System and method for shortening scan time for PROPELLER magnetic resonance imaging acquisition using deep learning reconstruction

Abstract

The present disclosure relates to systems and methods for shortening the scan time of PROPELLER magnetic resonance imaging acquisition using deep learning reconstruction. A system and method for shortening a scan time for periodic rotation overlapping parallel line-accompanying enhanced reconstruction (PROPELLER) imaging, the system and method including acquiring a plurality of k-space data blades of a region of interest during a PROPELLER sequence via a magnetic resonance imaging (MRI) scanner in a manner that rotates about a center of k-space, each blade of the plurality of k-space data blades includes a plurality of parallel phase encoding lines sampled in a phase encoding order. Each blade of the plurality of k-space data blades is undersampled. The system and method includes reconstructing each of the plurality of k-space data leaves individually and separately using a deep learning-based Cartesian-like reconstruction network to generate a plurality of fully sampled leaves. The system and method include generating a complex image from a plurality of fully sampled leaves using a PROPELLER reconstruction algorithm.

Description

Background Art

[0001] The subject matter disclosed herein relates to medical imaging, and more particularly, to a system and method for reducing scan time of PROPELLER magnetic resonance imaging acquisitions using deep learning reconstruction.

[0002] Non-invasive imaging techniques allow images of internal structures or features of a patient / object to be obtained without performing an invasive procedure on the patient / object. Specifically, such non-invasive imaging techniques rely on various physical principles (such as differential transmission of X-rays through a target volume, reflection of acoustic waves within a volume, paramagnetism of different tissues and materials within a volume, decomposition of target radionuclides within the body, etc.) to acquire data and construct an image or otherwise represent the observed internal features of the patient / object.

[0003] During magnetic resonance imaging (MRI), when a substance such as human tissue is subjected to a uniform magnetic field (polarization field B0), the individual magnetic moments of the spins in the tissue attempt to align with the polarization field, but precess around it in a random order at their characteristic Larmor frequency. If the substance or tissue is subjected to a magnetic field (excitation field B1) that is in the xy plane and close to the Larmor frequency, the net alignment moment or "longitudinal magnetization" M z can be rotated or "tilted" into the xy plane to produce a net transverse magnetic moment M t After the excitation signal B1 is terminated, a signal is emitted by the excited spins and can be received and processed to form an image.

[0004] When these signals are used to generate images, the magnetic field gradient (G x , G y and G z ). Typically, the area to be imaged is scanned in a series of measurement cycles in which these gradient fields are varied according to the particular localization method used. The resulting set of received nuclear magnetic resonance (NMR) signals is digitized and processed to reconstruct an image using one of a number of well-known reconstruction techniques.

[0005] Cartesian acquisition is the most widely used technique for k-space acquisition. However, there are several techniques that can acquire k-space data on a non-Cartesian grid. Periodically Rotated Overlapping Parallel Lines with Enhanced Reconstruction (PROPELLER) can be considered a hybrid of Cartesian and non-Cartesian. PROPELLER is an MR technique that provides high-resolution magnetic resonance imaging with reduced motion artifacts by providing the ability to remove leaves affected by motion and by oversampling low spatial frequencies. PROPELLER scans are critical scans at clinical sites, especially for anatomical structures that are at risk of being affected by motion, such as the abdomen, pelvis, and cervical spine. However, PROPELLER scans have longer scan times than those using Cartesian techniques. Summary of the invention

[0006] The following shows an overview of certain embodiments disclosed herein. It should be understood that these aspects are provided only to provide the reader with a brief overview of these specific embodiments, and these aspects are not intended to limit the scope of the present disclosure. In fact, the present disclosure may cover various aspects that may not be shown below.

[0007] In one embodiment, a computer-implemented method for shortening the scan time of periodically rotated overlapping parallel lines with enhanced reconstruction (PROPELLER) imaging. The computer-implemented method includes acquiring, via a processor, a plurality of k-space data leaves of a region of interest in an accelerated manner during a PROPELLER sequence via a magnetic resonance imaging (MRI) scanner in a manner rotating around the center of k-space, wherein each of the plurality of k-space data leaves includes a plurality of parallel phase-encoded lines sampled in a phase-encoding order. Each of the plurality of k-space data leaves is undersampled. The computer-implemented method also includes, via the processor, utilizing a deep learning-based Cartesian-like reconstruction network to individually and separately reconstruct each of the plurality of k-space data leaves to generate a plurality of fully sampled leaves. The computer-implemented method also includes utilizing, via the processor, a PROPELLER reconstruction algorithm to generate a complex image from the plurality of fully sampled leaves.

[0008] In another embodiment, a system for shortening the scan time of periodically rotated overlapping parallel lines with enhanced reconstruction (PROPELLER) imaging. The system includes a memory that encodes a processor executable routine. The system also includes a processor that is configured to access the memory and execute the processor executable routine, wherein the routine causes the processor to perform an action when executed by the processor. The action includes acquiring multiple k-space data leaves of a region of interest in an accelerated manner during a PROPELLER sequence via a magnetic resonance imaging (MRI) scanner in a manner rotating around the center of k-space, wherein each of the multiple k-space data leaves includes multiple parallel phase encoding lines sampled in a phase encoding order. Each of the multiple k-space data leaves is undersampled. The action also includes using a deep learning-based Cartesian-like reconstruction network to individually and separately reconstruct each of the multiple k-space data leaves to generate multiple fully sampled leaves. The action also includes generating a complex image from multiple fully sampled leaves using the PROPELLER reconstruction algorithm.

[0009] In another embodiment, a non-transitory computer-readable medium includes processor executable code that causes the processor to perform an action when executed by the processor. The action includes acquiring multiple k-space data leaves of a region of interest in an accelerated manner during a PROPELLER sequence via a magnetic resonance imaging (MRI) scanner in a manner rotating around the center of k-space, wherein each of the multiple k-space data leaves includes multiple parallel phase encoding lines sampled in a phase encoding order. Each of the multiple k-space data leaves is undersampled. The action also includes using a deep learning-based Cartesian-like reconstruction network to individually and separately reconstruct each of the multiple k-space data leaves to generate multiple fully sampled leaves. The action also includes generating a complex image from the multiple fully sampled leaves using the PROPELLER reconstruction algorithm. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] These and other features, aspects, and advantages of the present subject matter will be better understood when the following detailed description is read with reference to the accompanying drawings, in which like characters refer to like parts throughout the several views, and in which:

[0011] Figure 1 An embodiment of a magnetic resonance imaging (MRI) system suitable for use with the disclosed technology is illustrated;

[0012] Figure 2 A flow chart illustrating a method for reducing scan time for Periodically Rotated Overlapping Parallel Lines with Enhanced Reconstruction (PROPELLER) imaging according to aspects of the present disclosure;

[0013] Figure 3 A more detailed example of the method according to various aspects of the present disclosure is Figure 2 Schematic diagram of the method in;

[0014] Figure 4 A flow chart illustrating a method for training a deep learning based Cartesian-like reconstruction network according to aspects of the present disclosure;

[0015] Figure 5 A schematic diagram illustrating a deep learning-based Cartesian-like reconstruction network and its utilization according to various aspects of the present disclosure;

[0016] Figure 6 An example of a mask of a fully sampled mask acquired by Periodically Rotated Overlapping Parallel Lines with Enhanced Reconstruction (PROPELLER) according to aspects of the present disclosure is illustrated;

[0017] Figure 7 An example of non-uniform undersampling of Periodically Rotated Overlapping Parallel Lines with Enhanced Reconstruction (PROPELLER) acquisitions according to aspects of the present disclosure is illustrated;

[0018] Figure 8 An example of uniform undersampling of a periodically rotated overlapping parallel lines with enhanced reconstruction (PROPELLER) acquisition according to aspects of the present disclosure is illustrated;

[0019] Fig. 9 depicts a Periodically Rotated Overlapping Parallel Lines Enhanced Reconstruction (PROPELLER) image obtained using zero-filled image reconstruction and deep learning-based leaf-level reconstruction of a subject's abdomen in accordance with aspects of the present disclosure;

[0020] Fig.10 depicts a periodic rotational overlapping parallel lines with enhanced reconstruction (PROPELLER) image of a subject's abdomen in accordance with aspects of the present disclosure;

[0021] Fig.11 depicts a periodic rotational overlaid parallel lines with enhanced reconstruction (PROPELLER) image of a subject's spine in accordance with aspects of the present disclosure;

[0022] Fig.12 depicts a periodically rotated overlaid parallel lines with enhanced reconstruction (PROPELLER) image of a subject's pelvis in accordance with aspects of the present disclosure; and

[0023] Fig.13Depicted are Periodically Rotated Overlapping Parallel Lines with Enhanced Reconstruction (PROPELLER) images of a subject's pelvis in accordance with aspects of the present disclosure (eg, depicting the compatibility of artificial intelligence-based denoising with deep learning-based blade-level reconstruction). DETAILED DESCRIPTION

[0024] One or more specific embodiments will be described below. In order to provide a concise description of these embodiments, not all features of an actual implementation are described in the specification. It should be understood that in the development of any such actual implementation, as in any engineering or design project, many implementation-specific decisions must be made to achieve the developer's specific goals, such as complying with system-related and business-related constraints that may vary from implementation to implementation. In addition, it should be understood that such development efforts may be complex and time-consuming, but are still routine tasks for design, fabrication, and manufacturing for ordinary technicians who benefit from this disclosure.

[0025] When introducing elements of various embodiments of the present subject matter, the articles "a," "an," "the," and "said" are intended to indicate that there are one or more elements. The terms "comprising," "including," and "having" are intended to be inclusive, and mean that there may be additional elements in addition to the listed elements. Furthermore, any numerical examples in the following discussion are intended to be non-limiting, and thus the additional numerical values, ranges, and percentages are within the scope of the disclosed embodiments.

[0026] Although various aspects of the following discussion are provided in the context of medical imaging, it should be understood that the technology disclosed in the present invention is not limited to such medical contexts. In fact, the examples and explanations are provided in such medical contexts only to facilitate explanation by providing examples of realistic specific implementations and applications. However, the technology disclosed in the present invention can also be used in other contexts, such as non-destructive inspection of manufactured parts or goods (i.e., quality control or quality review application scenarios) and / or image reconstruction of non-invasive inspection of packages, boxes, suitcases, etc. (i.e., security inspection or screening application scenarios). In general, the technology disclosed in the present invention can be used in any imaging or screening context or image processing or photography field, in which a group or class of collected data undergoes a reconstruction process to generate an image or volume.

[0027] The deep learning (DL) methods discussed herein may be based on artificial neural networks and thus may encompass one or more of the following: deep neural networks, fully interconnected networks, convolutional neural networks (CNNs), unfolded neural networks, perceptrons, codecs, recurrent networks, transformer networks, wavelet filter banks, u-nets, generative adversarial networks (GANs), dense neural networks (e.g., residual dense networks (RDNs)), or other neural network architectures. Neural networks may include shortcuts, activations, batch normalization layers, and / or other features. These techniques are referred to herein as DL techniques, although the term may also be used with particular reference to the use of deep neural networks, which are neural networks with multiple layers.

[0028] As discussed herein, DL techniques (which may also be referred to as deep machine learning, hierarchical learning, or deep structured learning) are a branch of machine learning techniques that employ mathematical representations of data and artificial neural networks for learning and processing such representations. For example, DL methods may be characterized as using one or more algorithms to extract or model highly abstract concepts of a class of data of interest. This may be accomplished using one or more processing layers, where each layer typically corresponds to a different level of abstraction, and thus may employ or utilize different aspects of the initial data or the output of the previous layer (i.e., a hierarchical or cascaded structure of layers) as the target of the process or algorithm of a given layer. In the context of image processing or reconstruction, this may be characterized as different layers corresponding to different feature levels or resolutions in the data. In general, the processing of one representation space to the next level representation space may be viewed as a "stage" of the process. Each stage of the process may be performed by a separate neural network or by different parts of a larger neural network.

[0029] The present disclosure provides systems and methods for reducing the scan time of periodic rotation overlapping parallel lines with enhanced reconstruction (PROPELLER) imaging. Periodic rotation overlapping parallel lines with enhanced reconstruction is an MR technique that provides high-resolution magnetic resonance imaging with reduced motion artifacts by providing the ability to remove leaves affected by motion and by oversampling low spatial frequencies. In periodic rotation overlapping parallel lines with enhanced reconstruction, overlapping leaves are acquired by rotating around k-space.

[0030] The systems and methods disclosed herein include acquiring multiple k-space data leaves of a region of interest in an accelerated manner via a magnetic resonance imaging (MRI) scanner in a manner rotating around the center of k-space during a periodic rotation overlapping parallel lines with enhanced reconstruction (PROPELLER) sequence, wherein each of the multiple k-space data leaves includes multiple parallel phase encoding lines sampled in a phase encoding order. Each of the multiple k-space data leaves is undersampled (e.g., has fewer phase encoding lines relative to a fully sampled leaf). In some embodiments, the multiple k-space data leaves may be sampled in a uniform manner. In some embodiments, the multiple k-space data leaves may be sampled in a non-uniform manner. The systems and methods disclosed herein also include using a deep learning-based Cartesian-like reconstruction network to individually and separately reconstruct each of the multiple k-space data leaves to generate multiple fully sampled leaves. The systems and methods disclosed herein also include generating a complex image from multiple fully sampled leaves using a PROPELLER reconstruction algorithm.

[0031] In some embodiments, the deep learning-based Cartesian reconstruction network is configured to be performed using any number of k-space data leaves. In some embodiments, the multiple k-space data leaves are acquired from a single receiver coil. In some embodiments, wherein the multiple k-space data leaves are acquired from multiple receiver coils. In some embodiments, the multiple k-space data leaves include a skew aspect ratio.

[0032] In some embodiments, the deep learning-based Cartesian-like reconstruction network includes a deep learning-based network based on an unfolding algorithm. In some embodiments, the systems and methods disclosed herein train the deep learning-based Cartesian-like reconstruction network on input-output data pairs using supervised learning. The input-output data pairs include undersampled k-space leaf images and corresponding fully sampled k-space images acquired using a periodically rotated overlapping parallel line companion enhancement reconstruction sequence. The undersampled k-space leaf images are generated from the corresponding fully sampled k-space images.

[0033] In certain embodiments, a non-transitory computer-readable medium includes processor executable code that causes the processor to perform an action when executed by the processor. The action includes acquiring multiple k-space data leaves of a region of interest in an accelerated manner via a magnetic resonance imaging (MRI) scanner in a manner rotating around the center of k-space during a PROPELLER sequence, wherein each of the multiple k-space data leaves includes multiple parallel phase encoding lines sampled in a phase encoding order. Each of the multiple k-space data leaves is undersampled (e.g., there are fewer phase encoding lines relative to a fully sampled leaf). The action also includes using a deep learning-based Cartesian-like reconstruction network to individually and separately reconstruct each of the multiple k-space data leaves to generate multiple fully sampled leaves. The action also includes generating a complex image from multiple fully sampled leaves using the PROPELLER reconstruction algorithm.

[0034] Embodiments disclosed herein provide a technique utilizing a Cartesian-like reconstruction scheme at the leaf level for performing accelerated Periodic Rotation Overlapping Parallel Lines with Enhanced Reconstruction (PROPELLER) acquisition while maintaining all remaining elements of the PROPELLER image reconstruction undisturbed in the reconstruction chain. Embodiments disclosed herein enable faster magnetic resonance imaging scans utilizing periodic rotation overlapping parallel lines with enhanced reconstruction scans. Embodiments disclosed herein also increase system throughput using magnetic resonance imaging systems. Embodiments disclosed herein may also be utilized with both Cartesian under-sampling reconstruction and PROPELLER under-sampling reconstruction.

[0035] Considering the above, Figure 1 , a magnetic resonance imaging (MRI) system 100 is schematically illustrated as including a scanner 102, scanner control circuitry 104, and system control circuitry 106. According to embodiments described herein, the magnetic resonance imaging system 100 is generally configured to perform magnetic resonance imaging.

[0036] The system 100 also includes a remote access and storage system or device, such as a picture archiving and communication system (PACS) 108, or other devices, such as teleradiology equipment, that enable on-site or off-site access to data acquired by the system 100. In this way, MR data can be acquired and then processed and evaluated on-site or off-site. Although the magnetic resonance imaging system 100 can include any suitable scanner or detector, in the illustrated embodiment, the system 100 includes a whole body scanner 102 having a housing 120 through which an aperture 122 is formed. An examination table 124 can be moved into the aperture 122 to allow a patient 126 (e.g., a subject) to be positioned therein to image selected anatomical structures within the patient's body.

[0037] The scanner 102 includes a series of associated coils for generating a controlled magnetic field for exciting gyromagnetic materials within the anatomical structure of the patient being imaged. Specifically, a primary magnetic coil 128 is provided for generating a primary magnetic field B0 that is generally aligned with the orifice 122. A series of gradient coils 130, 132, and 134 allow controlled gradient magnetic fields to be generated during an examination sequence for position encoding of certain gyromagnetic nuclei within the patient 126. A radio frequency (RF) coil 136 (e.g., an RF transmit coil) is configured to generate radio frequency pulses for exciting certain gyromagnetic nuclei within the patient. In addition to the coils that may be located locally at the scanner 102, the system 100 also includes a set of receiving coils or RF receiving coils 138 (e.g., a coil array) configured to be placed proximal to the patient 126 (e.g., against the patient). For example, the receiving coil 138 may include a cervical / thoracic / lumbar (CTL) coil, a head coil, a single-sided spine coil, etc. Generally, the receive coil 138 is placed near or on top of the patient 126 to receive the weak RF signals (weak relative to the transmit pulses generated by the scanner coil) generated by certain gyromagnetic nuclei in the patient's body when the patient 126 returns to his or her relaxed state.

[0038] The various coils of the system 100 are controlled by external circuitry to generate the required fields and pulses and to read the emissions from the gyromagnetic material in a controlled manner. In the illustrated embodiment, the main power supply 140 provides power to the primary field coil 128 to generate the main magnetic field B0. The power input (e.g., power from a utility or grid), a power distribution unit (PDU), a power supply (PS), and a drive circuit 150 can provide power together to pulse the gradient field coils 130, 132, and 134. The drive circuit 150 may include an amplification and control circuit system for supplying current to the coils as defined by the digitized pulse sequence output by the scanner control circuit system 104.

[0039] Another control circuit 152 is provided for regulating the operation of the RF coil 136. The circuit 152 includes a switching device for alternating between an active mode of operation and a passive mode of operation, wherein the RF coil 136 transmits and does not transmit a signal, respectively. The circuit 152 also includes an amplification circuit system configured to generate RF pulses. Similarly, the receiving coil 138 is connected to a switch 154 that is capable of switching the receiving coil 138 between a receiving mode and a non-receiving mode. Thus, in the receiving mode, the receiving coil 138 resonates with the RF signal generated by the release of the gyromagnetic nuclei in the patient 126, and in the non-receiving mode, they do not resonate with the RF energy from the transmitting coil (i.e., the coil 136) to prevent undesired operation. In addition, the receiving circuit 156 is configured to receive data detected by the receiving coil 138, and may include one or more multiplexing and / or amplification circuits.

[0040] It should be noted that although the scanner 102 and control / amplification circuit system described above are illustrated as being coupled by a single line, in actual examples there may be many such lines. For example, separate lines may be used for control, data communication, power transmission, etc. In addition, appropriate hardware may be provided along each type of line for properly processing data and current / voltage. In practice, various filters, digitizers, and processors may be provided between the scanner and either or both of the scanner control circuit 104 and the system control circuit system 106.

[0041] As shown, the scanner control circuit system 104 includes an interface circuit 158 ​​that outputs signals for driving the gradient field coils and the RF coils and for receiving data representing magnetic resonance signals generated in the examination sequence. The interface circuit 158 ​​is coupled to a control and analysis circuit 160. Based on a defined scheme selected via the system control circuit 106, the control and analysis circuit 160 executes commands for driving the circuits 150 and 152.

[0042] The control and analysis circuitry 160 is also used to receive magnetic resonance signals and perform subsequent processing before transmitting the data to the system control circuitry 106. The scanner control circuitry 104 also includes one or more memory circuits 162 that store configuration parameters, pulse sequence descriptions, examination results, etc. during operation.

[0043] Interface circuit 164 is coupled to control and analysis circuit 160 for exchanging data between scanner control circuit system 104 and system control circuit system 106. In some embodiments, control and analysis circuit 160, although illustrated as a single unit, may include one or more hardware devices. System control circuit 106 includes interface circuit 166, which receives data from scanner control circuit system 104 and transmits data and commands back to scanner control circuit system 104. Control and analysis circuit 168 may include a CPU in a general or special purpose computer or workstation. Control and analysis circuit 168 is connected to memory circuit 170 to store programming code for operating magnetic resonance imaging system 100, and to store processed image data for later reconstruction, display and transmission. The programming code may execute one or more algorithms that are configured to perform reconstruction of acquired data as described below when executed by a processor. In some embodiments, memory circuit 170 may store one or more neural networks (e.g., similar Cartesian reconstruction networks based on deep learning) for processing and / or reconstructing acquired data, as described below. In certain embodiments, image reconstruction may occur on a separate computing device having processing circuitry and memory circuitry.

[0044] The processing components (e.g., microprocessors or processing circuitry) and memory (such as may be present in the scanner control circuitry 104 and / or the system control circuitry 106) of the magnetic resonance imaging system 100 may be used to execute stored software code, instructions or routines for acquiring and processing MR data. As used herein, the term "code" or "software code" refers to any instructions or instruction sets for controlling the magnetic resonance imaging system 100. The code or software code may be in a computer executable form, such as machine code, which is a set of instructions and data that is directly executed by the processing components of the scanner control circuitry 104 and / or the system control circuitry 106; a human understandable form, such as source code, which may be compiled for execution by the processing components of the scanner control circuitry 104 and / or the system control circuitry 106; or an intermediate form, such as object code, which is generated by a compiler. In some embodiments, the magnetic resonance imaging system 100 may include multiple controllers.

[0045] For example, the memory may store processor executable software code or instructions (e.g., firmware or software) that are tangibly stored on a non-transitory computer readable medium. Additionally or alternatively, the memory 46 may store data. For example, the memory may include volatile memory (such as random access memory (RAM)) and / or non-volatile memory (such as read-only memory (ROM), flash memory, hard drive, or any other suitable optical, magnetic or solid-state storage medium or a combination thereof. In addition, the processing unit may include multiple microprocessors, one or more "general" microprocessors, one or more special-purpose microprocessors, and / or one or more application-specific integrated circuits (ASICS) or some combination thereof. For example, the processing unit may include one or more reduced instruction set (RISC) or complex instruction set (CISC) processors. The processing unit may include multiple processors and / or the memory may include multiple memory devices.

[0046] The processing component is configured to acquire multiple k-space data leaves of a region of interest in an accelerated manner via a magnetic resonance imaging (MRI) scanner in a manner rotating around the center of k-space during a periodic rotation overlapping parallel lines with enhanced reconstruction (PROPELLER) sequence, wherein each of the multiple k-space data leaves includes multiple parallel phase encoding lines sampled in a phase encoding order. Each of the multiple k-space data leaves is undersampled (e.g., has fewer phase encoding lines relative to a fully sampled leaf). In some embodiments, the multiple k-space data leaves may be sampled in a uniform manner. In some embodiments, the multiple k-space data leaves may be sampled in a non-uniform manner. The processing component is also configured to utilize a deep learning-based Cartesian-like reconstruction network to individually and separately reconstruct each of the multiple k-space data leaves to generate multiple fully sampled leaves. The processing component is also configured to generate a complex image from the multiple fully sampled leaves using a PROPELLER reconstruction algorithm.

[0047] In some embodiments, the deep learning-based Cartesian reconstruction network is configured to be performed using any number of k-space data leaves. In some embodiments, the multiple k-space data leaves are acquired from a single receiver coil. In some embodiments, wherein the multiple k-space data leaves are acquired from multiple receiver coils. In some embodiments, the multiple k-space data leaves include a skew aspect ratio.

[0048] In some embodiments, the deep learning-based Cartesian reconstruction network includes a deep learning-based network based on an unfolding algorithm. In some embodiments, the processing component is configured to train the deep learning-based Cartesian reconstruction network on input-output data pairs using supervised learning. The input-output data pairs include undersampled k-space leaf images (e.g., fewer phase encoding lines relative to fully sampled leaves) and corresponding fully sampled k-space images acquired using a periodically rotated overlapping parallel line companion enhancement reconstruction sequence. Undersampled k-space leaf images are generated from the corresponding fully sampled k-space images.

[0049] Additional interface circuitry 172 may be provided for exchanging image data, configuration parameters, etc. with external system components (such as remote access and storage device 108). Finally, system control and analysis circuitry 168 may be communicatively coupled to various peripheral devices for facilitating operator interface and producing hard copies of reconstructed images. In the illustrated embodiment, these peripheral devices include a printer 174, a monitor 176, and a user interface 178, which includes devices such as a keyboard, a mouse, a touch screen (e.g., integral to the monitor 176), and the like.

[0050] Figure 2 1 is a flow chart of a method 180 for reducing scan time for periodically rotated overlapping parallel lines with enhanced reconstruction (PROPELLER) imaging. One or more steps of the method 180 may be performed by Figure 1 The processing circuitry of the magnetic resonance imaging system 100 or a remote computing system may be executed. For example, the processing circuitry may be part of the scanner control circuitry 104 and / or the system control circuitry 106 of the magnetic resonance imaging system 100. Figure 2 One or more steps of method 180 are performed in different orders as shown. Method 180 is performed for data acquired from a single receiver coil (e.g., an RF receive coil) or multiple receiver coils (e.g., an RF receive body coil or an RF receive surface coil). Method 180 is independent of the number of blades in the PROPELLER image. Therefore, the number of blades in the PROPELLER image is not limited, and method 180 can be used with any number of blades. Method 180 is performed in image space.

[0051] Method 180 includes transmitting a magnetic resonance imaging scanner (e.g., Figure 1A magnetic resonance imaging scanner 102 in the apparatus (eg, in a scanner 102) acquires a plurality of k-space data blades of a region of interest in a rotational manner about the center of k-space (eg, rotating approximately 10 to 20 degrees between each blade acquisition) (block 182). Each of the plurality of k-space data blades includes a plurality of parallel phase encoding lines (eg, having a straight line shape) sampled in a phase encoding order using a fast spin echo method or a gradient echo method (ie, each blade is filled by an echo chain of a corresponding MR pulse sequence). Each of the plurality of k-space blades is undersampled. Specifically, fewer phase encoding lines are acquired for each of the plurality of k-space data blades. In some embodiments, the plurality of k-space data blades are undersampled in a uniform manner. In some embodiments, the plurality of k-space data blades are undersampled in a non-uniform manner. The plurality of undersampled k-space data blades have a skewed aspect ratio. Each blade is acquired in a true Cartesian manner.

[0052] The method 180 also includes utilizing a deep learning-based Cartesian-like reconstruction network to individually and separately reconstruct each of the plurality of k-space data leaves to generate a plurality of fully sampled leaves (block 184). The deep learning-based Cartesian-like reconstruction network is trained to reconstruct fully sampled k-space leaf images from the undersampled k-space leaf images. Figure 4 Training a deep learning based Cartesian-like reconstruction network is described in . The deep learning based Cartesian-like reconstruction network is configured to be performed (and also trained) with any number of k-space data leaves. The deep learning based Cartesian-like reconstruction network includes a deep learning based network based on an unfolding algorithm.

[0053] Method 180 also includes generating a complex image of the region of interest from multiple fully sampled (reconstructed) leaves using a PROPELLER reconstruction algorithm (box 186). For example, the PROPELLER reconstruction algorithm includes: phase correction for each leaf to ensure that its rotation point is precisely located at the center of k-space; correction for the object's overall in-plane rotation and in-plane translation; and correlation weighting to minimize data from leaves that contain motion or displacement errors. In some embodiments, the complex image is further processed before the image is written to a Digital Imaging and Communications in Medicine (DICOM) file. It should be noted that since the Cartesian-like reconstruction is a leaf-level reconstruction (in which the leaves are also acquired in a Cartesian manner), in addition to the PROPELLER undersampled reconstruction, a deep learning-based Cartesian-like reconstruction network can also be used to perform Cartesian undersampled reconstruction (using a different reconstruction pipeline after using the network).

[0054] Figure 3 This is a more detailed example of the method. Figure 2Schematic diagram of method 180. Figure 3 As shown, in the case of leaf-level undersampling, accelerated periodic rotation overlapped parallel line with enhanced reconstruction acquisition is performed using k leaves of k-space data. Periodic rotation overlapped parallel line with enhanced reconstruction acquisition is a hybrid of Cartesian acquisition and non-Cartesian acquisition. During the periodic rotation overlapped parallel line with enhanced reconstruction sequence, the image is acquired by a magnetic resonance imaging scanner (e.g., Figure 1 A magnetic resonance imaging scanner 102 in the PROPELLER space acquires multiple k-space data leaves of a region of interest in a manner rotating about the center of k-space (e.g., rotating approximately 10 to 20 degrees between each leaf acquisition). The leaves in PROPELLER space (i.e., PROPELLER images) are represented by reference numeral 188. Each of the multiple k-space data leaves includes multiple parallel phase encoding lines (e.g., having a straight line shape) sampled in a phase encoding order using a fast spin echo method or a gradient echo method (i.e., each leaf is filled by an echo chain of a corresponding MR pulse sequence). Each k-space data leaf is acquired in a true Cartesian manner. Each of the multiple k-space leaves is undersampled. Specifically, fewer phase encoding lines are acquired for each of the multiple k-space data leaves. In some embodiments, the multiple k-space data leaves are undersampled in a uniform manner. In some embodiments, the multiple k-space data leaves are undersampled in a non-uniform manner. The multiple undersampled k-space data leaves have a skewed aspect ratio. Each leaf is acquired in a true Cartesian manner.

[0055] A plurality of undersampled k-space data blades (or undersampled leaf images) are extracted from the acquired periodically rotated overlapping parallel line enhanced reconstruction data, as shown in reference numeral 190. Each undersampled k-space data blade (or undersampled leaf image) represented by reference numeral 192 is separated and individually input into a leaf-level reconstruction unit 194. For each undersampled corresponding k-space data blade (or undersampled leaf image) 192, the leaf-level reconstruction unit 194 outputs a reconstructed and fully sampled leaf (or fully sampled leaf image) of the k-space data represented by reference numeral 196. Specifically, the leaf-level reconstruction unit 194 includes a deep learning-based Cartesian-like reconstruction network 198 to individually and separately reconstruct each undersampled k-space data blade (or undersampled leaf image) 192 to generate a corresponding fully sampled k-space data blade (or fully sampled leaf image) 196. Figure 3 As shown, this process is repeated individually and separately for each undersampled leaf (up to leaf k). A similar Cartesian reconstruction network 198 based on deep learning is trained to reconstruct a fully sampled k-space leaf image from the undersampled k-space leaf images. Figure 4The training of a deep learning based Cartesian-like reconstruction network 198 is described in . The deep learning based Cartesian-like reconstruction network 198 is configured to perform (and is also trained) with any number of k-space data leaves because the leaves are separated and reconstructed individually (rather than processed as multiple channels simultaneously). The deep learning based Cartesian-like reconstruction network 198 includes a deep learning based network based on an unfolding algorithm. It should be noted that since the Cartesian-like reconstruction is a leaf-level reconstruction (reconstructed at the leaf level, the leaves are also acquired in a Cartesian manner), in addition to PROPELLER undersampling reconstruction, the deep learning based Cartesian-like reconstruction network 198 can also be used to perform Cartesian undersampling reconstruction (using a different reconstruction pipeline after using the network 198). For example, the deep learning based Cartesian-like reconstruction network 198 can be used in conjunction with automatic calibration reconstruction for Cartesian-like imaging (ARC) undersampling, random undersampling, variable density undersampling, or other forms of undersampling.

[0056] The reconstructed fully sampled leaves 196 are reassembled (as represented by reference numeral 200) into a non-Cartesian grid, as represented by reference numeral 202. The reconstructed PROPELLER acquisition result (with fully sampled leaves) 202 (i.e., a reconstructed PROPELLER image) is affected by the remaining elements of the PROPELLER reconstruction chain as represented by reference numeral 204. Specifically, a complex image (reconstructed image) 206 of the region of interest is generated from multiple fully sampled (reconstructed) leaves using the PROPELLER reconstruction algorithm. For example, the PROPELLER reconstruction algorithm includes: phase correction for each leaf to ensure that its rotation point is precisely located at the center of k-space; correction for the object's overall in-plane rotation and in-plane translation; and correlation weighting to minimize data from leaves that contain motion or displacement errors. In certain embodiments, the complex image 206 is further processed before the image is written to a Digital Imaging and Communications in Medicine (DICOM) file.

[0057] As described above, the deep learning based leaf level reconstruction is independent of the number of leaves present in the PROPELLER data. This is because the reconstruction is performed in the acquisition space itself, once at a time. Each leaf is considered to be exactly Cartesian data and reconstructed accordingly. No leaves are concatenated at any point. This makes the disclosed technique independent of the number of leaves present, both from a training perspective and from an inference data perspective. In both the training phase and the inference phase, each leaf is treated separately as a reconstruction. Likewise, the back propagation loss for each leaf is calculated separately. Furthermore, the proposed technique is performed only in the image space.

[0058] Figure 4A method for training a similar Cartesian reconstruction network based on deep learning (e.g., Figure 3 A flowchart of a method 208 of a deep learning-based Cartesian reconstruction network 198). One or more steps of the method 208 may be performed by Figure 1 The processing circuitry of the magnetic resonance imaging system 100 or a remote computing system may be performed. For example, the processing circuitry may be part of the scanner control circuitry 104 and / or the system control circuitry 106 of the magnetic resonance imaging system 100.

[0059] The method 208 includes obtaining a fully sampled k-space leaf image acquired using a periodically rotated overlapping parallel line with enhanced reconstruction sequence (box 209). The method 208 also includes generating a partially sampled k-space leaf image from the fully sampled k-space leaf image (box 210). The method 208 also includes inputting an input-output (e.g., labeled) data pair 212 into a neural network 214 (box 216). The neural network 214 is a deep learning-based network based on an unfolding algorithm. The input-output data pair 212 is an undersampled k-space leaf image and a corresponding fully sampled k-space image acquired using a PROPELLER sequence. The fully sampled k-space image is used as a ground truth. The input-output data pair 212 has a skewed aspect ratio. The method 208 also includes training the neural network 214 on the input-output data pair 212 using supervised learning to generate a deep learning-based Cartesian-like reconstruction network or model 198 (box 218).

[0060] Figure 5 A schematic diagram of a Cartesian reconstruction network 198 based on deep learning and its utilization is illustrated. Figure 5 As shown, the deep learning-based Cartesian reconstruction network 198 is a deep learning-based network based on an unfolding algorithm. The deep learning-based network based on the unfolding algorithm is trained or made to learn to reconstruct fully sampled k-space leaves 220 (or fully sampled k-space leaf images) from undersampled k-space leaves 222 (or undersampled k-space leaf images). Each undersampled leaf is individually and separately reconstructed into a fully sampled leaf via the deep learning-based network based on the unfolding algorithm. Figure 5 As shown, the undersampled k-space data is transformed into undersampled image space 224 before being input into the expansion step 226. This reconstruction is similar to a Cartesian reconstruction. Figure 4The described PROPELLER leaf data trains a deep learning based Cartesian-like reconstruction network 198 to ensure effective anti-aliasing and high image quality reconstruction. As mentioned above, because the reconstruction is performed leaf by leaf (rather than processing all leaves simultaneously as multiple channels), the reconstruction described herein is not limited to PROPELLER reconstructions with a fixed number of leaves, but works for PROPELLER reconstructions with an arbitrary number of leaves.

[0061] The leaf-level deep learning reconstruction network 198 based on the unfolding algorithm is trained using the loss function. As shown in the following equation (Equation 1):

[0062]

[0063] I represents the benchmark truth image, Characterizes the estimated or reconstructed image, Re characterizes the real channel, and Im characterizes the imaginary channel. Moreover, α and β characterize the loss weights. These loss weights can be any floating point values. In the following disclosure, α is equal to 0.5, and β is equal to 1.0. The first part of the equation (to the left of the second + sign) characterizes the mean absolute error (MAE) loss on the real channel and the imaginary channel. This ensures that the complex reconstruction is accurate. Therefore, both the amplitude and phase are preserved. The second part of the equation (to the right of the second + sign) characterizes the structural similarity index (SSIM). The SSIM of the amplitude ensures that the structure in the reconstructed image is accurately preserved.

[0064] The unfolding algorithm used to perform the reconstruction is described in more detail by the following equation (Equation 2):

[0065]

[0066] Where x represents the fully sampled blade k-space, y represents the acquired undersampled blade k-space, A represents the MRI signal forming operator such that A = M·F, where M represents the undersampled mask used for blade acquisition and F represents the Fourier operation, λ represents the regularization weight, and z is the deep learning regularization term output. The deep learning regularization term is trained to perform the task of removing aliasing and blurring artifacts in the blade image due to the undersampling operation. In Equation 2, the first term is the data consistency term obtained from the basic MR image formation process of the blade. The second term is the data fidelity term obtained from the output of the deep learning regularization term. The update step of the unfolding algorithm is described by the following equation (Equation 3):

[0067]

[0068] Among them, X rec(k), Z(k), X acq (k) are the fully reconstructed image at the end of the expansion step, the deep learning predicted image of the expansion step, and the k-space information of the acquired original image at the kth position, respectively. The expansion is repeated N times. As used in the following results, N is equal to 10, and the deep learning regularization term is a deep learning network with residual channel attention architecture (# groups = 5, # blocks = 5). Moreover, λ is trainable.

[0069] There are several ways to undersample the blades. Figure 6 An example of a mask 228 of a fully sampled mask acquired by Periodically Rotated Overlapping Parallel Lines with Enhanced Reconstruction (PROPELLER) is illustrated. Figure 7 An example of non-uniform undersampling of a Periodically Rotated Overlapping Parallel Lines with Enhanced Reconstruction (PROPELLER) acquisition is illustrated. Figure 7 Both a non-uniformly undersampled mask 230 of a Periodically Rotated Overlapping Parallel Lines with Enhanced Reconstruction (PROPELLER) acquisition and blades 232 from non-uniformly unsampled PROPELLER data are depicted. Figure 8 An example of uniform undersampling of a Periodically Rotated Overlapping Parallel Lines with Enhanced Reconstruction (PROPELLER) acquisition is illustrated. Figure 8 Depicted are both a uniformly undersampled mask 234 of a Periodically Rotating Overlapping Parallel Lines with Enhanced Reconstruction (PROPELLER) acquisition and blades 236 from uniformly unsampled PROPELLER data.

[0070] Figures 9 to 13 The effect of performing deep learning-based leaf-level reconstruction is illustrated. Specifically, Figures 9 to 13 The effect of performing deep learning-based leaf-level reconstruction on different parts of the anatomy (e.g., pelvis, abdomen, cervical spine, etc.) is illustrated. Fig. 9 A periodically rotated overlapping parallel lines with enhanced reconstruction (PROPELLER) image obtained using zero-filled image reconstruction and deep learning-based leaf-level reconstruction of the abdomen of a subject is depicted. Fig. 9 The top row 238 of depicts zero-filled k-space leaf images obtained from a PROPELLER acquisition with an accelerated undersampling factor (or acceleration factor) of 2 and the corresponding zero-filled reconstructed images. Specifically, Fig. 9 The top row 238 of FIG. 2 depicts a zero-filled k-space leaf magnitude image 240 , a zero-filled k-space leaf real image 242 , and a zero-filled k-space leaf virtual image 244 . Fig. 9The top row 238 of FIG. 2 also depicts the corresponding zero-filled reconstructed magnitude image 246, zero-filled reconstructed real image 248, and zero-filled reconstructed virtual image 250. Zero-filled image reconstruction suffers from aliasing and blurring artifacts. Fig. 9 The bottom row 252 of depicts k-space leaf images after deep learning-based leaf-level reconstruction obtained from a PROPELLER acquisition with an accelerated undersampling factor (or acceleration factor) of 2 and the corresponding deep learning-based reconstructed images. Specifically, Fig. 9 The bottom row 252 of depicts a deep learning based reconstructed k-space leaf amplitude image 254 , a deep learning based reconstructed k-space leaf real image 256 , and a deep learning based reconstructed k-space leaf virtual image 258 . Fig. 9 The bottom row 252 of FIG. 2 also depicts a corresponding deep learning-based reconstructed magnitude image 260, a deep learning-based reconstructed real image 262, and a deep learning-based reconstructed virtual image 264. Fig. 9 As shown, the image quality of images 260, 262 and 264 reconstructed based on deep learning is better than that of images 246, 248 and 250 reconstructed by zero filling. Specifically, there are obvious blurring and undersampling artifacts (e.g., stripes) in the images 246, 248 and 250 reconstructed by zero filling. The images 260, 262 and 264 reconstructed based on deep learning restore clarity and do not have undersampling artifacts.

[0071] Fig.10 Periodic Rotation Overlapping Parallel Lines with Enhanced Reconstruction (PROPELLER) images of the abdomen of a subject are depicted. Images of the abdomen of the subject were acquired using a 1.5 Tesla (T) magnetic resonance imaging scanner using a body coil. Image 266 is a fully sampled axial T2 respiratory-triggered PROPELLER image of the abdomen of the subject. Image 268 is the corresponding zero-filled axial T2 respiratory-triggered PROPELLER image of the abdomen of the subject (the undersampling factor or acceleration factor is 2). The same undersampling mask was used for all leaves. The zero-filled image reconstruction suffers from aliasing and blurring artifacts. Image 270 is a fully sampled axial T2 respiratory-triggered PROPELLER image of the abdomen of the subject using Figure 2 Axial T2 respiration-triggered PROPELLER images reconstructed at the leaf level using deep learning-based reconstruction (PROPELLER data acquired with an undersampling factor or acceleration factor of 2) as described in method 180 of FIG. 272 ​​are the 4x difference between the ground truth (i.e., the fully sampled image 266) and the deep learning-based reconstruction image 270. Fig.10As shown, the image quality of the image 270 reconstructed based on deep learning is better than that of the image 268 reconstructed by zero filling. Specifically, the image 268 reconstructed by zero filling has obvious blur and undersampling artifacts (e.g., stripes). The image 270 reconstructed based on deep learning restores clarity and does not have undersampling artifacts. The image quality of the image 270 reconstructed based on deep learning is similar to that of the fully sampled image 266.

[0072] Fig.11 A periodic rotation overlapping parallel lines with enhanced reconstruction (PROPELLER) image of the subject's spine is depicted. Images of the subject's spine were acquired using a 1.5 Tesla (T) magnetic resonance imaging scanner using a body coil. Image 274 is a fully sampled sagittal T1 PROPELLER image of the subject's spine. Image 276 is a corresponding zero-filled sagittal T1 PROPELLER image of the subject's spine (the undersampling factor or acceleration factor is 2). The zero-filled image reconstruction suffers from aliasing and blurring artifacts. Image 278 is a fully sampled sagittal T1 PROPELLER image of the subject's spine using a body coil. Figure 2 Sagittal T1 PROPELLER image reconstructed based on deep learning and using leaf-level reconstruction as described in method 180 of FIG. 1 (PROPELLER data acquired with an undersampling factor or acceleration factor of 2). Fig.11 As shown, the image quality of the image 278 reconstructed based on deep learning is better than that of the image 276 reconstructed by zero filling. Specifically, the image 276 reconstructed by zero filling has obvious blur and undersampling artifacts (e.g., stripes). The image 278 reconstructed based on deep learning restores clarity and does not have undersampling artifacts. The image quality of the image 278 reconstructed based on deep learning is similar to that of the fully sampled image 274.

[0073] Fig.12 A periodically rotated overlapping parallel lines with enhanced reconstruction (PROPELLER) image of the subject's pelvis is depicted. Images of the subject's pelvis were acquired using a 1.5 Tesla (T) magnetic resonance imaging scanner using a body coil. Image 280 is a fully sampled sagittal T2PROPELLER image of the subject's pelvis. Image 282 is a corresponding zero-filled sagittal T2PROPELLER image of the subject's pelvis (the undersampling factor or acceleration factor is 2). The zero-filled image reconstruction suffers from aliasing and blurring artifacts. Image 284 is a fully sampled sagittal T2PROPELLER image of the subject's pelvis using a body coil. Figure 2 Sagittal T2 PROPELLER images reconstructed using deep learning based methods (PROPELLER data acquired with an undersampling factor or acceleration factor of 2) using leaf-level reconstruction as described in method 180 of FIG. Fig.12As shown, the image quality of the image 284 reconstructed based on deep learning is better than that of the image 282 reconstructed by zero filling. Specifically, the image 282 reconstructed by zero filling has obvious blur and undersampling artifacts (e.g., stripes). The image 284 reconstructed based on deep learning restores clarity and does not have undersampling artifacts. The image quality of the image 284 reconstructed based on deep learning is similar to that of the fully sampled image 280.

[0074] Fig.13 A Periodic Rotation Overlapping Parallel Lines with Enhanced Reconstruction (PROPELLER) image of a subject's pelvis is depicted (e.g., depicting the compatibility of artificial intelligence-based denoising with deep learning-based blade-level reconstruction). Images of the subject's pelvis were acquired using a 1.5 Tesla (T) magnetic resonance imaging scanner using a body coil. Image 286 is a fully sampled sagittal T2 PROPELLER image of the subject's pelvis. Image 288 is a fully sampled sagittal T2 PROPELLER image of the subject's pelvis using a body coil. Figure 2 Fig. 2 is a deep learning-based reconstructed sagittal T2 PROPELLER image with leaf-level reconstruction as described in method 180 of Fig. 2 (PROPELLER data acquired with an undersampling factor or acceleration factor of 2). Image 290 is the fully sampled image 286 after using artificial intelligence-based denoising. Image 292 is the deep learning-based reconstructed image 288 after using artificial intelligence-based denoising. The artificial-based denoising model or network is trained on the PROPELLER images. Since the artificial-based denoising model or network is able to denoise the deep learning-based reconstructed image 288, this indicates that the deep learning-based reconstruction of the accelerated data does not change the noise characteristics as learned by the PROPELLER denoising model or network.

[0075] The technical effects of the subject matter disclosed in the present invention include providing a system and method for shortening the scan time of Periodic Rotation Overlapping Parallel Lines with Enhanced Reconstruction (PROPELLER) imaging. Specifically, the technical effects of the subject matter disclosed in the present invention include providing a technique utilizing a Cartesian-like reconstruction method at the leaf level for performing accelerated Periodic Rotation Overlapping Parallel Lines with Enhanced Reconstruction (PROPELLER) acquisition while maintaining all remaining elements of the PROPELLER image reconstruction undisturbed in the reconstruction chain. The technical effects of the subject matter disclosed in the present invention also include enabling faster magnetic resonance imaging scans utilizing periodic rotation overlapping parallel lines with enhanced reconstruction scanning. The technical effects of the subject matter disclosed in the present invention also include increasing the system throughput of a magnetic resonance imaging system using the magnetic resonance imaging system. The subject matter disclosed in the present invention may also be utilized with both Cartesian undersampling reconstruction and PROPELLER undersampling reconstruction.

[0076] Reference is made to the technology presented and claimed herein and applied to physical and specific examples of a practical nature that clearly improves upon the present state of the art and, therefore, is not abstract, intangible, or purely theoretical. In addition, if any claim appended to the end of this specification contains one or more elements designated as "means for [performing] the function of ..." or "steps for [performing] the function of ...," it is intended that such elements be interpreted under 35 U.S.C. § 112(f). However, for any claim containing elements designated in any other manner, it is not intended that such elements be interpreted under 35 U.S.C. § 112(f).

[0077] This written description uses examples to disclose the subject matter, including the best mode, and also to enable those skilled in the art to practice the subject matter, including making and using any devices or systems and performing any included methods. The patentable scope of the subject matter is defined by the claims, and may include other examples that occur to those skilled in the art. Such other examples are intended to be within the scope of the claims if they have structural elements that do not differ from the literal language of the claims, or if they include equivalent structural elements with insignificant differences from the literal language of the claims.

Claims

1. A computer-implemented method for reducing scan time for periodically rotated overlapping parallel lines with enhanced reconstruction (PROPELLER) imaging, the computer-implemented method comprising: acquiring, via a processor, a plurality of k-space data leaves of a region of interest in a rotational manner about a center of k-space via a magnetic resonance imaging (MRI) scanner in an accelerated manner during a PROPELLER sequence, wherein each of the plurality of k-space data leaves comprises a plurality of parallel phase encoding lines sampled in a phase encoding order, and wherein each of the plurality of k-space data leaves is undersampled; reconstructing, via the processor, each of the plurality of k-space data leaves individually and separately using a deep learning based Cartesian-like reconstruction network to generate a plurality of fully sampled leaves; and A complex image is generated from the plurality of fully sampled leaves using a PROPELLER reconstruction algorithm via the processor.

2. The computer-implemented method of claim 1 , wherein: The deep learning based Cartesian-like reconstruction network is configured to perform with an arbitrary number of k-space data leaves.

3. The computer-implemented method of claim 1 , wherein: The plurality of k-space data blades are acquired from a single receiver coil.

4. The computer-implemented method of claim 1 , wherein: The plurality of k-space data blades are acquired from a plurality of receiver coils.

5. The computer-implemented method of claim 1 , wherein: The deep learning based Cartesian-like reconstruction network includes a deep learning based network based on an unfolding algorithm.

6. The computer-implemented method of claim 5, further comprising training, via the processor, the deep learning based Cartesian-like reconstruction network on input-output data pairs using supervised learning, wherein The input-output data pair includes an undersampled k-space leaf image and a corresponding fully sampled k-space image acquired using the PROPELLER sequence, and the undersampled k-space leaf image is generated from the corresponding fully sampled k-space image.

7. The computer-implemented method of claim 1 , wherein: The plurality of k-space data blades include a skew aspect ratio.

8. A system for reducing the scan time of periodically rotated overlapping parallel lines with enhanced reconstruction (PROPELLER) imaging, the system comprising: a memory encoding a processor-executable routine; and a processor configured to access the memory and to execute the processor-executable routine, wherein the processor-executable routine, when executed by the processor, causes the processor to: acquiring a plurality of k-space data leaves of a region of interest in a rotational manner about a center of k-space via a magnetic resonance imaging (MRI) scanner in an accelerated manner during a PROPELLER sequence, wherein each of the plurality of k-space data leaves comprises a plurality of parallel phase encoding lines sampled in a phase encoding order, and wherein each of the plurality of k-space data leaves is undersampled; utilizing a deep learning based Cartesian-like reconstruction network to individually and separately reconstruct each of the plurality of k-space data leaves to generate a plurality of fully sampled leaves; and A complex image is generated from the plurality of fully sampled leaves using the PROPELLER reconstruction algorithm.

9. The system according to claim 8, wherein: The deep learning based Cartesian-like reconstruction network is configured to perform with an arbitrary number of k-space data leaves.

10. The system according to claim 8, wherein: The plurality of k-space data blades are acquired from a single receiver coil.

11. The system according to claim 8, wherein: The plurality of k-space data blades are acquired from a plurality of receiver coils.

12. The system according to claim 8, wherein: The deep learning based Cartesian-like reconstruction network includes a deep learning based network based on an unfolding algorithm.

13. The system according to claim 12, wherein: The processor executable routine, when executed by the processor, also causes the processor to train the deep learning based Cartesian-like reconstruction network using supervised learning on input-output data pairs, wherein the input-output data pairs include undersampled k-space leaf images and corresponding fully sampled k-space images acquired using the PROPELLER sequence, and the undersampled k-space leaf images are generated from the corresponding fully sampled k-space images.

14. The system according to claim 8, wherein: The plurality of k-space data blades include a skew aspect ratio.

15. A non-transitory computer readable medium comprising processor executable code which, when executed by a processor, causes the processor to: A plurality of k-space data leaves of a region of interest are acquired in an accelerated manner in a manner rotating around the center of k-space via a magnetic resonance imaging (MRI) scanner during a periodic rotation overlapping parallel lines with enhanced reconstruction (PROPELLER) sequence, wherein: Each blade of the plurality of k-space data blades comprises a plurality of parallel phase encoding lines sampled in a phase encoding order, and wherein each blade of the plurality of k-space data blades is undersampled; utilizing a deep learning based Cartesian-like reconstruction network to individually and separately reconstruct each of the plurality of k-space data leaves to generate a plurality of fully sampled leaves; and A complex image is generated from the plurality of fully sampled leaves using the PROPELLER reconstruction algorithm.

16. The non-transitory computer readable medium of claim 15, wherein: The deep learning based Cartesian-like reconstruction network is configured to perform with an arbitrary number of k-space data leaves.

17. The non-transitory computer readable medium of claim 15, wherein: The plurality of k-space data blades are acquired from a single receiver coil.

18. The non-transitory computer readable medium of claim 15, wherein: The plurality of k-space data blades are acquired from a plurality of receiver coils.

19. The non-transitory computer readable medium of claim 15, wherein: The deep learning based Cartesian-like reconstruction network includes a deep learning based network based on an unfolding algorithm.

20. The non-transitory computer readable medium of claim 15, wherein: The processor executable code, when executed by the processor, also causes the processor to train the deep learning-based Cartesian-like reconstruction network using supervised learning on input-output data pairs, wherein the input-output data pairs include undersampled k-space leaf images and corresponding fully sampled k-space images acquired using the PROPELLER sequence, and the undersampled k-space leaf images are generated from the corresponding fully sampled k-space images.

Images