Two-stage noise reduction for magnetic resonance imaging.

Abstract

Disclosed herein is a medical system 100, 300 having a memory 110 storing machine-executable instructions 120. The medical system further includes a computing system 104. Execution of the machine-executable instructions causes the computing system to perform the following steps: receiving (200) a series of complex-valued magnetic resonance images 126 representing each of a field of view of a subject according to a magnetic resonance imaging protocol; receiving (202) a series of complex-valued magnetic resonance images 128 by inputting the series of complex-valued magnetic resonance images to a first denoising module 122; calculating (204) a correction mapping 130 from the series of complex-valued denoised images; reconstructing (206) a corrected magnetic resonance image 132 using the correction mapping and the series of complex-valued magnetic resonance images; and receiving (208) a denoised magnetic resonance image 134 by inputting the corrected magnetic resonance images to a second noise filter module 124.

Description

[Technical Field]

[0001] The present invention relates to magnetic resonance imaging, and more particularly to denoising magnetic resonance images. [Background technology]

[0002] Magnetic resonance imaging (MRI) often suffers from poor signal-to-noise ratio (SNR) and various types of image artifacts due to subject motion, inhomogeneity of the main magnetic field B, imperfections in the gradient system, etc. Various methods aim to reduce noise, suppress artifacts, or both. Summary of the Invention [Problem to be solved by the invention]

[0003] For example, U.S. Patent Application Publication No. 2022026516A1 discloses a computer-implemented method for correcting phase and reducing noise in magnetic resonance (MR) phase images. The method includes executing a neural network model for analyzing MR images, where the neural network model is trained on pairs of original and corrupted images, where the corrupted images include corrupted phase information, the original images are corrupted images with the corrupted phase information subtracted, and a target output image of the neural network model is the original images. The method further includes receiving MR images including the corrupted phase information and analyzing the received MR images using the neural network model. The method also includes deriving an original phase image of the received MR images based on the analysis, where the derived original phase image has the corrupted phase information subtracted compared to the received MR images, and outputting the MR images based on the derived original phase image. [Means for solving the problem]

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

[0005] Many magnetic resonance imaging techniques rely on multiple acquisitions or sub-acquisitions of k-space data. The k-space data obtained from these acquisitions or sub-acquisitions is often inconsistent for various reasons, making it difficult to later combine them into a single resulting image. Furthermore, many magnetic resonance imaging techniques produce noisy k-space data, making it difficult to make the k-space data from these acquisitions or sub-acquisitions consistent, much less interpret the resulting image. Embodiments may provide an improved means for reducing noise and suppressing artifacts resulting from inconsistent k-space data in the resulting image. To accomplish this, a series of complex-valued magnetic resonance images is input to a first denoising module, which provides a series of complex-valued denoised images. A correction mapping is then performed using the series of complex-valued denoised images. This may be, for example, a phase correction mapping or a motion correction mapping for each image. The correction mapping is then applied to the original series of complex-valued magnetic resonance images to reconstruct a corrected magnetic resonance image. The denoised magnetic resonance images are then received by inputting the corrected magnetic resonance images to a second denoising module.

[0006] In the framework of the present invention, the concept of a denoised (complex-valued) image means that the denoised image is formed from an original noisy image by filtering out (uncorrelated or other small-scale) fluctuations that adversely affect the image content. This can be achieved by a conventional noise filter or by a trained denoising network that is trained on a large number of corresponding noisy and noisy images. This filtering out of noise preserves details of the image content at small spatial scales. Since the denoised complex-valued magnetic resonance images still contain the high spatial frequency structure of the image content, the correction mapping (e.g., for the phase component of the complex-valued image) also takes into account small high spatial frequency details. The correction mapping is then applied to the original series of complex-valued magnetic resonance images.

[0007] The first and second denoising modules can function differently. The first denoising module can remove uncorrelated noise. The second denoising module can remove noise from the corrected magnetic resonance image with its voxel or pixel phase values ​​already corrected. The second denoising model can be different, for example, in the case of a propeller (or MultiVane) acquisition, the first denoising module can remove uncorrelated noise from individual blades in the k-space acquired in the Cartesian acquisition, and the second denoising module can remove or reduce spatially correlated noise due to the blades acquired in the Cartesian acquisition, thereby forming a non-Cartesian acquisition with a radial acquisition in k-space.

[0008] The first and second denoising modules may function similarly. For example, in a diffusion-weighted image, denoising by the first denoising module may be highly error-prone if processed with a very poor signal-to-noise ratio. Despite the poor signal-to-noise ratio, the first denoising makes the correction mapping more robust. The phase-corrected average generated with the relatively robust correction mapping then produces a diffusion-weighted image with an improved signal-to-noise ratio, since the signal is largely unaffected while the noise is reduced. The results of the second denoising module then become more reliable, since the processing of the first denoising module prevents or at least reduces the propagation of artifacts in the final diffusion-weighted image.

[0009] The first denoising module can operate on complex-valued magnetic resonance images with spatial resolution equivalent to or similar to that of the final denoised magnetic resonance image of diagnostic quality, excluding any interpolation or super-resolution steps that may be performed in post-processing or incorporated into image reconstruction. When applied in combination with a propeller (MultiVane) acquisition of each blade at a different rotation in k-space, the denoising of the present invention can be applied to (intermediate) images reconstructed for each blade. That is, each intermediate image is reconstructed from the k-space data of each blade. For each blade, the k-space may have reduced spatial resolution due to limited k-space extension in the short axis direction of the k-space blade. However, the ultimate set of magnetic resonance images reconstructed from all k-space blades has full spatial resolution comparable to diagnostic image quality.

[0010] In one aspect, the present invention provides a medical system having a memory storing machine-executable instructions, the memory further storing a first denoising module, the memory further storing a second denoising module, and a computing system, wherein execution of the machine-executable instructions causes the computing system to receive a series of complex-valued magnetic resonance images acquired according to a magnetic resonance imaging protocol for each of a field of view of a subject.

[0011] The magnetic resonance imaging protocol can be specified or identified by the context of the complex-valued magnetic resonance images. For example, the format of the images may implicitly specify the magnetic resonance imaging protocol. A series of complex-valued magnetic resonance images may be accompanied by data or metadata identifying the magnetic resonance imaging protocol and / or other acquisition details, or there may be metadata accompanying the series of complex-valued magnetic resonance images that identifies the magnetic resonance imaging protocol.

[0012] Execution of the machine-executable instructions further causes the computing system to receive a series of complex-valued denoised images by inputting the series of complex-valued magnetic resonance images to a first denoising module. Execution of the machine-executable instructions further causes the computing system to calculate a correction mapping from the series of complex-valued denoised images. Execution of the machine-executable instructions further causes the computing system to reconstruct a corrected magnetic resonance image using the correction mapping and the series of complex-valued magnetic resonance images.

[0013] Execution of the machine-executable instructions further causes the computing system to receive denoised magnetic resonance images by inputting the corrected magnetic resonance images into a second denoising module. This embodiment may be advantageous because it may provide a means of generating denoised magnetic resonance images of superior quality. In some embodiments, a series of complex-valued magnetic resonance images is first processed through a first denoising module to provide a series of complex-valued denoised images. A correction mapping is calculated from the series of complex-valued denoised images. A corrected magnetic resonance image is then reconstructed from the original series of complex-valued magnetic resonance images using this correction mapping. Thus, the corrected magnetic resonance images are not corrupted by the first denoising module. This two-step process results in denoised magnetic resonance images with fewer artifacts.

[0014] In some examples, the first denoising module and the second denoising module may be the same. In other examples, the first denoising module and the second denoising module may be separate. As used herein, a denoising module refers to a software module that receives an image as input and provides a denoised image as output. The first denoising module and the second denoising module may be implemented using an algorithm or a neural network.

[0015] Examples of neural networks used for image denoising include DnCNN, Recurrent CNN, UNet, and Simplified DenseNet. In particular, Mohan et al., "Robust and interpretable blind image denoising via bias-free convolutional neural networks" (arXiv:1906.05478), describe deep convolutional neural networks suitable for implementing the first denoising neural network and / or the second denoising neural network.

[0016] In another embodiment, the correction mapping comprises a phase correction mapping calculated separately for each of the series of complex-valued denoised images. Execution of the machine-executable instructions further causes the computing system to calculate a series of phase-corrected images by applying the phase correction mapping to each of the series of complex-valued magnetic resonance images. The corrected magnetic resonance images are calculated from the series of phase-corrected images. This embodiment may be useful when the series of complex-valued magnetic resonance images are not phase-consistent.

[0017] In another embodiment, the phase correction mapping is calculated individually for each image in the sequence of complex-valued denoised images by fitting a parameterized polynomial to the spatially dependent phase. This embodiment may be advantageous as it provides a robust means of calculating the phase correction mapping individually for each image in the sequence of complex-valued denoised images.

[0018] As an example, a parameterized polynomial p'(x,y) of degree N for individually describing the spatially dependent phase p(x,y) of each image in a sequence of complex-valued denoised images is given by: TIFF2025528106000002.tif19124Here, the parameters cm and n are TIFF2025528106000003.tif19127, where m(x,y) is the spatially dependent magnitude.

[0019] In another embodiment, the magnetic resonance imaging protocol is a diffusion-weighted magnetic resonance imaging protocol. The corrected magnetic resonance image is calculated from the series of phase-corrected images by averaging the series of phase-corrected images. This embodiment may be beneficial when complex averaging of the series of complex-valued magnetic resonance images leads to signal cancellation due to phase variations in the series of complex-valued magnetic resonance images.

[0020] In another embodiment, the magnetic resonance imaging protocol is a bipolar readout magnetic resonance imaging protocol. Executing the machine-executable instructions causes the computing system to further calculate a corrected magnetic resonance image from the series of phase-corrected images by applying a phase correction mapping to the series of complex-valued magnetic resonance images to calculate a series of phase-corrected magnetic resonance images. Executing the machine-executable instructions causes the computing system to further calculate a corrected magnetic resonance image by calculating phase-corrected k-space data by transforming the series of phase-corrected images to k-space data. The corrected magnetic resonance image is then reconstructed from a combination of the phase-corrected k-space data. Essentially, each of the series of phase-corrected images is transformed to k-space. The corrected magnetic resonance image is then calculated using the combined k-space data. This combination can be achieved in different ways. In some examples, the k-space data is simply grouped and combined. In other examples, the k-space data is very similar and can be averaged.

[0021] In another embodiment, the correction mapping includes a main field inhomogeneity correction mapping calculated from the series of complex-valued denoised images. Execution of the machine-executable instructions further causes the computing system to calculate a series of phase-corrected images by applying the main field inhomogeneity correction mapping to each image in the series of complex-valued magnetic resonance images. The corrected magnetic resonance images are calculated from the series of phase-corrected images.

[0022] In another embodiment, the magnetic resonance imaging protocol is a Dixon magnetic resonance imaging protocol, and the corrected magnetic resonance image can be a Dixon magnetic resonance image, such as a water-only image or a fat-only image. This embodiment can be beneficial because it can provide an improved means of Dixon imaging.

[0023] In another embodiment, the magnetic resonance imaging protocol is an echo-planar magnetic resonance imaging protocol configured to acquire a series of complex-valued magnetic resonance images having two blip gradient polarities. The echo-planar magnetic resonance imaging protocol has a series of readout gradients of alternating polarity, during which blip gradients are applied perpendicular to the readout gradients to acquire k-space data along multiple lines in opposite directions after a single excitation. The polarity of the blip gradients determines the direction in which distortion occurs in the presence of main magnetic field inhomogeneities. The correction mapping includes a distortion correction mapping. Execution of the machine-executable instructions causes the computing system to further calculate the distortion correction mapping by sorting the series of complex-valued magnetic resonance images into a first group having a first blip gradient polarity and a second group having a second blip gradient polarity. Execution of the machine-executable instructions causes the computing system to further calculate the distortion correction mapping by registering images from the first group with images from the second group.

[0024] Execution of the machine-executable instructions further causes the computing system to calculate a series of distortion-corrected images by applying the distortion correction mapping to the series of complex-valued magnetic resonance images. A corrected magnetic resonance image is reconstructed by averaging the series of distortion-corrected images. The average of the distortion-corrected images can be either complex-valued or real-valued. For example, the average can be the average of the magnitude values ​​of the series of distortion-corrected images.

[0025] In another embodiment, the correction mapping includes a motion correction mapping to a reference image calculated from the series of complex-valued denoised images. In one example, the reference image can be selected from the series of complex-valued denoised images. In another example, the reference image can be another image from another source, such as a previously acquired image of the subject. One of the complex-valued denoised images is selected as a reference for all other images. A motion correction mapping is then determined that maps those images to the reference image. This embodiment can be beneficial because motion correction is more accurate when denoising is applied to these individual images.

[0026] In another embodiment, execution of the machine-executable instructions further causes the computing system to calculate a series of motion-corrected complex-valued magnetic resonance images by applying a motion correction mapping to the series of complex-valued magnetic resonance images. Execution of the machine-executable instructions further causes the computing system to calculate motion-corrected k-space data by transforming the series of motion-corrected magnetic resonance images into k-space. A corrected magnetic resonance image is reconstructed from the combination of the motion-corrected k-space data.

[0027] In this embodiment, a series of motion-corrected magnetic resonance images are first each converted to k-space data to form motion-corrected k-space data. The k-space data from the individual images is then combined together. This may involve simply aggregating the k-space data, or it may involve averaging the k-space data.

[0028] In another embodiment, the magnetic resonance imaging protocol is a segmented multi-shot magnetic resonance imaging protocol.

[0029] In another embodiment, the magnetic resonance imaging protocol is the PROPELLER magnetic resonance imaging protocol. In the PROPELLER magnetic resonance imaging protocol, a series of complex-valued magnetic resonance images are images reconstructed from individual "blades" of k-space data.

[0030] In another embodiment, the motion-compensated mapping is a rigid body mapping, in which the image-to-image transformation is a rigid body mapping without deformation.

[0031] In another embodiment, the motion compensated mapping is an affine mapping. In this embodiment, the image-to-image transformation is an affine mapping.

[0032] In another embodiment, the motion-compensated mapping is a displacement field.

[0033] Various algorithms can be used for rigid mapping, affine mapping, and displacement fields. For example, an anatomical atlas can be used to map points between different images. Alternatively, a shape-deformable model can be used.

[0034] In another embodiment, the medical system further comprises a magnetic resonance imaging system, and the memory further comprises pulse sequence commands configured to control the magnetic resonance imaging system to acquire measured k-space data according to a magnetic resonance imaging protocol, and execution of the machine-executable instructions causes the computing system to further control the magnetic resonance imaging system using the pulse sequence commands to acquire the measured k-space data, the k-space data comprising a plurality of discrete acquisition data each depicting a field of view of the subject.

[0035] Execution of the machine-executable instructions further causes the computing system to reconstruct a series of complex-valued magnetic resonance images representing each field of view of the subject from the measured k-space data, and in some examples, each of the discrete acquisitions is used to reconstruct a different image in the series of complex-valued magnetic resonance images.

[0036] In another aspect, the present invention provides a computer program comprising machine-executable instructions, the computer program being capable of being stored, for example, on a non-transitory storage medium, the computer program further comprising a first denoising module and a second denoising module.

[0037] Execution of the machine-executable instructions causes the computing system to receive a series of complex-valued magnetic resonance images representing each field of view of a subject in accordance with a magnetic resonance imaging protocol. Execution of the machine-executable instructions also causes the computing system to receive a series of complex-valued de-noised images by inputting the series of complex-valued magnetic resonance images to a first de-noising module. Execution of the machine-executable instructions also causes the computing system to calculate a correction mapping from the series of complex-valued de-noised images. Execution of the machine-executable instructions also causes the computing system to reconstruct a corrected magnetic resonance image using the correction mapping and the series of complex-valued magnetic resonance images. Execution of the machine-executable instructions also causes the computing system to receive a de-noised magnetic resonance image by inputting the corrected magnetic resonance images to a second de-noising module.

[0038] In another aspect, the present invention provides a method of medical imaging. The method includes receiving a series of complex-valued magnetic resonance images representing each of a field of view of a subject according to a magnetic resonance imaging protocol. The method further includes receiving a series of complex-valued de-noised images by inputting the series of complex-valued magnetic resonance images into a first de-noising module. The method further includes calculating a correction mapping from the series of complex-valued de-noised images. The method further includes reconstructing a corrected magnetic resonance image using the correction mapping and the series of complex-valued magnetic resonance images. The method further includes receiving a de-noised magnetic resonance image by inputting the corrected magnetic resonance images into a second de-noising module.

[0039] It will 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.

[0040] As will be appreciated by those skilled in the art, aspects of the present invention may be embodied as an apparatus, a method, or a 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, all of which may be referred to generally as a "circuit," "module," or "system" herein. Furthermore, aspects of the present invention may take the form of a computer program product embodied in one or more computer-readable medium(s) having computer-executable code embodied thereon.

[0041] 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 capable of storing instructions executable by a processor or computing system of a computing device. The computer-readable storage medium may also be referred to as a 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 store data accessible by the computing system 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, read-only memory, optical disks, magneto-optical disks, and computing system register files. 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 storage media that can be accessed by a computer device over a network or communications link. For example, data can be retrieved via a modem, over the Internet, or over a local area network. Computer-executable code embodied on a computer-readable medium can be transmitted using any suitable medium, including, but not limited to, wireless, wired, fiber optic cable, RF, etc., or any suitable combination of the foregoing.

[0042] A computer-readable signal medium may include, for example, a propagated data signal having computer-executable code embodied therein, 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, and such computer-readable medium is not a computer-readable storage medium, but is capable of communicating, propagating, or transmitting a program for use by or in connection with an instruction execution system, apparatus, or device.

[0043] "Computer memory" or "memory" is one example of a computer-readable storage medium. Computer memory is any memory directly accessible to a computing system. "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 be computer memory, or vice versa.

[0044] 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, including examples of a "computing system," should be interpreted as possibly including two or more computing systems or processing cores. A computing system may be, for example, a multi-core processor. A computing system may also refer to a collection of computing systems, either within a single computer system or distributed among multiple computer systems. The term computing system should also be interpreted as referring to a collection or network of computing devices, each of which includes a processor or computing system. Machine-executable code or instructions can be executed by multiple computing systems or processors, which may be within the same computing device or distributed across multiple computing devices.

[0045] Machine-executable instructions or computer-executable code may include instructions or programs that cause a processor or other computing system to perform aspects of the present invention. Computer-executable code for carrying out processes according to aspects of the present invention may be written in any combination of 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, compiled into machine-executable instructions. In some examples, the computer-executable code may be in the form of a high-level language or pre-compiled, or may be used in conjunction with an interpreter that generates machine-executable instructions on the fly. In other examples, the machine-executable instructions or computer-executable code may form a program for a programmable logic gate array.

[0046] The computer executable code may run entirely on the user's computer, partially on the user's computer, as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the latter situation, the remote computer may 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 may be connected to an external computer (e.g., via the Internet using an Internet Service Provider).

[0047] Aspects of the present invention will be described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block or portion of a block in the flowcharts, illustrations, and / or block diagrams, where applicable, can be implemented by computer program instructions in the form of computer-executable code. It will also be understood that combinations of blocks in different flowcharts, illustrations, and / or block diagrams may be combined, if not mutually exclusive. These computer program instructions are provided to a general-purpose computer, special-purpose computer, or other programmable data processing device computing system to generate a machine, such that the instructions, when executed via the computer or other programmable data processing device computing system, generate means for performing the functions / acts specified in one or more blocks of the flowcharts and / or block diagrams.

[0048] These machine-executable instructions or computer program instructions may be stored on a computer-readable medium that can cause a computer, other programmable data processing apparatus, or other device to function in a particular manner, such that the instructions stored on the computer-readable medium produce an article of manufacture that includes instructions that implement the functions / acts specified in the flowchart and / or block diagram blocks or blocks.

[0049] Furthermore, the machine-executable instructions or computer program instructions may be loaded into a computer, other programmable data processing device, or other device to cause the computer, other programmable device, or other device to perform a series of operational steps to create a computer-executed process, such that the instructions executing on the computer or other programmable device provide a process for performing the functions / operations specified in one or more blocks of the flowcharts and / or block diagrams.

[0050] 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" is also referred to as a "human interface device," and a user interface can provide information or data to an operator and / or receive information or data from an operator. A user interface can allow a computer to receive input from an operator and provide output from the computer to a user. In other words, a user interface allows an operator to control or manipulate a computer, and the interface may allow a computer to show 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, headset, pedals, wired gloves, remote control, and accelerometer are all examples of user interface components that allow receiving information or data from an operator.

[0051] As used herein, a "hardware interface" includes an interface that allows a computing system to interact with and / or control external computing devices and / or equipment. A hardware interface may allow a computing system to send control signals or instructions to external computing devices and / or equipment. A hardware interface may also allow a computing system to exchange data with external computing devices and / or equipment. 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 local area network connection, a TCP / IP connection, an Ethernet connection, a control voltage interface, a MIDI interface, an analog input interface, and a digital input interface.

[0052] As used herein, "display" or "display device" encompasses an output device or user interface adapted to display images or data. A display can 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), storage tubes, bi-stable displays, electronic paper, vector displays, flat panel displays, vacuum fluorescent displays (VFs), light emitting diode (LED) displays, electroluminescent displays (ELDs), plasma display panels (PDPs), liquid crystal displays (liquids), organic light emitting diode displays (OLEDs), projectors, and head-mounted displays.

[0053] K-space data is defined herein as the Fourier transform of a magnetic resonance image. In one example, k-space data may be recorded measurements of radio frequency signals emitted by atomic spins using the antenna of a magnetic resonance machine during a magnetic resonance imaging scan. In another example, k-space data is formed by taking the inverse Fourier transform of an existing magnetic resonance image.

[0054] A magnetic resonance image is defined herein as a reconstructed two-dimensional or three-dimensional visualization of anatomical data contained within magnetic resonance image data, which visualization can be performed, for example, by a computer.

[0055] Preferred embodiments of the present invention will now be described, by way of example only, with reference to the drawings in which: [Brief explanation of the drawings]

[0056] [Figure 1] FIG. 1 is a diagram illustrating an example of a medical system. [Figure 2] 2 is a flow chart illustrating a method of using the medical system of FIG. 1 . [Figure 3] FIG. 10 is a diagram showing another example of a medical system. [Figure 4] 4 is a flow chart illustrating a method of using the medical system of FIG. 3. [Figure 5] 10 is a flowchart illustrating another example of a method. [Figure 6] 10 is a flowchart illustrating another example of a method. [Figure 7] 10 is a flowchart illustrating another example of a method. [Figure 8] 10 is a flowchart illustrating another example of a method. [Figure 9] 10 is a flowchart illustrating another example of a method. [Figure 10] 10 is a flowchart illustrating another example of a method. [Figure 11] FIG. 1 shows the calculation of diffusion-weighted magnetic resonance images. [Figure 12] FIG. 1 illustrates denoising of individual images used to construct a diffusion-weighted magnetic resonance image. [Figure 13] FIG. 13 is a diagram showing phase correction of the image in FIG. 12. [Figure 14] 1 illustrates the complex combination of individual images for diffusion-weighted magnetic resonance imaging. DETAILED DESCRIPTION OF THE INVENTION

[0057] Like numbered components in these figures are either equivalent components or perform the same function. An aforementioned component is not necessarily described in a subsequent figure if the functionality is equivalent.

[0058] FIG. 1 illustrates an example of a medical system 100. The medical system 100 is shown as having a computer 102. The computer 102 is intended to represent one or more computers in one or more locations. The computer 102 is shown as having a computing system 104. Similarly, the computing system 104 is intended to represent one or more computing systems or cores located in one or more locations. The computing system 104 is shown as being in communication with an optional hardware interface 106 and an optional user interface 108. The hardware interface 106, if present, can be used to communicate with or control other components of the medical system 100, if present. For example, the hardware interface 106 can be used to control a magnetic resonance imaging system. The user interface 108 can be used by a user or technician to control the operation and functionality of the medical system 100.

[0059] Computing system 104 is further shown as being in communication with memory 110. Memory 110 is intended to represent one or more types of memory accessible by computing system 104. For example, memory 110 may be a non-transitory storage medium.

[0060] Memory 110 is shown as storing machine-executable instructions 120. Machine-executable instructions 120 enable computing system 104 to perform basic tasks, such as calculations and image processing tasks.

[0061] The memory 110 is further shown as including a first noise removal module 122 and a second noise removal module 124. Both noise removal modules 122 and 124 are configured to receive an image as an input and then output a denoised image. These noise removal modules 122 and 124 may be algorithmic or neural network based. In some examples, the first noise filter module 122 and the second noise removal module 124 are the same noise removal module.

[0062] The memory 110 is further shown as including a series of complex-valued magnetic resonance images 126. Each of these images 126 is a field of view of the subject according to a magnetic resonance imaging protocol. The magnetic resonance imaging protocol defines the technique used to acquire the series of complex-valued magnetic resonance images 126. The magnetic resonance imaging protocol may also specify relationships between the various images. For example, the series of complex-valued magnetic resonance images 126 may be accompanied by metadata that specifies the magnetic resonance imaging protocol. The series of complex-valued magnetic resonance images 126 may be encapsulated in a DICOM file that provides this information.

[0063] The memory 110 is further shown as containing a series of complex-valued denoised images 128. The series of complex-valued denoised images was obtained by inputting a series of complex-valued magnetic resonance images 126 into the first denoising module 122.

[0064] The memory 110 is further shown as including a correction mapping 130 calculated from the series of complex-valued denoised images 128. This correction mapping 130 can take different forms in different examples. In some examples, the correction mapping 130 can be a motion correction or a correction for distortion within the individual images 128. The memory 110 is further shown as including a corrected magnetic resonance image 132 calculated or reconstructed using the correction mapping 130 and the series of complex-valued magnetic resonance images 126. The memory 110 is further shown as including a denoised magnetic resonance image 134. The denoised magnetic resonance image 134 was constructed by inputting the corrected magnetic resonance image 132 into a second denoising module 124. It should be noted that in some examples, the first denoising module 122 and the second denoising module 124 are the same. In another example, the first noise removal module 122 can be tuned to a series of complex-valued magnetic resonance images 126, and the second noise removal module 124 can be tuned to a corrected magnetic resonance image 132.

[0065] Figure 2 is a flowchart illustrating a method of operation of the medical system 100 of Figure 1. First, in step 200, a series of complex-valued magnetic resonance images 126 is received. Then, in step 202, a series of complex-valued de-noised images 128 is received by inputting the series of complex-valued magnetic resonance images 126 into a first de-noising module 122. Then, in step 204, a correction mapping 130 is calculated from the series of complex-valued de-noised images 128. Then, in step 206, a corrected magnetic resonance image 132 is reconstructed using the correction mapping 130 and the series of complex-valued magnetic resonance images 126. Finally, in step 208, a de-noised magnetic resonance image 134 is received by inputting the corrected magnetic resonance image 132 into a second de-noising module 124.

[0066] Figure 3 shows another example of a medical system 300. The medical system 300 shown in Figure 3 is similar to the medical system 100 of Figure 1, except that it additionally includes a magnetic resonance imaging system 302 controlled by the computing system 104.

[0067] The magnetic resonance imaging system 302 includes a magnet 304. The magnet 304 is a cylindrical superconducting magnet with a bore 306 extending therethrough. Both segmented cylindrical magnets and so-called open magnets can be used. A segmented cylindrical magnet is similar to a standard cylindrical magnet except that the cryostat is divided into two sections to allow access to the magnet's isosurface. 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 between them large enough to accommodate a subject. The arrangement of the two sections is similar to that of a Helmholtz coil. Open magnets are popular because they provide less subject confinement. Inside the cryostat of a cylindrical magnet is a collection of superconducting coils.

[0068] Within the bore 306 of the cylindrical magnet 304 is an imaging zone 308 where the magnetic field is strong and uniform enough to perform magnetic resonance imaging. A field of view 309 is shown within the imaging zone 308. K-space data is acquired for the field of view 309. The region of interest may be identical to the field of view 309 or may be a sub-volume of the field of view 309. A subject 318 is shown supported by a subject support 320 such that at least a portion of the subject 318 is within the imaging zone 308 and the field of view 309.

[0069] Also within the magnet bore 306 are a set of magnetic field gradient coils 310 used for acquiring measured k-space data to spatially encode magnetic spins within the imaging zone 308 of the magnet 304. The magnetic field gradient coils 310 are connected to a magnetic field gradient coil power supply 312. The magnetic field gradient coils 310 are intended to be an example. Typically, the magnetic field gradient coils 310 have three separate coil sets for spatial encoding in three orthogonal spatial directions. The magnetic field gradient power supply supplies current to the magnetic field gradient coils 310. The current supplied to the magnetic field gradient coils 310 is controlled as a function of time and may be ramped or pulsed.

[0070] Adjacent to the imaging zone 308 is a radio frequency coil 314 for manipulating the direction of magnetic spins within the imaging zone 308 and for receiving radio transmissions from the spins within the imaging zone 308. A radio frequency antenna can have multiple coil elements. A radio frequency antenna can also be referred to as a channel or antenna. The radio frequency coil 314 is connected to a radio frequency transceiver 316. The radio frequency coil 314 and the radio frequency transceiver 316 can be replaced with separate transmit and receive coils and separate transmit and receive coils. It is understood that the radio frequency coil 314 and the radio frequency transceiver 316 are generic. The radio frequency coil 314 is also intended to represent a dedicated transmit antenna and a dedicated receive antenna. Similarly, the transceiver 316 may represent a separate transmitter and receiver. The radio frequency coil 314 can also have multiple receive / transmit elements, and the radio frequency transceiver 316 can have multiple receive / transmit channels. The transceiver 316 and the gradient controller 312 are shown as being connected to the hardware interface 106 of the computer system 102 .

[0071] Memory 110 is further shown as including pulse sequence commands 330 and measured k-space data 124. Pulse sequence commands are commands or data that can be converted into commands that can control magnetic resonance imaging system 302 to acquire measured k-space data 124.

[0072] Figure 4 is a flow chart illustrating a method of operation of the medical system 300 of Figure 3. The method illustrated in Figure 4 is similar to the method illustrated in Figure 2. The method of Figure 4 begins with step 400. In step 400, the magnetic resonance imaging system 302 is controlled by pulse sequence commands 330 to acquire measured k-space data 332. Then, in step 402, a series of complex-valued magnetic resonance images are reconstructed from the measured k-space data 332. After step 402 is performed, steps 200, 202, 204, 206, and 208 are performed as shown in Figure 2.

[0073] 5 shows another example of a method applied for diffusion-weighted imaging. Several acquired data 400 with discrete measurements 500 provide measured k-space data 332. These measured k-space data are then reconstructed 402 into a series of complex-valued magnetic resonance images 126, which are then denoised 202 by a first denoising module 122 to generate a series of complex-valued denoised images 128. This series of complex-valued denoised images 128 is then used to calculate or estimate 204 a correction mapping 130. In the example of diffusion-weighted imaging, the correction mapping 130 is a phase correction of the individual images 128. This phase correction 130 is applied to the series of complex-valued magnetic resonance images 126 to generate a series of phase-corrected images 502, and then an averaging process 206 is applied to generate an averaged corrected image or corrected magnetic resonance image 132. The corrected magnetic resonance image 132 is finally input to a second denoising module 124 to produce a denoised magnetic resonance image 134 .

[0074] FIG. 6 shows another example of a method applied to PROPELLER images. The acquisition includes a plurality of individual blades 400, each providing an acquisition of measured k-space data 332. These acquisitions of measured k-space data are reconstructed (402) into a series of complex-valued magnetic resonance images 126, one image per blade. Each of these blades is then denoised (202) using the first denoising module 122 to obtain a series of complex-valued denoised images 128. A correction mapping 130 is then calculated (204); in this example, the correction mapping 130 is motion corrected. The motion correction 130 is applied to the original series of complex-valued magnetic resonance images 126 to generate a series of motion-corrected images 602, which are then combined (206) to generate a combined or corrected magnetic resonance image 132. The motion-corrected images 602 can be transformed, for example, into k-space, and the resulting k-space data can be combined together. The combined k-space data is then reconstructed into a corrected magnetic resonance image 132. The corrected magnetic resonance image 132 is then denoised (208) to produce a denoised magnetic resonance image 134.

[0075] 7 illustrates another method applied for segmented multi-shot magnetic resonance imaging. Acquisition data is segmented into multiple interleaved shots 400, each providing an acquisition of measured k-space data 332. Again, these individual acquisitions are reconstructed (402) into a series of complex-valued magnetic resonance images 126, resulting in one image per shot. Each of these images 126 is then denoised (202) using a first denoising module 122 to generate a series of complex-valued denoised images 128. Motion estimation 204 is then performed to calculate a correction mapping 130 from the series of complex-valued denoised images 128. Again, the correction mapping 130 is motion corrected. This motion correction 130 is then applied individually to the original series of complex-valued magnetic resonance images 126 to generate a motion-corrected image 602. These motion-corrected images 602 are transformed back to k-space, and the resulting k-space data can be combined collectively 206. The combined k-space data is then reconstructed into a corrected magnetic resonance image 132. The corrected magnetic resonance image 132 is then denoised 208 using the second denoising module 124 to generate a denoised composite image or denoised magnetic resonance image 134.

[0076] 8 illustrates another method adapted for imaging with bipolar readout. In this example, two acquisitions 400 are performed with different readout gradient polarities. The measured k-space data 332 from these two acquisitions are again reconstructed (402) to provide two complex-valued magnetic resonance images 126, which are then denoised by a first denoising module 122 to generate a complex-valued denoised image 128. A correction mapping 130 is then calculated (204), which in this example is a phase correction. This correction mapping 130 is then applied to the complex-valued magnetic resonance image 126 to calculate a phase-corrected image 502. These two phase-corrected images 502 are then converted to k-space data and combined (206) to reconstruct the corrected magnetic resonance image 132. The corrected magnetic resonance image 132 is then denoised 208 by a second denoising module 124 to produce a denoised magnetic resonance image 134 .

[0077] FIG. 9 illustrates another method applied to echo-planar imaging with two blip gradient polarities. In this example, two acquisitions 400 are performed with different blip gradient polarities. Measured k-space data 332 from these two acquisitions are reconstructed (402) into two complex-valued magnetic resonance images 126. A first denoising module 122 is used to process the two complex-valued magnetic resonance images 126 to generate two complex-valued denoised images 128. Registration is then performed on these two complex-valued denoised images to calculate a correction mapping 130. In this example, the correction mapping 130 is a distortion correction mapping. This correction mapping 130 is then applied to the complex-valued magnetic resonance images 126 to generate a distortion-corrected image 902, which are then combined (206) to obtain the corrected magnetic resonance image 132. The corrected magnetic resonance image 132 is then denoised ( 208 ) using the second denoising module 124 to produce a denoised magnetic resonance image 134 .

[0078] 10 illustrates an example of a method applied to Dixon magnetic resonance imaging. In this example, two or more acquisitions 400 with different echo times are performed. The measured k-space data 332 from these acquisitions are reconstructed (402) into a series of complex-valued magnetic resonance images 126. Each of these images is then denoised (202) using a first denoising module 122. This produces a series of complex-valued denoised images 128. In this example, the calculation of the correction mapping 204 includes estimating a main magnetic field inhomogeneity correction mapping 130. This correction mapping 130 is then applied to the series of complex-valued magnetic resonance images 126 to produce a series of phase-corrected images 502. These phase-corrected images 502 are then used to calculate individual Dixon images 132, such as water-only and fat-only images. These individual Dixon images 132 are then provided with a denoised magnetic resonance image 134 by inputting them into a second denoising module 124, which in this example is a denoised Dixon magnetic resonance image.

[0079] Figures 11 to 14 illustrate the method applied to diffusion-weighted magnetic resonance imaging (DWI), which is primarily performed with single-shot echo-planar imaging (EPI) to reduce sensitivity to motion. However, even at moderate spatial resolution, the signal-to-noise ratio (SNR) is often low, necessitating multiple averaging. The phases of the individual averages are usually inconsistent due to motion, preventing immediate complex combination.

[0080] Under these circumstances, a diffusion-weighted magnetic resonance image 1100 is typically generated as shown in FIG. 11. In this prostate case, a total of 30 images were acquired for each slice, with 10 images acquired for each of the three orthogonal directions of the diffusion-sensitized gradient (M, P, S). The amplitude 1104 and phase 1106 of one selected image for each direction are shown in the first and second columns, respectively, with each row representing one direction. First, the amplitude images (M) are summed for each direction, and the phase images (P) are ignored. The amplitude sum 1108 for each direction is shown in the third column. The trace is then calculated as the geometric mean for each direction to obtain the trace image (T), i.e., the diffusion-weighted magnetic resonance image.

[0081] Combining the amplitudes of individual averages leads to the accumulation of unwanted background signals, while instantaneous complex combinations risk cancellation of foreground signals. Furthermore, the low SNR of the individual averages often precludes reliable phase correction and adequate instantaneous noise removal.

[0082] When the SNR is low, diffusion-weighted imaging (DWI) may undergo two stages of denoising. In the first stage, denoising of the individual averages is performed for reliable phase correction only. This phase correction is applied to the individual averages, followed by complex combination per direction. In the second stage, denoising of the resulting images further improves the SNR before combination per direction. In this way, the superior performance of complex averaging and denoising in low and high SNR regions, respectively, can be exploited, while avoiding the possibility of signal cancellation.

[0083] 12 shows a single complex-valued magnetic resonance image 1200 that is input into a noise filter 202 to produce a single complex-valued denoised image 1202. Because this is a complex image, the real component 1204, imaginary component 1206, amplitude 1208, and phase 1210 are shown for both images.

[0084] It is worth noting that due to the moderate spatial resolution of the acquisition and the very limited image area with perceptible signal, it is not practical to improve the SNR by simply reducing the spatial resolution.

[0085] 13 shows the magnitude image 1208 and phase image 1210 for a single image 1202 of the series of complex-valued denoised images 128, which are used to provide the phase correction applied to the single complex-valued magnetic resonance image 1200. The phase 1210 of this image and the resulting image 1300 are also shown. The phase estimation, in this example, involves a weighted least-squares fit of a constant and linear phase.

[0086] Figure 14 shows the complex combinations of the individual means per direction 1102. The amplitude of the complex combinations of the individual means for each direction is shown. The images in column 1400 represent the series of complex-valued magnetic resonance images 126. The images in column 1402 represent the series of complex-valued denoised images 128. The images in column 1404 represent the denoised magnetic resonance images 134. A significant improvement in image quality can be seen in comparison to Figure 11.

[0087] 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.

[0088] Other variations to the disclosed embodiments can be understood and effected by those skilled in the art in practicing the claimed invention, from a study of the drawings, the disclosure, and the appended claims. In the claims, the word "comprising" does not exclude other elements or steps, and the indefinite article "a" or "an" does not exclude a plurality. A single processor or other unit may fulfill the functions of several items recited in the claims. The mere fact that certain means are recited in mutually different dependent claims does not indicate that a combination of these means cannot be used to advantage. A computer program can be stored / distributed on a suitable medium, such as an optical storage medium or a solid-state medium, supplied together with 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. [Explanation of symbols]

[0089] 100 Medical Systems 102 Computer 104 Computing Systems 106 Hardware Interface 108 User Interface 110 memory 120 machine executable instructions 122 First noise reduction module 124 Second Denoising Module A series of 126 complex-valued magnetic resonance images A series of 128 complex-valued denoised images 130 Correction Mapping 132 Corrected Magnetic Resonance Imaging 134 Denoised Magnetic Resonance Images 200 receiving a series of complex-valued magnetic resonance images, each representing a field of view of the subject, according to a magnetic resonance imaging protocol; 202 receiving a series of complex-valued denoised images by inputting the series of complex-valued magnetic resonance images into a first denoising module; 204 Compute the correction mapping from a sequence of complex-valued denoised images 206 Reconstructing a corrected magnetic resonance image using the correction mapping and a series of complex-valued magnetic resonance images 208 receiving a denoised magnetic resonance image by inputting the corrected magnetic resonance image into a second denoising module; 300 Medical Systems 302 Magnetic Resonance Imaging System 304 Magnet 306 Magnet Bore 308 Imaging Zone 309 viewing angle 310 Magnetic Gradient Coil 312 Magnetic field gradient coil power supply 314 RF coil 316 Transceiver 318 subjects 320 Subject Support 330 Pulse Sequence Commands 332 measured k-space data 400 Controlling the magnetic resonance imaging system with pulse sequence commands to acquire measured k-space data 402 Reconstruct complex-valued magnetic resonance images for each field of view of the subject from the measured k-space data. 500 measured k-space data 332 individual acquisitions 502 series of phase-corrected images 602 Motion Compensated Images 902 Distortion Corrected Image 1100 Diffusion Weighted Imaging 1102 Orthogonal direction 1104 Amplitude Image 1106 Phase Image 1108 large and small images in total 1200 Complex-Valued Magnetic Resonance Images 1202 Complex-valued denoised image 1204 real component 1206 Imaginary component 1208 Amplitude 1210 Phase 1300 phase corrected images A total of 1400 complex-valued images Sum of 1402 complex-valued denoised images 1404 denoised image

Claims

1. 1. A medical system comprising: a memory storing machine-executable instructions, a first denoising module, and a second denoising module; A computing system; and wherein execution of the machine-executable instructions causes the computing system to: receiving a series of complex-valued magnetic resonance images, each representing a field of view of the subject, according to a magnetic resonance imaging protocol; receiving a series of complex-valued denoised images by inputting the series of complex-valued magnetic resonance images into a first denoising module; calculating a correction mapping from the series of complex-valued denoised images; reconstructing a corrected magnetic resonance image using the correction mapping and the series of complex-valued magnetic resonance images; receiving a denoised magnetic resonance image by inputting the corrected magnetic resonance image into a second denoising module; A medical system that performs the above.

2. 2. The medical system of claim 1, wherein the correction mapping comprises a phase correction mapping calculated individually for each of the series of complex-valued denoised magnetic resonance images, and wherein execution of the machine-executable instructions causes the computing system to further perform the step of calculating a series of phase-corrected images by applying the phase correction mapping to each of the series of complex-valued magnetic resonance images, and the corrected magnetic resonance images are calculated from the series of phase-corrected images.

3. 3. The medical system of claim 2, wherein the phase correction mapping is calculated individually for each of the series of complex-valued denoised images by fitting a parameterized polynomial to the spatially dependent phase of the series of complex-valued denoised images.

4. 4. The medical system of claim 2 or 3, wherein the magnetic resonance imaging protocol is a diffusion-weighted magnetic resonance imaging protocol, and the corrected magnetic resonance image is calculated from the series of phase-corrected images by averaging the series of phase-corrected images.

5. the correction mappings collectively comprise phase correction mappings calculated for the series of complex-valued denoised images, the magnetic resonance imaging protocol is a bipolar readout magnetic resonance imaging protocol, and execution of the machine-executable instructions further causes the computing system to: calculating a series of phase-corrected images by applying the phase-correction mapping to the series of complex-valued magnetic resonance images; calculating phase-corrected k-space data by transforming the series of phase-corrected images into k-space data, wherein the corrected magnetic resonance image is reconstructed from a combination of the phase-corrected k-space data; 2. The medical system of claim 1, further comprising the step of calculating the corrected magnetic resonance image from the series of phase-corrected images by performing:

6. 6. The medical system of claim 1, wherein the correction mapping comprises a main magnetic field inhomogeneity correction mapping calculated from the series of complex-valued denoised images, and wherein execution of the machine-executable instructions causes the computing system to further perform the step of calculating a series of phase-corrected images by applying the main magnetic field inhomogeneity correction mapping to each of the series of complex-valued magnetic resonance images, and the corrected magnetic resonance images are calculated from the series of phase-corrected images.

7. The medical system of claim 6 , wherein the magnetic resonance imaging protocol is a Dixon magnetic resonance imaging protocol and the corrected magnetic resonance image is a Dixon magnetic resonance image.

8. The magnetic resonance imaging protocol is an echo-planar magnetic resonance imaging protocol that acquires a series of complex-valued magnetic resonance images with two blip gradient polarities, and the correction mapping includes a distortion correction mapping, and execution of the machine-executable instructions further causes the computing system to: classifying the series of complex-valued magnetic resonance images into a first group according to a first blip gradient polarity and a second group according to a second blip gradient polarity; calculating a distortion correction mapping by registering the first group of images to the second group of images; calculating a series of distortion-corrected magnetic resonance images by applying the distortion correction mapping to the series of complex-valued magnetic resonance images, wherein the corrected magnetic resonance images are reconstructed by averaging the series of distortion-corrected magnetic resonance images; The medical system of claim 1 , wherein the step of calculating the distortion correction mapping is performed by executing:

9. The medical system of claim 1 , wherein the correction mapping comprises a motion correction mapping calculated from the series of complex-valued denoised images with respect to a reference image.

10. Execution of the machine-executable instructions further causes the computing system to: applying the motion correction mapping to the series of complex-valued magnetic resonance images to calculate a series of motion-corrected magnetic resonance images; calculating motion-corrected k-space data by transforming the series of motion-corrected magnetic resonance images into k-space data, wherein the corrected magnetic resonance image is reconstructed from a combination of the motion-corrected k-space data; The medical system of claim 9 , wherein the medical system executes the following:

11. The medical system of claim 10 , wherein the magnetic resonance imaging protocol is one of a segmented multi-shot magnetic resonance imaging protocol or a PROPELLER magnetic resonance imaging protocol.

12. The medical system of claim 9 , wherein the motion-corrected mapping is one of a rigid body mapping, an affine mapping, and a displacement field.

13. the medical system further comprises a magnetic resonance imaging system, the memory further comprises pulse sequence commands configured to control the magnetic resonance imaging system to acquire measured k-space data according to the magnetic resonance imaging protocol, and execution of the machine-executable instructions further causes the computing system to: controlling the magnetic resonance imaging system using the pulse sequence commands to acquire measured k-space data, the measured k-space data comprising a plurality of discrete acquisitions each depicting a field of view of a subject; reconstructing from the measured k-space data a series of complex-valued magnetic resonance images, each representing a field of view of the subject; The medical system according to claim 1 , further comprising:

14. 1. A computer program product having machine-executable instructions, a first denoising module, and a second denoising module, wherein execution of the machine-executable instructions causes the computing system to: receiving a series of complex-valued magnetic resonance images, each representing a field of view of the subject, according to a magnetic resonance imaging protocol; receiving a series of complex-valued denoised images by inputting the series of complex-valued magnetic resonance images into a first denoising module; calculating a correction mapping from the series of complex-valued denoised images; reconstructing a corrected magnetic resonance image using the correction mapping and the series of complex-valued magnetic resonance images; receiving a denoised magnetic resonance image by inputting the corrected magnetic resonance image into a second denoising module; A computer program that executes

15. 1. A medical imaging method comprising: receiving a series of complex-valued magnetic resonance images each representing a field of view of a subject according to a magnetic resonance imaging protocol; receiving a series of complex-valued denoised images by inputting the series of complex-valued magnetic resonance images into a first denoising module; calculating a correction mapping from the series of complex-valued denoised images; reconstructing a corrected magnetic resonance image using the correction mapping and the series of complex-valued magnetic resonance images; receiving a denoised magnetic resonance image by inputting the corrected magnetic resonance image into a second denoising module; A method having the following.

Images