System and method for region layer of calibration images

Abstract

The invention relates to a system and a method for regional stratification of a calibration image. A method and system includes obtaining calibration scan data or low resolution images of a subject acquired with an MR scanner of a magnetic resonance (MR) imaging system. The method and the system also include inputting the calibration data or the low resolution image into a trained deep learning based multi-mask segmentation network. The method and the system also include outputting marker mask images for different anatomical stations, wherein a respective mask of a respective marker mask image highlights an anatomical marker of interest in each respective anatomical station of the different anatomical stations. The method and the system even further include determining a range of a respective anatomical landmark of interest in each of the different anatomical stations for a respective locator scan for each respective anatomical station based at least on the respective marker mask image for each respective anatomical station.

Description

Background Art

[0001] The subject matter disclosed herein relates to medical imaging, and more particularly to a system and method for regional stratification of a calibration image.

[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 images or otherwise represent observed internal features of a 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 the well-known reconstruction techniques.

[0005] Whole body scanning in MRI (e.g., of the head, spine, abdomen, pelvis, and heart) enables MRI of the entire body to be performed in a single imaging session. However, whole body scanning in MRI is not straightforward due to the need for multiple stations (e.g., anatomical stations) or anatomical structure layering on multiple locator images. Similarly, spinal scanning using MRI is cumbersome due to the involvement of multiple stations and the need to find the extent of the cervical, lumbar, and thoracic regions. Therefore, technicians must acquire multiple locators and then work across multiple groups of locators to determine the boundaries across stations and scan to obtain higher resolution images. This may also involve additional processing regarding image cropping, image stitching (e.g., digitally stitching image segments together), and other processing 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 performing regional stratification on a calibration image or a low-resolution image is provided. The computer-implemented method includes obtaining calibration scan data or a low-resolution image of a subject acquired by an MR scanner of a magnetic resonance (MR) imaging system at a processor. The computer-implemented method also includes inputting the calibration data or the low-resolution image into a trained multi-mask segmentation network based on deep learning via the processor. The computer-implemented method also includes outputting a labeled mask image for different anatomical stations via the processor, wherein the corresponding mask of the corresponding labeled mask image highlights the anatomical landmarks of interest in each corresponding anatomical station in the different anatomical stations. The computer-implemented method even includes determining, via the processor, the range of the corresponding anatomical landmarks of interest in each corresponding anatomical station in the different anatomical stations for the corresponding locator scan for each corresponding anatomical station.

[0008] In another embodiment, a system for performing regional stratification on a calibration image or a low-resolution image is provided. 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 obtaining calibration scan data or a low-resolution image of a subject acquired by an MR scanner of a magnetic resonance (MR) imaging system. The action also includes inputting the calibration data or the low-resolution image into a trained multi-mask segmentation network based on deep learning. The action also includes outputting a labeled mask image for different anatomical stations, wherein the corresponding mask of the corresponding labeled mask image highlights the anatomical landmarks of interest in each corresponding anatomical station in the different anatomical stations. The action even includes determining the range of the corresponding anatomical landmarks of interest in each corresponding anatomical station in the different anatomical stations for the corresponding locator scan for each corresponding anatomical station, based at least on the corresponding labeled mask image for each corresponding anatomical station.

[0009] In yet another embodiment, a non-transitory computer-readable medium includes processor executable code that causes the processor to perform an action when executed by a processor. The action includes obtaining calibration scan data or a low-resolution image of a subject acquired using an MR scanner of a magnetic resonance (MR) imaging system. The action also includes inputting the calibration data or the low-resolution image into a trained multi-mask segmentation network based on deep learning. The action also includes outputting a labeled mask image for different anatomical stations, wherein the corresponding mask of the corresponding labeled mask image highlights the anatomical landmarks of interest in each corresponding anatomical station in the different anatomical stations. The action even includes determining the range of the corresponding anatomical landmarks of interest in each corresponding anatomical station in the different anatomical stations for the corresponding locator scan for each corresponding anatomical station, based at least on the corresponding labeled mask image for each corresponding anatomical station. 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 A schematic diagram illustrating a magnetic resonance imaging (MRI) system suitable for use with the disclosed techniques is illustrated;

[0012] Figure 2 is a schematic diagram illustrating an example of a deep learning based multi-mask segmentation network according to aspects of the present disclosure;

[0013] Figure 3 is an example of an MR image including the entire spine according to aspects of the present disclosure;

[0014] Figure 4 is a schematic diagram illustrating how to create ground truth values ​​for labeling multiple stations for a spinal scan according to aspects of the present disclosure;

[0015] Figure 5 The present disclosure illustrates various aspects of the present disclosure for utilizing Figure 1 A flowchart of a method for performing a scan on a patient by an MRI system;

[0016] Figure 6 The present disclosure illustrates various aspects of the present disclosure for utilizing Figure 1 A flowchart of a method for performing a spinal scan on a patient using an MRI system;

[0017] Figure 7 is an example of a coil calibration image of a subject's spine according to aspects of the present disclosure;

[0018] Figure 8 is an example of an image with masks for the thoracic region and the lumbar region of the spine according to aspects of the present disclosure;

[0019] Fig. 9 According to various aspects of the present disclosure, there is a method for applying Figure 7 Coil calibration image of the spine Figure 8 An example of an image of a mask of a thoracic region and a lumbar region of a spine in an image of FIG.

[0020] Fig.10 is an example of a coil calibration image and a corresponding ground truth image marked with a mask identifying a thoracic region of the spine according to aspects of the present disclosure;

[0021] Fig.11 is another example of a coil calibration image and a corresponding ground truth image marked with a mask identifying a thoracic region of the spine according to aspects of the present disclosure;

[0022] Fig.12 is yet another example of a coil calibration image and a corresponding ground truth image marked with a mask identifying a thoracic region of the spine according to aspects of the present disclosure; and

[0023] Fig.13 are examples of coil calibration images of a first portion and a second portion of a subject's body according to aspects of the present disclosure. 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 additional elements may be present 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 embodiments disclosed herein.

[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 provided in such medical contexts are only for the convenience of 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 acquired 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 may therefore include one or more of the following: deep neural networks, fully interconnected networks, convolutional neural networks (CNNs), unfolded neural networks, perceptrons, codecs, recurrent networks, wavelet filter banks, u-nets, generative adversarial networks (GANs), dense neural networks 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 can be characterized as using one or more algorithms to extract or model a class of high-level abstractions of a class of data of interest. This can 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., the hierarchical or cascaded structure of the layers) as the target of the process or algorithm of a given layer. In the context of image processing or reconstruction, this can 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 can be viewed as a "stage" of the process. Each stage of the process can be performed by a separate neural network or by different parts of a larger neural network.

[0029] The present disclosure provides a system and method for performing regional stratification on a calibration image. Specifically, the system and the method obtain coil calibration scan data (e.g., a coil calibration image) of a subject acquired during a pre-scan of an MR scanner using a magnetic resonance (MR) imaging system. The method and the system also include inputting the coil calibration data into a trained multi-mask segmentation network based on deep learning. The method and the system also include outputting a labeled mask image for different anatomical stations (or regions marked with boundaries), wherein a corresponding mask of the corresponding labeled mask image highlights an anatomical landmark of interest in each corresponding anatomical station in the different anatomical stations. In certain embodiments, more than one anatomical landmark of interest may be marked in a particular anatomical station. In addition, the mask may be a point mark or a multi-point station delineation region. The method and the system even include determining the range of a corresponding anatomical landmark of interest in each corresponding anatomical station in the different anatomical stations for a corresponding locator scan for each corresponding anatomical station, based at least on a corresponding labeled mask image for each corresponding anatomical station.

[0030] In some embodiments, the systems and methods can be used for low-resolution images (as opposed to calibration images). Low-resolution images are non-diagnostic images that have a lower resolution than diagnostic images. In some embodiments, the systems and methods can be used for calibration data other than coil calibration data (e.g., ASSET / PURE calibration). In some embodiments, the calibration data can be obtained from a pre-scan calibration or a mission calibration. The calibration data can be raw or processed.

[0031] The multi-mask segmentation network based on deep learning is trained to detect and mark the corresponding anatomical landmarks of interest in each corresponding anatomical station in different anatomical stations. In some embodiments, the system and method also include obtaining the requirements for the scan of the subject (e.g., scanning protocol, region of interest to be scanned, scanning purpose, etc.). In some embodiments, based on the corresponding marking mask image for each corresponding anatomical station and the requirements for the scan of the subject, for the corresponding locator scan for each corresponding anatomical station, the range of the corresponding anatomical landmarks of interest in each corresponding anatomical station in different anatomical stations is determined. In some embodiments, coil calibration scan data is obtained from a single radio frequency coil. In some embodiments, coil calibration data is obtained from multiple radio frequency coils. The calibration data can be original or processed. The radio frequency coil may include a single channel or multi-channel radio frequency body or surface or a transmit / receive or receive-only coil. In some embodiments, for each of the multiple radio frequency coils used for the corresponding locator scan for each corresponding anatomical station, the range of the corresponding anatomical landmarks of interest in each corresponding anatomical station in different anatomical stations is determined. In some embodiments, different anatomical stations are used for spinal scanning using an MR scanner. In some embodiments, different anatomical stations are used for whole body scans using an MR scanner. In some embodiments, different anatomical stations are used for MRI-based attenuation correction (MRAC) scans using an MR scanner and a positron emission tomography (PET) scanner, wherein the MR imaging system is part of a PET / MR imaging system.

[0032] The disclosed embodiments provide automatic patient-specific range estimation for the correct locator or high-resolution scan during the pre-scan stage. For example, the disclosed embodiments enable the technology to automatically determine the range of the spine station (e.g., cervical, thoracic, lumbar) from the calibration image as part of the pre-scan, and the correct station locator is scanned for further use by the technician when planning the spine examination. These techniques can be extended to determine other anatomical boundaries in the context of whole-body imaging or PET / MR MRAC boundaries. Using these techniques, all presets necessary for regional stratification can be obtained from the pre-scan data and geometry, where the geometric and patient-specific locators are available to the technician at any time. This ensures that the correct station (e.g., only cervical or only thoracic or only lumbar) or station (e.g., cervical-thoracic, thoracic-lumbar or entire spine) locator is automatically available to the user for spine planning. This is especially true for the thoracic station, which is usually separated across two locators in current practice. The method allows a single locator to be automatically generated, eliminates the need for splicing, and facilitates easy setup for planning inspections. Embodiments disclosed herein may also be used to automatically establish patient anatomy specific head / foot coverage presets for multi-station whole body imaging examinations (e.g., PET / MR MRAC). Embodiments disclosed herein further reduce the time to establish subsequent scans (e.g., localizer scans and / or higher resolution scans (e.g., diagnostic scans)).

[0033] Although the techniques described below are used with an MRI system, in certain embodiments, the techniques can be used with a PET / MR imaging system. For example, the techniques described below can be used for PET / MR MRAC scanning.

[0034] 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 MRI system 100 is generally configured to perform MR imaging.

[0035] Additionally, the system 100 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 MRI 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.

[0036] The scanner 102 includes a series of associated coils for generating a controlled magnetic field, which is used to excite the gyromagnetic material 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 the generation of a controlled gradient magnetic field during an examination sequence for position encoding of certain gyromagnetic nuclei in 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 in the patient. In addition to the coils that may be located locally in 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, receive coil 138 is placed near or on top of patient 126 to receive the weak RF signals (relative to the transmit pulses generated by the scanner coils) generated by certain gyromagnetic nuclei in the patient's body as patient 126 returns to his or her relaxed state.

[0037] The various coils of the system 100 are controlled by external circuitry to generate the required fields and pulses and to read the emission 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 primary magnetic field Bo. 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.

[0038] 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.

[0039] 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.

[0040] 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.

[0041] The control and analysis circuitry 160 is also used to receive magnetic resonance signals and perform subsequent processing before sending 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.

[0042] 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 certain embodiments, control and analysis circuit 160, although shown as a single unit, may include one or more hardware devices. System control circuit 106 includes interface circuit 166 that receives data from scanner control circuit system 104 and sends 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 coupled to memory circuit 170 to store programming code for operating MRI 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 acquisition data as described below when executed by a processor. In some embodiments, the memory circuit 170 may store one or more neural networks (e.g., a deep learning-based multi-mask segmentation network for spine scans, a deep learning-based multi-mask segmentation network for whole body scans, a deep learning-based multi-mask segmentation network for PET / MR MRAC scans, a deep learning-based multi-mask segmentation network for regions of interest, etc.) for performing regional stratification on the calibration image, as described below. In some embodiments, image reconstruction may occur on a separate computing device having processing circuitry and memory circuitry.

[0043] The programming code may enable regional stratification to be performed on the calibration image. Specifically, the programming code may also enable coil calibration scan data (e.g., coil calibration image) of a subject acquired during a pre-scan of an MR scanner utilizing a magnetic resonance (MR) imaging system to be obtained. The programming code may also enable the coil calibration data to be input into a trained multi-mask segmentation network based on deep learning. The calibration data may be raw or processed. The programming code may also enable the output of a marker mask image for different anatomical stations, wherein a corresponding mask of a corresponding marker mask image highlights an anatomical landmark of interest in each corresponding anatomical station in the different anatomical stations. The mask may be a point marker or a multi-point station delineation region. The programming code may also enable the range of a corresponding anatomical landmark of interest in each corresponding anatomical station in the different anatomical stations to be determined for a corresponding locator scan for each corresponding anatomical station, based at least on a corresponding marker mask image for each corresponding anatomical station.

[0044] The multi-mask segmentation network based on deep learning is trained to detect and mark the corresponding anatomical landmarks of interest in each corresponding anatomical station in different anatomical stations. In some embodiments, the programming code can also enable the requirements for the scanning of the subject (e.g., scanning protocol, region of interest to be scanned, scanning purpose, etc.). In some embodiments, based on the corresponding marking mask image for each corresponding anatomical station and the requirements for the scanning of the subject, for the corresponding locator scan for each corresponding anatomical station, the range of the corresponding anatomical landmarks of interest in each corresponding anatomical station in different anatomical stations is determined. In some embodiments, coil calibration scan data is obtained from a single radio frequency coil. In some embodiments, coil calibration data is obtained from multiple radio frequency coils. The calibration data can be original or processed. The radio frequency coil may include a single channel or multi-channel radio frequency body or surface or a transmit / receive or receive-only coil. In some embodiments, for each of the multiple radio frequency coils used for the corresponding locator scan for each corresponding anatomical station, the range of the corresponding anatomical landmarks of interest in each corresponding anatomical station in different anatomical stations is determined. In some embodiments, different anatomical stations are used for spinal scanning using an MR scanner. In some embodiments, different anatomical stations are used for whole body scans using an MR scanner. In some embodiments, different anatomical stations are used for PET / MR MRAC scans using an MR scanner and a positron emission tomography (PET) scanner, wherein the MR imaging system is part of the PET / MR imaging system.

[0045] 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 generating 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.

[0046] Figure 2is a schematic diagram illustrating an example of a multi-mask segmentation network based on deep learning. Each multi-mask segmentation network based on deep learning is trained via supervised learning of input-output data pairs. Each multi-mask segmentation network based on deep learning is configured to receive an input of coil calibration data (e.g., a coil calibration image). In some embodiments, the coil calibration data is obtained from a single coil (e.g., a surface RF coil or a body RF coil) during a pre-scan. In some embodiments, the coil calibration data is obtained from multiple coils (e.g., a surface RF coil and a body RF coil) during a pre-scan. The calibration data may be raw or processed. The radio frequency coil may include a single-channel or multi-channel radio frequency body or surface or a transmit / receive or receive-only coil. Each multi-mask segmentation network based on deep learning is also configured to output a labeled mask image for different anatomical stations, wherein the corresponding mask of the corresponding labeled mask image highlights the anatomical landmarks of interest in each corresponding anatomical station in the different anatomical stations. Specifically, each multi-mask segmentation network based on deep learning is trained to detect and mark the corresponding anatomical landmarks of interest in each corresponding anatomical station in the different anatomical stations. The mask can be a point marker or a multi-point station delineated area.

[0047] like Figure 2 As depicted, a deep learning-based multi-mask segmentation network 180 of the spine is trained for use with spinal scans. Specifically, the deep learning-based multi-mask segmentation network 180 of the spine is configured to detect and mark the cervical portion of the spine, the thoracic portion of the spine, and the lumbar portion of the spine for the cervical spine station, the thoracic spine station, and the lumbar spine station, respectively. In certain embodiments, the deep learning-based multi-mask segmentation network 180 of the spine is configured to detect and mark both the cervical and thoracic portions of the spine for the cervical-thoracic spine station, detect and mark both the thoracic portion and the lumbar portion of the spine for the thoracic-lumbar spine station, and / or detect and mark the entirety of the spine for the spine station. Figure 3 is an example of an MR image 183 that includes the entire spine. The MR image 183 is along the sagittal plane. The MR image 183 has been generated by stitching (e.g., digitally stitching or merging image segments together) different images from a scanning session of a subject. The MR image 183 can be used as a ground truth in training a deep learning-based spine multi-mask segmentation network 180.

[0048] like Figure 2As depicted, a full body multi-mask segmentation network 182 based on deep learning is trained for use with full body scans. Specifically, a full body multi-mask segmentation network 182 based on deep learning is configured to detect and mark different regions of the body along the head-to-foot direction. The number of anatomical stations and the anatomical landmarks of interest in each anatomical station may vary depending on the scanning protocol. In one example, anatomical stations may be used for the head, chest, abdomen, and pelvic regions. In certain embodiments, a full body multi-mask segmentation network 182 based on deep learning may be configured to detect and mark different regions of the body along other directions (e.g., right-left direction or front-back direction).

[0049] like Figure 2 As depicted, a deep learning based PET / MR MRAC multi-mask segmentation network 184 is trained for use with a PET / MR MRAC scan. Specifically, the deep learning based PET / MR MRAC multi-mask segmentation network 184 is configured to detect and label different regions of the body along a head-front direction. The number of anatomical stations and the anatomical landmarks of interest in each anatomical station may vary depending on the scan protocol. In one example, anatomical stations may be used for the head, chest, abdomen, and pelvic regions.

[0050] like Figure 2 As depicted, a deep learning-based region of interest (ROI) multi-mask segmentation network 186 is trained for use with scans of a region of interest. Specifically, the deep learning-based region of interest multi-mask segmentation network 186 is configured to detect and mark different anatomical landmarks in the region of interest. For example, the deep learning-based region of interest multi-mask segmentation network 186 can be used for different parts of the lungs. In another example, the deep learning-based region of interest multi-mask segmentation network 186 can be used for different parts of the shoulder or another region of interest.

[0051] In certain embodiments, an MRI system (e.g., Figure 1 The memory circuit system of the MRI system 100 in the remote computing device or the remote computing device may include only one of the deep learning-based spine multi-mask segmentation network 180, the deep learning-based whole body multi-mask segmentation network 182, the deep learning-based PET / MR MRAC multi-mask segmentation network 184, and the deep learning-based region of interest multi-mask segmentation network 186 for use by the processing circuit system. In some embodiments, the MRI system (e.g., Figure 1The memory circuit system of the MRI system 100 in the medical device or a remote computing device may include two or more of a deep learning-based spine multi-mask segmentation network 180, a deep learning-based whole body multi-mask segmentation network 182, a deep learning-based PET / MRMRAC multi-mask segmentation network 184, and a deep learning-based region of interest multi-mask segmentation network 186 for utilization by the processing circuit system.

[0052] Figure 4 FIG. 1 is a schematic diagram illustrating how to create a ground truth for labeling multiple stations for a spinal scan. Figure 4 As depicted, a vertebra segmentation mask image 188 is generated. The vertebra segmentation mask image 188 is generated by utilizing a deep learning-based model that is trained to segment vertebrae on an MR image (along a sagittal plane) that includes stitched data of the entire spine. As depicted in the vertebra segmentation mask image 188, all vertebrae of the spine are marked (e.g., highlighted) with a segmentation mask.

[0053] In addition, if Figure 4 As depicted, a disc segmentation mask image 190 is generated. The disc segmentation mask image 190 is generated by utilizing a deep learning-based model that is trained to segment the disc on an MR image (along the sagittal plane) including stitched data of the entire spine. As depicted in the disc segmentation mask image 190, all discs of the spine are marked (e.g., herniated) with a segmentation mask.

[0054] In addition, if Figure 4 As depicted, the spine segmentation mask image 188 is combined with the disc segmentation mask image 190 (as indicated by arrow 192) to generate a combined spine segmentation mask image 194. As depicted in the combined spine segmentation mask image 194, all vertebrae and discs are marked (e.g., highlighted) with the segmentation mask.

[0055] Furthermore, if Figure 4 As depicted, a cervical spine segmentation mask image 196 is generated. The cervical spine segmentation mask image 196 is generated by utilizing a deep learning-based model that is trained to segment cervical vertebrae on an MR image (along a sagittal plane) that includes stitched data of the entire spine. As depicted in the cervical spine segmentation mask image 196, all cervical vertebrae of the spine are marked (e.g., highlighted) with a segmentation mask.

[0056] Furthermore, if Figure 4As depicted, a lumbar segmentation mask image 198 is generated. The lumbar segmentation mask image 198 is generated by utilizing a deep learning-based model that is trained to segment lumbar vertebrae on an MR image (along a sagittal plane) that includes stitched data of the entire spine. As depicted in the lumbar segmentation mask image 198, all of the lumbar vertebrae of the spine are marked (e.g., highlighted) with a segmentation mask.

[0057] like Figure 4 As depicted, the combined spinal segmentation mask image 194 is separated using the cervical segmentation mask image 196 and the lumbar segmentation mask image 198, as indicated by arrow 200, to generate both a cervical segmentation mask image 202 and a lumbar segmentation mask image 204. In the cervical segmentation mask image 202, both the cervical vertebrae and the intervertebral discs between the cervical vertebrae are marked (e.g., highlighted) with a segmentation mask. In the lumbar segmentation mask image 204, both the lumbar vertebrae and the intervertebral discs between the lumbar vertebrae are marked (e.g., highlighted) with a segmentation mask. The remaining portion enables the generation of a thoracic segmentation mask image 206. In the thoracic segmentation mask image 208, both the thoracic vertebrae and the intervertebral discs between the thoracic vertebrae are marked (e.g., highlighted) with a segmentation mask.

[0058] The cervical segmentation mask image 202, the lumbar segmentation mask image 204, and the thoracic segmentation mask image 208 can be used as reference truth values ​​in training a deep learning-based spine multi-mask segmentation network to detect and label the cervical portion of the spine, the thoracic portion of the spine, and the lumbar portion of the spine for the cervical spine station, the thoracic spine station, and the lumbar spine station, respectively. In certain embodiments, the cervical segmentation mask image 202, the lumbar segmentation mask image 204, and the thoracic segmentation mask image 208 can be used as reference truth values ​​in training a deep learning-based spine multi-mask segmentation network to detect and label both the cervical and thoracic portions of the spine for the cervical-thoracic spine station, to detect and label both the thoracic and lumbar portions of the spine for the thoracic-lumbar spine station, and / or to detect and label the entirety of the spine for the spine station.

[0059] Figure 5 Illustrated for use Figure 1 Flow chart of method 210 for performing a scan on a patient using MRI system 100 in FIG. One or more steps of method 210 may be performed by Figure 1 One or more of the steps of method 210 may be performed simultaneously or sequentially. Figure 5The depicted sequence is performed in a different order. In some embodiments, method 210 can be used for low-resolution images (as opposed to calibration images). Low-resolution images are non-diagnostic images with lower resolution than diagnostic images. In some embodiments, method 210 can be used for calibration data other than coil calibration data (e.g., ASSET / PURE calibration). In some embodiments, the calibration data can be obtained from a pre-scan calibration or a mission calibration. The calibration data can be raw or processed.

[0060] Method 210 includes initiating the use of Figure 1 A pre-scan of a subject is performed using an MR scanner (eg, the MR scanner 102) in an MRI scanner (block 212). The pre-scan includes performing a coil calibration scan using one or more coils (eg, an RF surface coil and / or an RF body coil).

[0061] The method 210 also includes obtaining coil calibration scan data (e.g., coil calibration images) of the subject acquired during a pre-scan using an MR scanner (box 214). The coil calibration data is large field of view coil calibration data. The method 210 also includes inputting the coil calibration data into a trained deep learning-based multi-mask segmentation network 216 (box 218). In some embodiments, the deep learning-based multi-mask segmentation network is trained for spine scanning. In some embodiments, the deep learning-based multi-mask segmentation network is trained for whole body scanning. In some embodiments, the deep learning-based multi-mask segmentation network 216 is trained for PET / MR MRAC scanning. In some embodiments, the deep learning-based multi-mask segmentation network 216 is trained for non-PET / MR whole body planning workflows. In some embodiments, the deep learning-based segmentation network 216 is trained for scanning of different and specific regions of interest (e.g., lungs, shoulders, etc.). The deep learning-based multi-mask segmentation network is trained to detect and mark corresponding anatomical landmarks of interest in each corresponding anatomical station in different anatomical stations.

[0062] The method 210 also includes outputting a labeled mask image for different anatomical stations, wherein the corresponding masks of the corresponding labeled mask images highlight the anatomical landmarks of interest in each of the different anatomical stations (box 220). In some embodiments, more than one anatomical landmark of interest may be marked in a particular anatomical station. The method 210 also includes obtaining a requirement or information for a scan (e.g., a subsequent scan, such as a localizer scan or a diagnostic scan) of the subject (box 222). The requirement or information may include a scan protocol, a region of interest to be scanned, a scan purpose, or other information. The requirement or information may be obtained from an electronic medical record, a radiology information system, and / or other source.

[0063] The method 210 even includes determining (e.g., automatically) the range of the corresponding anatomical landmarks of interest in each of the different anatomical stations (e.g., appropriate fields of view (based on head / feet)) for the corresponding localizer scan for each of the different anatomical stations based at least on the corresponding marker mask image for each of the corresponding anatomical stations (box 224). In some embodiments, the range of the corresponding anatomical landmarks of interest in each of the different anatomical stations is determined for the corresponding localizer scan for each of the corresponding anatomical stations based on both the corresponding marker mask image for each of the corresponding anatomical stations and the requirements for the scan of the subject. In some embodiments (when the coil calibration data is obtained from a single RF coil), for a single RF coil, the range is determined for the corresponding anatomical landmarks of interest in each of the different anatomical stations. In some embodiments (when the coil calibration data is obtained from multiple RF coils), for each of the multiple RF coils, the range is determined for the corresponding anatomical landmarks of interest in each of the different anatomical stations.

[0064] Method 210 also includes initiating a locator scan (e.g., a non-diagnostic scan) of the subject using the MRI system to obtain corresponding locators or locator images for one or more anatomical stations in different anatomical stations based on the determined ranges of the corresponding anatomical landmarks of interest (box 226). The locator image is a series of images (e.g., two-dimensional (2D) images) in the axial plane, the sagittal plane, and the coronal plane (e.g., a 3-plane locator scan). Method 210 even includes obtaining a locator image that includes anatomical landmarks of interest of the subject for one or more anatomical stations in different anatomical stations (box 228). Method 210 also still includes determining a geometric structure plan (e.g., a prescribed slice including the center, orientation, and range of the anatomical landmarks of interest) of a scan (e.g., a high-resolution scan, such as a diagnostic scan) of the anatomical landmarks of interest of the subject based on the locator images for one or more anatomical stations in different anatomical stations (box 230). In some embodiments, an intelligent prescription module (such as, AIR from General Electric Healthcare) x TM ) may be used to determine the geometry plan. The method 210 further includes initiating a scan (e.g., a high-resolution or diagnostic scan) to obtain a higher-resolution image of corresponding anatomical landmarks of interest for one or more of the different anatomical stations (block 232). The higher-resolution image (e.g., diagnostic image) has a higher resolution than the locator image or localizer image and the calibration image.

[0065] Figure 6 Illustrated for use Figure 1A flowchart of a method 234 for performing a spinal scan on a patient using an MRI system in FIG. Figure 1 One or more of the steps of method 234 may be performed simultaneously or sequentially. Figure 6 The depicted sequence is performed in a different order. In some embodiments, method 234 can be used for low resolution scans (as opposed to calibration scans). In some embodiments, method 234 can be used for calibration data other than coil calibration data (e.g., ASSET / PURE calibration). In some embodiments, the calibration data can be obtained from a pre-scan calibration or a mission calibration.

[0066] Method 234 includes initiating an exploit Figure 1 A pre-scan of a subject is performed using an MR scanner (eg, the MR scanner 102) in the apparatus (block 236). The pre-scan includes performing a coil calibration scan using one or more coils (eg, an RF surface coil and / or an RF body coil).

[0067] The method 234 also includes obtaining coil calibration scan data (eg, coil calibration images) of the subject acquired during a pre-scan with the MR scanner (block 238). The coil calibration data is large field of view coil calibration data. Figure 7 is an example of a coil calibration image 240 of the spine of the subject. Method 234 also includes inputting the coil calibration data into a trained deep learning-based spine multi-mask segmentation network 180 (box 242). As described above, the deep learning-based spine multi-mask segmentation network 180 is trained for spinal scanning. Specifically, the deep learning-based spine 180 multi-mask segmentation network is trained to detect and mark the cervical part of the spine, the thoracic part of the spine, and the lumbar part of the spine for the cervical spine station, the thoracic spine station, and the lumbar spine station, respectively. In certain embodiments, the deep learning-based spine multi-mask segmentation network 180 is configured to detect and mark both the cervical and thoracic parts of the spine for the cervical-thoracic spine station, detect and mark both the thoracic part and the lumbar part of the spine for the thoracic-lumbar spine station, and / or detect and mark the entirety of the spine for the spine station.

[0068] The method 234 also includes outputting labeled mask images for the different spine stations, wherein respective masks of respective labeled mask images highlight specific regions of the spine in each respective one of the different spine stations (block 244). Figure 8 is an example of an image 246 with masks for the thoracic and lumbar regions of the spine. Fig. 9 Is applied to Figure 7 Coil calibration image 240 of the spine Figure 8246 in the image 248 of the thoracic region and the lumbar region of the spine. The method 234 also includes obtaining requirements or information for a spinal scan of the subject (block 250). The requirements or information may include a scan protocol, a region of interest to be scanned, a scan purpose, or other information. The requirements or information may be obtained from an electronic medical record, a radiology information system, and / or other source.

[0069] Method 234 even includes determining (e.g., automatically) the extent of the corresponding region of the spine in each corresponding spine station in the different spine stations (e.g., an appropriate field of view (based on head / feet)) for the corresponding locator scan for each corresponding spine station based at least on the corresponding marker mask image for each corresponding spine station (box 252). In some embodiments, the extent of the corresponding region of the spine in each corresponding spine station in the different spine stations is determined for the corresponding locator scan for each corresponding spine station based on both the corresponding marker mask image for each corresponding spine station and the requirements for the scan of the subject. In some embodiments (when coil calibration data is obtained from a single RF coil), for a single RF coil, the extent is determined for the corresponding region of the spine in each corresponding spine station in the different spine stations. In some embodiments (when coil calibration data is obtained from multiple RF coils), for each RF coil in the multiple RF coils, the extent is determined for the corresponding region of the spine in each corresponding spine station in the different spine stations.

[0070] Method 234 also includes initiating a locator scan (e.g., a non-diagnostic scan) of the subject using the MRI system to obtain corresponding locator or locator images for one or more of the different spinal stations based on the determined extent of the corresponding region of the spine (box 254). The locator image is a series of images (e.g., two-dimensional (2D) images) in the axial plane, the sagittal plane, and the coronal plane (e.g., a 3-plane locator scan). Method 234 even includes obtaining a locator image of the corresponding region of the subject's spine for one or more of the different spinal stations (box 256). Method 234 also still includes determining a geometric structure plan (e.g., prescribed slices including the center, orientation, and extent of the spinal region of interest) of a spinal scan (e.g., a high-resolution spinal scan, such as a diagnostic spinal scan) of the subject's spine based on the locator images for one or more of the different spinal stations (box 258). In some embodiments, a smart prescription module (such as, AIR from General Electric Healthcare) x TM) may be used to determine the geometry plan. The method 234 further includes initiating a spine scan (e.g., a high-resolution or diagnostic spine scan) to obtain higher-resolution images of corresponding regions of the spine for one or more of the different spine stations (block 260). The higher-resolution images (e.g., diagnostic images) have a higher resolution than the locator or localizer images and the calibration images.

[0071] Fig.10 is to use Fig. 9 An example of a method 234 in which a coil calibration image 262 is used with mask markings identifying the thoracic region of the spine and a corresponding ground truth image 264. The ground truth image 264 is a higher resolution image obtained by stitching different images together. Fig.11 is to use Fig. 9 Another example of method 234 utilizing a mask-marked coil calibration image 266 identifying the thoracic region of the spine and a corresponding ground truth image 268. Ground truth image 268 is a higher resolution image obtained by stitching different images together. Fig.12 is to use Fig. 9 The method 234 in FIG. 2 is another example of utilizing a mask-marked coil calibration image 270 identifying the thoracic region of the spine and a corresponding ground truth image 272. The ground truth image 272 is a higher resolution image obtained by stitching different images together.

[0072] The above techniques can be extended to other applications of anatomical zoning, such as station identification in a PET / MR MRAC setting or a whole-body examination setting. Fig.13 274 and 276 of a first portion and a second portion of a subject's body, respectively. As marked, the coil calibration image 274 includes the head, neck, and lungs. As marked, the coil calibration image 276 includes the abdomen and pelvis. Figure 5 The method 210 described above is used to mark corresponding anatomical landmarks of interest in different anatomical stations (e.g., using Figure 2 The method can then be used to determine the extent of each corresponding anatomical landmark of interest in each of the different anatomical stations using a whole body multi-mask segmentation network 182 or a PET / MR MRAC multi-mask segmentation network 184 as described in the foregoing. For example, the different anatomical stations can be the head, chest, abdomen, and pelvic regions. The number of anatomical stations can vary depending on the application. In addition, what part of the body is included in each anatomical station can vary depending on the application.

[0073] Similarly, the above techniques can be extended to other regions of interest in the body for partitioning. Figure 5The method 210 described above is used to mark the corresponding anatomical landmarks of interest within a specific region of interest (e.g., using Figure 2 The method can then be used to determine the range of each corresponding anatomical landmark of interest in the region of interest using the region of interest multi-mask segmentation network 186 described in . Figure 5 The method 210 described can be applied to the lungs to detect and mark and determine the corresponding extents of different parts of the lungs. In another example, Figure 5 The described method 210 may be applied to a shoulder to detect and mark as well as determine the corresponding extents of different portions of the shoulder.

[0074] The technical effects of the subject matter disclosed in the present invention include providing an automatic patient-specific range estimate for the correct locator or high-resolution scan during the pre-scan stage. For example, the embodiments disclosed in the present invention enable the technology to automatically determine the range of the spine station (e.g., cervical, thoracic, lumbar) from the calibration image as part of the pre-scan, and enable the correct station locator to be scanned for further use by the technician when planning the spine examination. These technologies can be extended to determine other anatomical boundaries in the context of whole-body imaging or PET / MR MRAC boundaries. Using these technologies, all presets necessary for regional stratification can be obtained from pre-scan data and geometry, where the geometric and patient-specific locators are available to technicians at any time. The technical effects of the subject matter disclosed in the present invention include ensuring that the correct station (e.g., only cervical or only thoracic or only lumbar) or station (e.g., cervical-thoracic, thoracic-lumbar or entire spine) locator is automatically available to the user for spine planning. This is especially true for the thoracic station, which is usually separated across two locators in current practice. The technical effects of this theme also include allowing a single locator to be automatically generated. The technical effects of this theme also include eliminating the need for splicing. Technical effects of the subject matter even include facilitating easy setup for planning exams. Embodiments disclosed herein may also be used to automatically establish patient anatomy specific head / foot coverage presets for multi-station whole body imaging exams (e.g., PET / MR MRAC). Technical effects of the subject matter disclosed herein also include reducing the time to establish subsequent scans (e.g., localizer scans and / or higher resolution scans (e.g., diagnostic scans)).

[0075] 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).

[0076] 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 performing regional layering on a calibration image or a low-resolution image, the computer-implemented method comprising: obtaining, at a processor, calibration scan data or a low-resolution image of a subject acquired using a magnetic resonance (MR) scanner of a magnetic resonance imaging system; inputting the calibration data or the low-resolution image into a trained deep learning-based multi-mask segmentation network via the processor; outputting, via the processor, labeled mask images for different anatomical stations, wherein respective masks of respective labeled mask images highlight anatomical landmarks of interest in each respective one of the different anatomical stations; as well as Determining, via the processor, a range of a respective anatomical landmark of interest in each respective one of the different anatomical stations for a respective localizer scan for each respective anatomical station based at least on a respective marker mask image for each respective anatomical station.

2. A computer-implemented method according to claim 1, wherein the trained deep learning-based multi-mask segmentation network is trained to detect and label the corresponding anatomical landmarks of interest in each corresponding anatomical station in the different anatomical stations.

3. The computer-implemented method of claim 1 , further comprising: A request for a scan of the subject is obtained at the processor.

4. A computer-implemented method according to claim 3, wherein the range of the corresponding anatomical landmarks of interest in each of the different anatomical stations is determined for the corresponding locator scan for each corresponding anatomical station based on both the corresponding marker mask image for each corresponding anatomical station and the requirements for the scan of the subject.

5. The computer implemented method of claim 1, wherein the calibration data comprises coil calibration data, and wherein the coil calibration scan data is obtained from a single channel or multi-channel radio frequency body or surface or transmit / receive or receive only coil. 6 . The computer-implemented method of claim 1 , wherein the calibration data comprises coil calibration data, and wherein the coil calibration data is obtained from a plurality of radio frequency coils.

7. The computer-implemented method of claim 6, wherein for each of the plurality of radio frequency coils used for the respective localizer scan for each respective anatomical station, the range of the respective anatomical landmark of interest in each respective one of the different anatomical stations is determined.

8. The computer-implemented method of claim 1, wherein the different anatomical stations are used for spinal scans utilizing the MR scanner.

9. The computer-implemented method of claim 1, wherein the different anatomical stations are used for whole body scanning using the MR scanner.

10. The computer-implemented method of claim 1, wherein the different anatomical stations are used for attenuation-corrected MR imaging-based scanning utilizing the MR scanner and a positron emission tomography (PET) scanner, wherein the MR imaging system is part of a PET / MR imaging system.

11. A system for performing region layering on a calibration image or a low-resolution image, the system comprising: a memory encoding a processor-executable routine; and a processor configured to access the memory and execute the processor-executable routine, wherein the processor-executable routine, when executed by the processor, causes the processor to: obtaining calibration scan data or a low-resolution image of a subject acquired using a magnetic resonance (MR) scanner of a magnetic resonance imaging system; Inputting the calibration data or the low-resolution image into a trained deep learning-based multi-mask segmentation network; outputting labeled mask images for different anatomical stations, wherein respective masks of respective labeled mask images highlight anatomical landmarks of interest in each respective anatomical station of said different anatomical stations; as well as An extent of a respective anatomical landmark of interest in each of the different anatomical stations is determined for a respective localizer scan for each respective anatomical station based at least on the respective marker mask image for each respective anatomical station.

12. The system of claim 11, wherein the trained deep learning-based multi-mask segmentation network is trained to detect and label the respective anatomical landmarks of interest in each respective one of the different anatomical stations.

13. A system according to claim 11, wherein the processor executable routine, when executed by the processor, also causes the processor to obtain a requirement for a scan of the subject, and wherein based on both the corresponding marker mask image for each corresponding anatomical station and the requirement for the scan of the subject, the range of the corresponding anatomical landmark of interest in each of the different anatomical stations is determined for the corresponding locator scan for each corresponding anatomical station.

14. The system of claim 11, wherein the calibration data comprises coil calibration data, and wherein the coil calibration scan data is obtained from a single channel or multi-channel radio frequency body or surface or transmit / receive or receive only coil.

15. The system of claim 11, wherein the calibration data comprises coil calibration data, wherein the coil calibration data is obtained from a plurality of radio frequency coils, and wherein for each of the plurality of radio frequency coils used for the respective localizer scan for each respective anatomical station, the range of the respective anatomical landmark of interest in each respective one of the different anatomical stations is determined.

16. The system of claim 11, wherein the different anatomical stations are used for spinal scanning using the MR scanner.

17. The system of claim 11, wherein the different anatomical stations are used for whole body scanning using the MR scanner.

18. The system of claim 11, wherein the different anatomical stations are used for MR imaging-based attenuation correction scanning using the MR scanner and a positron emission tomography (PET) scanner, wherein the MR imaging system is part of a PET / MR imaging system.

19. A non-transitory computer readable medium comprising processor executable code which, when executed by a processor, causes the processor to: obtaining calibration scan data or a low-resolution image of a subject acquired using a magnetic resonance (MR) scanner of a magnetic resonance imaging system; Inputting the calibration data or the low-resolution image into a trained deep learning-based multi-mask segmentation network; outputting labeled mask images for different anatomical stations, wherein respective masks of respective labeled mask images highlight anatomical landmarks of interest in each respective anatomical station of said different anatomical stations; as well as An extent of a respective anatomical landmark of interest in each of the different anatomical stations is determined for a respective localizer scan for each respective anatomical station based at least on the respective marker mask image for each respective anatomical station.

20. The non-transitory computer-readable medium of claim 19, wherein the trained deep learning-based multi-mask segmentation network is trained to detect and label the respective anatomical landmarks of interest in each respective one of the different anatomical stations.

Images