localizer magnetic resonance images into the predictor algorithm and in response to inputting the object metadata. The advantages of this embodiment have been previously discussed.

[0034] In another aspect, the present invention provides a method of operating a medical system. The medical system includes a memory storing a predictor algorithm. The predictor algorithm is configured to output predicted field of view alignment data in response to inputting one or more localizer magnetic resonance images and subject metadata. The predictor algorithm comprises a trainable machine learning algorithm. The memory also includes training data. The training data comprises training entries. Each of the training entries comprises one or more training magnetic resonance images, training subject metadata, and training field of view alignment data. The memory also includes a training algorithm configured to train the particular algorithm using a comparison between the predicted field of view alignment data and the training field of view alignment data.

[0035] The method further includes receiving the predicted field of view alignment data from the predictor algorithm in response to inputting the one or more training magnetic resonance images into the predictor algorithm and in response to inputting the training subject metadata. The method further includes determining a comparison between the predicted field of view alignment data and the training field of view alignment data. The method further includes training the predictor algorithm using the training algorithm by inputting the comparison between the predicted field of view alignment data and the training field of view alignment data.

[0036] In another embodiment, the method further comprises generating the training data by extracting the one or more training magnetic resonance images, the training subject metadata and the training field of view alignment data from a medical image database.

[0037] In another embodiment, the method further comprises receiving one or more localizer magnetic resonance images and subject metadata. The method further comprises receiving the predicted field of view alignment data from the predictor algorithm in response to inputting the one or more localizer magnetic resonance images into the predictor algorithm and in response to inputting the subject metadata.

[0038] It should be understood that one or more of the above-described embodiments of the present invention can be combined as long as the combined embodiments are not mutually exclusive.

[0039] As will be appreciated by those skilled in the art, various aspects of the present invention may be implemented as an apparatus, method, or computer program product. Accordingly, various 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 (all of which may be generally referred to herein as "circuits," "modules," or "systems"). Furthermore, various 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.

[0040] Any combination of one or more computer-readable media may be utilized. The computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. As used herein, "computer-readable storage medium" encompasses any tangible storage medium that can store instructions that can be executed by a processor of a computing device. A computer-readable storage medium may be referred to as a computer-readable non-transitory storage medium. A computer-readable storage medium may also be referred to as a tangible computer-readable medium. In some embodiments, a computer-readable storage medium may also be capable of storing data that can be accessed by a processor of a computing device. Examples of computer-readable storage media include, but are not limited to, floppy disks, magnetic hard disk drives, solid-state hard disks, flash memory, USB thumb drives, random access memory (RAM), read-only memory (ROM), optical disks, magneto-optical disks, and a processor's register file. Examples of optical disks include compact disks (CDs) and digital versatile disks (DVDs), such as CD-ROMs, CD-RWs, CD-Rs, DVD-ROMs, DVD-RWs, or DVD-R disks. The term computer-readable storage medium also refers to various types of recording media that can be accessed by a computing device via a network or communication link. For example, data can be retrieved on a modem, the Internet, or a local area network. Computer executable code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0041] A computer-readable signal medium may include a propagated data signal having computer-executable code embodied therein, for example, 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 is capable of conveying, propagating, or transmitting a program for use by or in connection with an instruction execution system, apparatus, or device.

[0042] "Computer memory" or "memory" is an example of a computer-readable storage medium. Computer memory is any memory directly accessible by a processor. "Computer storage" or "storage" is another example of a computer-readable storage medium. Computer storage is any non-volatile computer-readable storage medium. In some embodiments, computer storage may also be computer memory, or vice versa.

[0043] As used herein, "processor" encompasses an electronic component capable of executing a program or machine-executable instructions or computer-executable code. References to a computing device comprising a "processor" should be interpreted as being capable of containing more than one processor or processing core. The processor may, for example, be a multi-core processor. A processor may also refer to a collection of processors within a single computer system or distributed across multiple computer systems. The term computing device should also be interpreted as being capable of referring to a collection or network of computing devices each comprising one or more processors. Computer-executable code may be executed by multiple processors, which may be within the same computing device or even distributed across multiple computing devices.

[0044] The computer executable code may include machine executable instructions or programs that cause a processor to perform aspects of the present invention. The computer executable code for performing operations for aspects of the present invention may be written in any combination of one or more programming languages ​​and compiled into machine executable instructions, the one or more programming languages ​​including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as the "C" programming language or similar programming languages. In some instances, the computer executable code may be in the form of a high-level language or in a precompiled form and used in conjunction with an interpreter that generates machine executable instructions on the fly.

[0045] The computer-executable code may execute entirely on the user's computer, partly on the user's computer (as a stand-alone software package), partly on the user's computer and partly on a remote computer, or entirely on a remote computer or server. In the latter case, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., through the Internet using an Internet service provider).

[0046] Aspects of the present invention are described with reference to the flow chart, diagram and / or block diagram of the method, device (system) and computer program product according to an embodiment of the present invention.Should be understood that, when applicable, each square frame or part of the square frame of flow chart, diagram and / or block diagram can be implemented by the computer program instruction in the form of computer executable code.It should also be understood that, when mutually non-exclusive, the combination of the square frames in different flow charts, diagrams and / or block diagrams can be combined.These computer program instructions can be provided to the processor of other programmable data processing devices of general-purpose computers, special-purpose computers or machines, so that the instruction executed via the processor of computer or other programmable data processing devices creates a unit for implementing the function / action specified in flow chart and / or one or more block diagram frames.

[0047] These computer program instructions may also be stored in a computer-readable medium that can direct a computer, other programmable data processing apparatus, or other device to operate in a specific manner so 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.

[0048] The computer program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device to cause a series of operational steps to be performed on the computer, other programmable apparatus, or other device to produce a computer-implemented process, such that the instructions executed on the computer or other programmable apparatus provide a process for the functions / actions specified in the flowchart and / or one or more block diagram blocks.

[0049] As used herein, "user interface" is an interface that allows a user or operator to interact with a computer or computer system. "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 an operator. A user interface can enable the input from an operator to be received by a computer and can provide output to a user from the computer. In other words, the user interface can allow an operator to control or manipulate a computer, and the interface can allow the computer to indicate the effect 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 by a keyboard, mouse, tracking ball, touchpad, pointing stick, graphic input board, joystick, game controller, webcam, earphones, pedals, wired gloves, remote controllers and accelerometers is all an example of a user interface component that realizes receiving information or data from an operator.

[0050] As used herein, "hardware interface" encompasses an interface that enables a processor of a computer system to interact with and / or control an external computing device and / or apparatus. A hardware interface can allow a processor to send control signals or instructions to an external computing device and / or apparatus. A hardware interface can also enable a processor to exchange data with an external computing device and / or apparatus. Examples of hardware interfaces include, but are not limited to, a universal serial bus, an IEEE 1394 port, a parallel port, an IEEE 1284 port, a serial port, an RS-232 port, an IEEE-488 port, a Bluetooth connection, a wireless LAN connection, a TCP / IP connection, an Ethernet connection, a control voltage interface, a MIDI interface, an analog input interface, and a digital input interface.

[0051] As used herein, a "display" or "display device" encompasses an output device or user interface suitable for displaying images or data. A display may output visual, audio, and / or tactile data. Examples of displays include, but are not limited to, computer monitors, television screens, touch screens, tactile electronic displays, Braille screens, cathode ray tubes (CRTs), memory tubes, bi-stable displays, electronic paper, vectorscopes, flat panel displays, vacuum fluorescent displays (VFs), light emitting diode (LED) displays, electroluminescent displays (ELDs), plasma display panels (PDPs), liquid crystal displays (LCDs), organic light emitting diode displays (OLEDs), projectors, and head-mounted displays.

[0052] Magnetic resonance (MR) data is defined herein as the measurements of radio frequency signals emitted by atomic spins, recorded using the antenna of a magnetic resonance device during a magnetic resonance imaging scan. Magnetic resonance data is an example of medical image data. A magnetic resonance imaging (MRI) image or MR image is defined herein as a reconstructed two- or three-dimensional visualization of anatomical data contained within the magnetic resonance imaging data. This visualization can be performed using a computer. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In the following, preferred embodiments of the present invention will be described, by way of example only, and with reference to the accompanying drawings, in which:

[0054] Figure 1 Illustration of an example of a medical system;

[0055] Figure 2 The diagram shows the use of Figure 1 A flowchart of a method of a medical system;

[0056] Figure 3 Another example of a medical system is shown;

[0057] Figure 4 The diagram shows the use of Figure 3 A flowchart of a method of a medical system;

[0058] Figure 5 Another example of a medical system is shown;

[0059] Figure 6 The diagram shows the use of Figure 5 a flowchart of a method of a medical system; and

[0060] Figure 7 Another example of a medical system is shown.

[0061] Reference Signs List

[0062] 100 Medical Systems

[0063] 102 Computer

[0064] 104 processors

[0065] 106 hardware interfaces

[0066] 108 optional user interfaces

[0067] 110 memory

[0068] 120 machine-executable instructions

[0069] 122 Predictor Algorithm

[0070] 124 One or more localizer MRI images

[0071] 126 Object Metadata

[0072] 128 predicted field of view alignment data

[0073] 128' predicted field of view alignment data

[0074] 200 Receive one or more localizer magnetic resonance images and object metadata

[0075] 400 Receiving predicted field of view alignment data from a predictor algorithm in response to inputting one or more localizer magnetic resonance images into the predictor algorithm and in response to inputting subject metadata

[0076] 300 training algorithms

[0077] 302 training items

[0078] 304 One or more training magnetic resonance images

[0079] 306 training object metadata

[0080] 308 training field of view alignment data

[0081] 310 Comparison

[0082] 312 Medical Imaging Database

[0083] 400 Receiving predicted field of view alignment data from a predictor algorithm in response to inputting one or more training magnetic resonance images into the predictor algorithm and in response to inputting training subject metadata

[0084] 402 Determine the comparison between the predicted field of view alignment data and the training field of view alignment data

[0085] 404 Training the predictor algorithm using comparison between predicted field of view alignment data and training field of view alignment data

[0086] 500 Medical Systems

[0087] 502 Magnetic Resonance Imaging System

[0088] 504 magnet

[0089] 506 magnet bore

[0090] 508 imaging area

[0091] 509 Area of ​​Interest

[0092] 510 magnetic field gradient coil

[0093] 512 magnetic field gradient coil power supply

[0094] 514 RF coil

[0095] 516 transceiver

[0096] 518 objects

[0097] 520 object support

[0098] 530 Positioner pulse sequence command

[0099] 532 localizer MRI data

[0100] 534 Clinical Pulse Sequence Commands

[0101] 536 Modified pulse sequence command

[0102] 538 clinical magnetic resonance imaging data

[0103] 540 clinical magnetic resonance images

[0104] 600 acquires localizer magnetic resonance imaging data by controlling the magnetic resonance imaging system using the localizer pulse sequence command

[0105] 602 Reconstruct one or more localizer magnetic resonance images based on the localizer magnetic resonance imaging data

[0106] 604 Generate a modified pulse sequence command by modifying the clinical pulse sequence command using the predicted field of view alignment data

[0107] 606 Acquiring clinical magnetic resonance imaging data by controlling the magnetic resonance imaging system using the modified pulse sequence commands DETAILED DESCRIPTION

[0108] In the figures, like numbered elements are equivalent elements or perform the same function. If the function is equivalent, an element that has been discussed previously will not necessarily be discussed in later figures.

[0109] Figure 1An example of a medical system 100 is illustrated. The medical system 100 is shown as including a computer 102 having a processor 104. The processor 104 is intended to represent one or more processing cores and can be distributed across different computers or computing systems. The processor 104 is connected to a hardware interface 106. The hardware interface 106 can be used, for example, to enable the processor 104 to connect to and / or control other components of the medical system 100. The hardware interface 106 can also include elements that enable it to communicate with other computer systems or data systems via a network. The processor 104 is also shown as being connected to an optional user interface 108. Here, the processor 104 is also shown as being connected to a memory 110.

[0110] Memory 110 may, for example, represent different types of memory accessible to processor 104. Memory 110 may be any combination of memory accessible to processor 104. This may include things such as main memory, cache memory, and non-volatile memory such as flash RAM, a hard drive, or other storage devices. In some examples, memory 110 may be considered a non-transitory computer-readable medium.

[0111] The memory 110 is shown as containing machine-executable instructions 120. The machine-executable instructions 120 enable the processor 104 to control the medical system 100. The machine-executable instructions 120 may also enable the processor 104 to perform various data analysis and image processing tasks.

[0112] The memory 110 is also shown as containing an embodiment of a predictor algorithm 122. The memory is also shown as containing one or more localizer magnetic resonance images 124. The term localizer is used as a global label to identify a particular magnetic resonance image or group of magnetic resonance images. The memory 110 is also shown as containing subject metadata 126. The subject metadata 126 is metadata describing the subject, the magnetic resonance protocol used to acquire or reconstruct the one or more localizer magnetic resonance images 124, or other data. The memory 110 is also shown as containing predicted field of view alignment data 128. The predicted field of view alignment data 128 is provided by inputting the subject metadata 126 and the one or more localizer magnetic resonance images 124 into the predictor algorithm 122.

[0113] Figure 2 A flow chart is shown illustrating a method of operating the medical system 100. First, in step 200, one or more localizer magnetic resonance images 124 are received. In step 200, subject metadata 126 is also received. Next, in step 202, predicted field of view alignment data 128 is provided by inputting the subject metadata 126 and the one or more localizer magnetic resonance images 124 into the predictor algorithm 122.

[0114] Figure 3 Another example of a medical system 300 is shown. It should be noted that Figure 1 Medical systems 100 and Figure 3 The features of the medical system 300 in FIG. 3 can be freely combined. This can take the form of combining all elements into a single computer system, or Figure 1 and 3 The systems 100, 300 illustrated in FIG may be connected via a network connection.

[0115] exist Figure 3 , the memory 110 is again shown as containing machine-executable instructions 120 and a predictor algorithm 122. The memory 110 is also shown as containing predicted field of view alignment data 128'. The memory 110 is also shown as containing a training algorithm 300 configured to modify or train the predictor algorithm 122. The memory 110 is also shown as containing one or more training entries 302. Each training entry 302 includes one or more training magnetic resonance images 304, training subject metadata 306, and training field of view alignment data 308.

[0116] One or more training magnetic resonance images 304 and training subject metadata 306 can be input into the predictor algorithm 122 to provide predicted field of view alignment data 128. The predicted field of view alignment data 128' can then be compared to the training field of view alignment data 308 and a comparison 310 can be made. For example, the field of view alignment data can include coordinates and / or orientations of the field of view. The comparison 310 can be a numerical comparison of these coordinates and orientations. The comparison 310 can then be input into the training algorithm 300, which then uses it to modify the predictor algorithm 122. In some examples, the comparison functionality is integrated directly into the training algorithm 300. In this case, the system operates by inputting the predicted field of view alignment data 128' and the training field of view alignment data 308 directly into the training algorithm 300, which then performs and modifies the predictor algorithm 122.

[0117] Memory 110 is also shown as optionally including a medical imaging database 312. Medical imaging database 312 may, for example, provide archived data such as DICOM images and other data and metadata acquired during use of the magnetic resonance imaging system. In some examples, machine-executable instructions 120 are programmed to mine or extract training entries 302 from the data contained within medical imaging database 312.

[0118] Figure 4 The diagram shows the operation Figure 3 Flowchart of a method of a medical system 300 . Figure 4 The method can be used with Figure 3 For example, Figure 3The steps can be Figure 4 Execute before or after the steps contained in .

[0119] First, in step 400, the predicted field of view alignment data 128 is received by inputting one or more training magnetic resonance images 304 and training subject metadata 306 into the predictor algorithm 122. Next, in step 402, a comparison 310 is calculated by comparing the training field of view alignment data 308 with the predicted field of view alignment data 128'. Finally, in step 404, the predictor algorithm 122 uses the comparison 310 to train or modify the predictor algorithm 122. The exact method of modifying the predictor algorithm 122 depends on the type of predictor algorithm. If the predictor algorithm comprises a trainable machine learning algorithm, then the type of algorithm determines how it is trained.

[0120] Figure 5 Another example of a medical system 500 is shown. In this example, the medical system 500 also includes a magnetic resonance imaging system 502. It should be noted that Figure 3 Features of the illustrated medical system 300 may also be used with Figure 5 The features shown are freely combinable.

[0121] Magnetic resonance imaging system 502 includes magnet 504. Magnet 504 is a superconducting cylindrical magnet with a bore 506 extending therethrough. It is also possible to use different types of magnets; for example, it is possible to use both split cylindrical magnets and so-called open magnets. A 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 isoplane of the magnet; such magnets can be used, for example, in conjunction with charged particle beam therapy. An open magnet has two magnet sections, one above the other, with a space large enough in between to accommodate the subject: the arrangement of the two sections is similar to that of Helmholtz coils. Open magnets are popular because the subject is less confined. Inside the cryostat of the cylindrical magnet is a collection of superconducting coils.

[0122] Within the bore 506 of the cylindrical magnet 504, there is an imaging zone 508 in which the magnetic field is sufficiently strong and uniform to perform magnetic resonance imaging. A region of interest 509 is shown within the imaging zone 508. Magnetic resonance data is typically acquired for a field of view. A subject 518 is shown supported by a subject support 520 such that at least a portion of the subject 518 is within the imaging zone 508 and the region of interest 509.

[0123] Also within the bore 506 of the magnet is a set of magnetic field gradient coils 510 for acquiring preliminary magnetic resonance data to spatially encode magnetic spins within the imaging region 508 of the magnet 504. The magnetic field gradient coils 510 are connected to a magnetic field gradient coil power supply 512. The magnetic field gradient coils 510 are intended to be representative. Typically, the magnetic field gradient coils 510 include three independent sets of 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 510 is controlled over time and can be ramped or pulsed.

[0124] Adjacent to the imaging zone 508 is a radio frequency coil 514, which is used to manipulate the orientation of magnetic spins within the imaging zone 508 and to receive radio frequency transmissions from spins also within the imaging zone 508. An radio frequency antenna may include multiple coil elements. An radio frequency antenna may also be referred to as a channel or antenna. The radio frequency coil 514 is connected to a radio frequency transceiver 516. The radio frequency coil 514 and the radio frequency transceiver 516 may be replaced by separate transmit and receive coils, or separate transmitters and receivers. It should be understood that the radio frequency coil 514 and the radio frequency transceiver 516 are representative. The radio frequency coil 514 is also intended to represent a dedicated transmit antenna and a dedicated receive antenna. Similarly, the transceiver 516 may also represent a separate transmitter and receiver. The radio frequency coil 514 may also have multiple transmit / receive elements, and the radio frequency transceiver 516 may have multiple transmit / receive channels. For example, if a parallel imaging technique such as SENSE is implemented, the radio frequency coil 514 will have multiple coil elements.

[0125] The transceiver 516 and gradient controller 512 are shown connected to the hardware interface 106 of the computer system 102 .

[0126] The memory 110 is also shown as containing localizer pulse sequence commands 530. The memory is also shown as containing localizer magnetic resonance imaging data 532 acquired by controlling the magnetic resonance imaging system 502 using the localizer pulse sequence commands 530. The memory 110 is also shown as containing one or more localizer magnetic resonance images 124. These are reconstructed based on the localizer magnetic resonance imaging data 532. The medical system 500 is shown as containing Figure 1 All features of medical system 100.

[0127] The memory 110 is also shown as containing clinical pulse sequence commands 534. The predicted field of view alignment data 128 can be used to modify the pulse sequence commands to set or modify the field of view. The memory 110 is also shown as containing modified pulse sequence commands 536, which are produced by modifying the clinical pulse sequence commands 534 using the predicted field of view alignment data 128. The memory 110 is also shown as containing clinical magnetic resonance imaging data 538 acquired by controlling the magnetic resonance imaging system 502 using the modified pulse sequence commands 536. The memory 110 is also shown as optionally containing a clinical magnetic resonance image 540 reconstructed based on the clinical magnetic resonance imaging data 538.

[0128] Figure 6 The diagram shows the operation Figure 5 Flowchart of a method of a medical system 500. First, in step 600, the MRI system 502 is controlled by using the localizer pulse sequence command 530 to acquire localizer MRI data 532. Next, in step 602, one or more localizer MRI images 124 are reconstructed based on the localizer MRI data 532. Next, the following is performed: Figure 2 Steps 200 and 202 are illustrated. After step 202, the method proceeds to step 604. In step 604, a modified pulse sequence command 536 is generated or created by modifying the clinical pulse sequence command 534 using the predicted field of view alignment data 128. Finally, in step 606, clinical magnetic resonance imaging data 538 is acquired by controlling the magnetic resonance imaging system 502 using the modified pulse sequence command 536.

[0129] Planning is a preliminary step in every MRI study on which the quality of clinical images (clinical magnetic resonance images 540) may depend. Automation of planning can allow one to achieve a high degree of repeatability in image orientation for quantitative comparison in subsequent studies. Examples can provide a new approach for training automatic sequence planning algorithms that exploit the content of radiology datasets. Examples can allow the development and training of automatic sequence planning algorithms without the involvement of pre-built anatomical models and expert knowledge.

[0130] A great deal of time and effort in radiology is spent on planning MR studies. The purpose of planning is to center the field of view (FOV) of the MR scanner relative to the region of interest (ROI) (509) and to orient the scan plane along the anatomical axes of the organs and systems of the subject (518). Planning allows one to maximize information, reduce the influence of patient position and individual anatomical features on the clinical image, and show the image from a convenient perspective. The main steps of sequence planning may include one or more of the following:

[0131] 1. Acquisition of low-resolution localizer images in three orthogonal planes.

[0132] 2. Visual identification and marking of anatomical landmarks.

[0133] 3. Alignment of the FOV relative to the identified anatomical landmarks. All subsequent clinical images were taken in the newly aligned FOV.

[0134] The paradigm can provide new approaches for the development of automatic sequence planning algorithms.

[0135] Manual sequence planning is a fast and cost-effective preliminary step in MR studies, but it has several disadvantages. The most important disadvantages of manual planning include:

[0136] 1) Human-related accuracy. The accuracy of sequence planning is an essential requirement for comprehensive and reliable MRI studies. However, it strongly depends on the qualification of the medical device, and there is no simple way to control it.

[0137] 2) Lack of reproducibility. Reproducibility of sequence planning is a key requirement for quantitative analysis and comparison of subsequent studies. However, there is significant inter- and intra-patient variability in FOV orientation, and it is almost impossible to ensure that subsequent scans are manually aligned in the same orientation as the previous study.

[0138] A possible way to overcome these shortcomings is through automatic sequence planning (ASP) algorithms. The purpose of these algorithms is to automatically suggest consistent and highly repeatable FOV orientations regardless of the peculiarities of the patient's anatomy and his / her position. So far, a series of ASP algorithms have been proposed, developed and implemented in commercial products. They are all based on pre-built human anatomical models and utilize automatic recognition of anatomical landmarks. As a result, existing ASP algorithms require specialized medical knowledge and cannot be easily adapted to the preferences and practices of specific medical organizations. The proposed paradigm can allow the development and training of automatic sequence planning algorithms (prediction algorithms 122), possibly without a pre-built anatomical model and possibly without the involvement of expert knowledge.

[0139] A possible element of the paradigm is a new approach for training automatic sequence planning (ASP) algorithms that leverages the content of radiology datasets. Most existing radiology datasets contain a localizer (or scout image), patient metadata (age, sex, weight, etc.), and clinical images of different modalities. The localizer shows the region of interest (ROI) in its initial position before FOV alignment and sequence planning. The clinical image shows the same ROI after sequence planning and contains information about the applied alignment steps (e.g., relative offset and tilt of the scan plane). Therefore, a radiology dataset can be viewed as a set of pairs of initial patient positions and optimal FOV orientations for a given position. In other words, such a set represents an implicit form of expert knowledge that can be extracted by modern machine learning algorithms (e.g., k-nearest neighbors, regression, computer vision, etc.). During the training of the machine learning algorithm, the localizer image and patient metadata are features; the position and orientation of the clinical image are target values. Once the algorithm has been optimized in this way, it is applied to new localizer images and generates the origin position and FOV orientation for subsequent radiology studies.

[0140] The example may include a predictor (predictor algorithm 122) and an optional teacher (training algorithm 300) (see below Figure 7 ). The predictor comprises a machine learning algorithm (a trainable machine learning algorithm) that receives as input the localizer image and returns the optimal position and orientation of the FOV for a subsequence of clinical images. Many modern machine learning algorithms (k-nearest neighbors, decision trees, convolutional networks, etc.) can be used as predictors. The teacher is an algorithm that compares the actual and predicted values ​​of the FOV position and orientation, estimates the degree of difference between the true and predicted values, and modifies the predictor to reduce the difference. The goal of the modification of the predictor can be to achieve the highest accuracy and precision of the target value. The specific implementation of the teacher depends on the predictor selected.

[0141] The embodiments may be applicable to planning of radiological studies of any part of an object. To implement the present invention, one or more of the following steps may be performed:

[0142] 1. Collect and prepare a suitable training radiology dataset. The localizer image and patient metadata serve as features, and the spatial FOV orientation of subsequent clinical images serves as the target standard data value.

[0143] 2. Using the selected features we select a predictor architecture that allows us to predict the spatial orientation of the FOV.

[0144] 3. Optimize the parameters of the predictor to minimize the difference between the predicted and actual orientations of the FOV.

[0145] 4. Store the ASP algorithm and its optimized parameters on a writable medium such as a computer hard drive.

[0146] To use the example, perform one or more of the following steps:

[0147] 1. Collect available patient metadata and obtain localizer images using appropriate radiological equipment.

[0148] 2. Load the stored ASP algorithm and its optimization parameters.

[0149] 3. Send patient meta information and localizer image to the predictor.

[0150] 4. Get the predicted optimal FOV orientation.

[0151] 5. Align FOV using the recommended optimal FOB orientation

[0152] 6. Obtain the required clinical images.

[0153] Figure 7 A functional view of a medical system 300 is shown. The medical system 300 may, for example, access a medical imaging database 312. This may contain historical magnetic resonance imaging data, such as a scout magnetic resonance image that may be used as one or more training magnetic resonance images 304. The medical imaging database 312 may also contain metadata describing a subject, which may be used as training subject metadata 306. The one or more training magnetic resonance images 304 and the training subject metadata 306 may be input into the predictor algorithm 122. This may provide predicted field of view alignment data 128′ for a particular one or more training magnetic resonance images 304 and training subject metadata 306. The predicted field of view alignment data 128′ output by the predictor 122 may then be provided to the training algorithm 300. The medical imaging database 312 may also provide field of view alignment data for a particular scout image 304. This may be used as training field of view alignment data 308. The training algorithm 300 may then use the training field of view alignment data 308 and the predicted field of view alignment data 128′ to train 404 the predictor algorithm 122.

[0154] In other cases, the localizer may represent one or more localizer magnetic resonance images 124 and subject metadata 126. These may be input into a predictor algorithm 122 to generate predicted field of view alignment data 128, which is then used to modify pulse sequence commands and to control the magnetic resonance imaging system 502. For example, a scanner console or user interface 108 of the magnetic resonance imaging system 502 may display the predicted field of view alignment data 128 for operator approval.

[0155] While the invention has been illustrated and described in detail in the drawings and foregoing description, such illustration and description are to be considered illustrative or exemplary and not restrictive; the invention is not limited to the disclosed embodiments.

[0156] Those skilled in the art will be able to understand and implement other variations of the disclosed embodiments when practicing the claimed invention by studying the drawings, the description and the claims. In the claims, the word "comprising" does not exclude other elements or steps, and the word "a" or "an" does not exclude a plurality. A single processor or other unit may satisfy the functions of several items recited in the claims. Although specific elements are recited in mutually different dependent claims, this does not indicate that a combination of these elements cannot be used to advantage. The computer program may be stored and / or distributed on an appropriate medium, such as an optical storage medium or solid-state medium provided with or as part of other hardware, but the computer program may also be distributed in other forms, such as via the Internet or other wired or wireless telecommunications systems. Any figure marks in the claims shall not be interpreted as limiting the scope.

Claims

1. A medical system (100, 300, 500), comprising: - a memory (110) storing machine-executable instructions (120) and a predictor algorithm (122), the predictor algorithm configured to output predicted field of view alignment data (128) for a magnetic resonance imaging system (502) in response to inputting one or more localizer magnetic resonance images (124) and subject metadata (126), wherein the predictor algorithm comprises a trainable machine learning algorithm, wherein the predictor algorithm is a convolutional neural network; - a processor (104) configured to control the medical system, wherein execution of the machine executable instructions causes the processor to: - receiving the one or more localizer magnetic resonance images and the object metadata (200); and - receiving the predicted field of view alignment data from the predictor algorithm in response to inputting the one or more localizer magnetic resonance images into the predictor algorithm and in response to inputting the subject metadata (202).

2. The medical system according to claim 1, wherein The memory further stores training data, wherein the training data comprises training entries, wherein each of the training entries comprises one or more training magnetic resonance images, training subject metadata, and training field of view alignment data; wherein the memory further comprises a training algorithm configured to train the predictor algorithm using a comparison between the predicted field of view alignment data and the training field of view alignment data; wherein execution of the machine-executable instructions further causes the processor to: - receiving the predicted field of view alignment data from the predictor algorithm in response to inputting the one or more training magnetic resonance images into the predictor algorithm and in response to inputting the training subject metadata (400); - determining said comparison between said predicted field of view alignment data and said training field of view alignment data (402); and - training the predictor algorithm using the comparison between the predicted field of view alignment data and the training field of view alignment data (404).

3. The medical system according to claim 2, wherein: Execution of the machine-executable instructions further causes the processor to generate the training data by extracting the one or more training magnetic resonance images, the training subject metadata, and the training field of view alignment data from a medical image database.

4. The medical system according to any one of the preceding claims, wherein The medical system is any one of: a medical imaging workstation and a cloud-based magnetic resonance imaging planning system.

5. The medical system according to claim 1, wherein The medical system also includes a magnetic resonance imaging system.

6. The medical system according to claim 5, wherein: The memory further includes localizer pulse sequence commands configured to control the magnetic resonance imaging system to acquire localizer magnetic resonance imaging data, wherein execution of the machine-executable instructions further causes the processor to: - acquiring the localizer magnetic resonance imaging data (600) by controlling the magnetic resonance imaging system using the localizer pulse sequence commands; - reconstructing the one or more localizer magnetic resonance images from the localizer magnetic resonance imaging data (602).

7. The medical system according to claim 5 or 6, wherein: The memory further includes clinical pulse sequence commands configured to control the magnetic resonance imaging system to acquire clinical magnetic resonance imaging data, wherein execution of the machine-executable instructions further causes the processor to: - generating a modified pulse sequence command by modifying the clinical pulse sequence command using the predicted field of view alignment data (604); and - acquiring the clinical magnetic resonance imaging data by controlling the magnetic resonance imaging system using the modified pulse sequence commands (606).

8. A computer program product comprising machine-executable instructions for execution by a processor controlling a medical system, wherein: The computer program product further comprises a predictor algorithm configured to output predicted field of view alignment data in response to inputting one or more localizer magnetic resonance images and subject metadata, wherein the predictor algorithm comprises a trainable machine learning algorithm, wherein execution of the machine-executable instructions further causes the processor to: - receiving the one or more localizer magnetic resonance images and the subject metadata; and - receiving the predicted field of view alignment data from the predictor algorithm in response to inputting the one or more localizer magnetic resonance images into the predictor algorithm and in response to inputting the subject metadata.

9. A method of operating a medical system, wherein: The medical system comprises a memory storing a predictor algorithm, wherein the predictor algorithm is configured to output predicted field of view alignment data in response to inputting one or more localizer magnetic resonance images and subject metadata, wherein the predictor algorithm comprises a trainable machine learning algorithm, wherein the memory further stores training data, wherein the training data comprises training entries, wherein each of the training entries comprises one or more training magnetic resonance images, training subject metadata, and training field of view alignment data; wherein the memory further comprises a training algorithm configured to train the predictor algorithm using a comparison between the predicted field of view alignment data and the training field of view alignment data, wherein the method comprises: - receiving the predicted field of view alignment data from the predictor algorithm in response to inputting the one or more training magnetic resonance images into the predictor algorithm and in response to inputting the training subject metadata; - determining said comparison between said predicted field of view alignment data and said training field of view alignment data; and - training said predictor algorithm using said training algorithm by inputting said comparison between said predicted field of view alignment data and said training field of view alignment data.

10. The method according to claim 9, wherein: The method further comprises generating the training data by extracting the one or more training magnetic resonance images, the training subject metadata and the training field of view alignment data from a medical image database.

11. The method according to claim 9 or 10, wherein: The method comprises: - receiving one or more localizer magnetic resonance images and subject metadata; and - receiving the predicted field of view alignment data from the predictor algorithm in response to inputting the one or more localizer magnetic resonance images into the predictor algorithm and in response to inputting subject metadata.

Images