Intelligent positioning and protocol recommendation method, device and system for magnetic resonance scanning and medium

Abstract

The invention provides an intelligent positioning and protocol recommendation method, device and system for magnetic resonance scanning and a medium, and the method comprises the steps: obtaining a positioning image of a patient, and recognizing the anatomical structure information in the positioning image through a pre-trained deep learning model; based on the identified anatomical structure information, according to a preset clinical rule base, generating at least one suggested scanning range frame on the positioning image; and displaying the generated scanning range frame in an interactive interface, and adjusting the scanning range frame by responding to a user operation instruction. According to the invention, the traditional manual operation is converted into a one-key confirmation mode or a fine adjustment mode, so that the scanning preparation time is greatly shortened, and the equipment turnover rate is improved. Based on an AI algorithm and a preset rule, man-made differences are eliminated, and consistency and normalization of scanning results in different scenes are guaranteed. Dependence on personal experience is reduced, new user training cost is reduced, low-level errors are avoided, and user experience is optimized through an intuitive and intelligent interaction interface.

Description

Technical Field

[0001] This application relates to the field of magnetic resonance scanning technology, and in particular to intelligent positioning and protocol recommendation methods, apparatus, systems and media for magnetic resonance scanning. Background Technology

[0002] In the field of medical imaging scanning, the traditional scanning process typically involves multiple steps, including localization scanning, manual selection of the scan area, protocol selection and parameter adjustment, and initiation of the scan. Specifically, the user first performs a localization scan to obtain a localization image, then manually drags a frame line on the localization image to determine the precise scan area. Next, based on the clinical information on the examination request form (such as a description of symptoms like "knee pain"), the user selects a suitable protocol from hundreds of preset scan protocols using their personal experience. In some scenarios, the parameters of the selected protocol also need to be manually adjusted. Only after completing these steps can the actual scan begin. This process has long been the standard operating procedure for medical imaging examinations and is widely used in various clinical imaging scanning scenarios.

[0003] However, traditional scanning methods have significant operational complexity and efficiency bottlenecks. (1) Complex and time-consuming operation: The process of manually positioning the scanning range and selecting and adjusting the protocol requires repeated operation by the user, which not only prolongs the patient's examination preparation time, but also increases the user's workload. (2) Results are affected by personal experience: The scanning results are highly dependent on the user's personal experience. Different users have different positioning operation habits and protocol selection preferences, which makes it difficult to form a unified standard for the scanning range and image quality of the same disease, thus affecting the consistency and accuracy of clinical diagnosis. (3) Prone to errors: In scenarios where medical resources are scarce and the work pace is busy, users are prone to selecting the wrong protocol or positioning deviation, resulting in unqualified scanning results, which requires rescanning. This not only reduces the overall diagnostic efficiency, but also increases the wear and tear cost of imaging equipment. (4) High learning threshold: New users need to spend a lot of time learning and mastering the applicable scenarios and positioning operation skills of hundreds of preset protocols, which further increases the human resource training cost of clinical imaging departments. Summary of the Invention

[0004] In view of the shortcomings of the prior art described above, the purpose of this application is to provide a method, apparatus, system and medium for intelligent positioning and protocol recommendation for magnetic resonance scanning, in order to solve the technical problems of traditional scanning methods, such as complex and time-consuming operation, results affected by personal experience, easy to make mistakes, and high learning threshold.

[0005] To achieve the above and other related objectives, a first aspect of this application provides an intelligent positioning and protocol recommendation method for magnetic resonance imaging (MRI) scans, comprising: acquiring a patient's positioning image; identifying anatomical structure information in the positioning image using a pre-trained deep learning model; generating at least one suggested scanning bounding box on the positioning image based on the identified anatomical structure information and according to a preset clinical rule base; displaying the generated scanning bounding box in an interactive interface; and adjusting the scanning bounding box in response to user operation commands.

[0006] In some embodiments of the first aspect of this application, the intelligent positioning and protocol recommendation method further includes: acquiring the patient's clinical information and physiological parameters corresponding to the patient's positioning image to form multimodal data with the patient's positioning image, and standardizing the multimodal data; inputting the standardized positioning image features, clinical information semantic features, and physiological parameter quantification features into a pre-trained deep learning model, and obtaining fused features through the feature fusion module built into the model; the pre-trained deep learning model adjusting the recognition strategy according to the fused features, and outputting accurate recognition results containing lesion-related regions.

[0007] In some embodiments of the first aspect of this application, the intelligent localization and protocol recommendation method further includes: adding an image enhancement module to the input end of the pre-trained deep learning model, for improving the anatomical structure clarity of the localization image through super-resolution reconstruction and noise suppression algorithms.

[0008] In some embodiments of the first aspect of this application, the image enhancement module at the input end of the deep learning model is configured to perform the following: noise suppression processing on the input original positioning image data to suppress noise caused by low-dose scanning of the positioning image; and super-resolution reconstruction of the noise-suppressed positioning image using a super-resolution algorithm based on generative adversarial networks or convolutional neural networks.

[0009] In some embodiments of the first aspect of this application, the intelligent positioning and protocol recommendation method further includes: after automatically generating a scanning range box, providing dedicated fine-tuning handles at the boundaries and corners of the box; the fine-tuning handles are configured as semantic controls associated with anatomical structures.

[0010] In some embodiments of the first aspect of this application, after generating and adjusting the scan range box, the intelligent positioning and protocol recommendation method further performs a protocol recommendation step, which includes: extracting structured clinical information and anatomical structure information identified in the positioning image from the examination request form to integrate them into a context data set; calling a preset protocol library, performing multi-dimensional matching and sorting based on the context data set, and recommending several protocols with the highest priority; and displaying the key parameters of the recommended protocols in an interactive interface for comparison.

[0011] In some embodiments of the first aspect of this application, the intelligent positioning and protocol recommendation method further performs the following: displaying different clinical task objectives through a protocol booster, each clinical task objective corresponding to a type of examination request.

[0012] To achieve the above and other related objectives, a second aspect of this application provides an intelligent positioning and protocol recommendation system for magnetic resonance imaging (MRI) scans, comprising: an anatomical structure information extraction module for acquiring a patient's positioning image and identifying anatomical structure information in the positioning image using a pre-trained deep learning model; a scan bounding box generation module for generating at least one suggested scan bounding box on the positioning image based on the identified anatomical structure information and according to a preset clinical rule base; and a bounding box adjustment module for displaying the generated scan bounding box in an interactive interface and adjusting the scan bounding box in response to user operation commands.

[0013] To achieve the above and other related objectives, a third aspect of this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the intelligent positioning and protocol recommendation method for magnetic resonance scanning.

[0014] To achieve the above and other related objectives, a fourth aspect of this application provides a computer program product comprising computer program code that, when executed on a computer, causes the computer to implement the intelligent positioning and protocol recommendation method for magnetic resonance imaging.

[0015] To achieve the above and other related objectives, a fifth aspect of this application provides a computer device including a memory, a processor, and a computer program stored in the memory; the processor executes the computer program to implement the intelligent positioning and protocol recommendation method for magnetic resonance scanning.

[0016] As described above, the intelligent positioning and protocol recommendation method, apparatus, system, and medium for magnetic resonance scanning of this application have the following beneficial effects:

[0017] (1) Significantly improve efficiency: The manual operation is transformed into "one-click confirmation" or "fine-tuning", which greatly shortens the preparation time and improves the equipment turnover rate.

[0018] (2) Ensure consistent quality: By using standardized algorithms and rules, human differences are eliminated, ensuring that different users and different devices can obtain consistent and standard scanning results.

[0019] (3) Reduced operation threshold and error rate: Reduced over-reliance on users' personal experience, reduced training costs for new users, and effectively avoided basic errors such as choosing the wrong protocol.

[0020] (4) Optimize user experience: Provide users with an intuitive, intelligent and efficient interactive interface, reducing workload and psychological pressure. Attached Figure Description

[0021] Figure 1 The diagram shown is a framework schematic of an intelligent positioning and protocol recommendation method for magnetic resonance scanning according to an embodiment of this application.

[0022] Figure 2 The diagram shown is a flowchart illustrating an intelligent positioning and protocol recommendation method for magnetic resonance scanning according to an embodiment of this application.

[0023] Figure 3A The diagram shown is an interface schematic of the original positioning image in one embodiment of this application.

[0024] Figure 3B The diagram shown is a schematic representation of an organ outline interface annotated by AI after automatic recognition in one embodiment of this application.

[0025] Figure 3C The diagram shown is an interface schematic showing the range of adjustable organ contours in one embodiment of this application.

[0026] Figure 3D The diagram shows an interface schematically illustrating the automatic addition of a scan range frame to the organ outline in one embodiment of this application.

[0027] Figure 4 The diagram shows the interface of the smart protocol recommendation function in one embodiment of this application.

[0028] Figure 5 The diagram shown is a structural schematic of an intelligent positioning and protocol recommendation system for magnetic resonance scanning according to an embodiment of this application.

[0029] Figure 6 The diagram shown is a structural schematic of a computer device according to an embodiment of this application. Detailed Implementation

[0030] The following specific examples illustrate the implementation of this application. Those skilled in the art can easily understand other advantages and effects of this application from the content disclosed in this specification. This application can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of this application. It should be noted that, unless otherwise specified, the following embodiments and features in the embodiments can be combined with each other.

[0031] Before providing a further detailed description of the present invention, the nouns and terms used in the embodiments of the present invention are explained, and the nouns and terms used in the embodiments of the present invention are subject to the following interpretations:

[0032] 1. Magnetic Resonance Imaging (MRI): This is a non-invasive imaging technique that uses strong magnetic fields and radio frequency pulses to cause hydrogen protons in human tissue to resonate, generating tomographic images by receiving the resonance signals. It does not emit ionizing radiation and can clearly show soft tissue details.

[0033] 2. Scout image: This is a rapid, low-resolution image acquired before an MRI scan to clearly define the anatomical location and extent of the scanned area. It provides coordinate references for subsequent scan sequences, helping technicians to accurately set the scan slices, extent, and parameters.

[0034] 3. U-Net: This is a symmetrical convolutional neural network. Its core consists of an encoding path (feature extraction) and a decoding path (resolution restoration), with skip connections fusing multi-scale features. It excels in medical image segmentation tasks, accurately segmenting target regions such as organs and lesions, providing structured data support for clinical diagnosis.

[0035] 4. Faster R-CNN: This is a target detection algorithm based on region proposal. It extracts image features through shared convolutional layers, generates candidate regions using an RPN network, and then completes target localization and recognition through classification and regression modules. This algorithm balances detection accuracy and speed, and can efficiently identify targets such as lesions and anatomical structures in medical images, making it suitable for various image analysis tasks.

[0036] 5. ResNet (Residual Network): This is a deep learning model that solves the vanishing gradient problem in deep networks by introducing residual connections. Its core is the construction of residual blocks, allowing the network to directly learn the residual mapping between input and output. It can deeply mine the complex features of medical images and is widely used in various tasks such as image classification, segmentation, and denoising.

[0037] 6. Super-resolution reconstruction: This is a technique that improves the spatial resolution of low-resolution images through algorithms. It is divided into traditional interpolation methods and deep learning methods (such as ESRGAN). By supplementing the missing details in the image, it transforms blurry medical images (such as localization images and low-dose scans) into high-resolution images, which helps to accurately identify subtle anatomical structures and lesions.

[0038] 7. Noise Suppression Algorithms: These are preprocessing techniques targeting interference such as salt-and-pepper noise and Gaussian noise in medical images. They include adaptive median filtering, non-local mean denoising, and deep learning denoising networks. Their core principle is to effectively suppress noise interference and improve the image signal-to-noise ratio while preserving anatomical details and edge features.

[0039] The technical solution of this application can be applied to various MRI scenarios, whether it is routine clinical diagnostic scans of the head, spine, joints, etc., or specialized examinations such as tumor screening, vascular imaging, and neurological function assessment, or adaptive scans for special populations such as children, the elderly, and obese individuals, it can leverage its intelligent advantages. Through core functions such as localization image enhancement, semantic recognition of anatomical structures, automatic scan range planning, clinical task-driven protocol recommendation, and interactive fine-tuning, this solution can adapt to the scanning needs of different examination purposes and different patient individual characteristics, effectively simplifying the operation process, improving the level of scanning standardization, and reducing human error. It provides efficient, accurate, and convenient technical support for MRI examinations in medical institutions at all levels, thereby improving the efficiency and accuracy of clinical diagnosis.

[0040] To facilitate understanding of the embodiments of this application, firstly, in conjunction with Figure 1 and Figure 2 Detailed explanation.

[0041] Figure 1 The overall framework diagram of the embodiments of the present invention is shown, which is divided into three core frameworks. First, the patient's medical image data is acquired by scanning with MRI equipment; then, the data is input into a deep learning module for processing. This module uses a pre-trained model to complete intelligent analysis such as anatomical structure recognition, scan range planning, and protocol matching; finally, the processing results are visualized on the interactive interface for users to view, fine-tune, and execute the scanning process, realizing full automation and intelligence from data acquisition to intelligent processing to interface presentation.

[0042] Figure 2 This demonstrates the implementation process of the intelligent positioning and protocol recommendation method for magnetic resonance scanning provided by the present invention in a specific embodiment, which mainly includes the following steps:

[0043] Step S21: Obtain the patient's localization image and identify the anatomical structure information in the localization image using a pre-trained deep learning model.

[0044] The aforementioned scout image is an initial scan image used in medical imaging to assist in precise localization. Also known as a reconnaissance image or localization film, it acquires the overall anatomical contour of the target examination site through low-dose, rapid scanning, without requiring high-resolution imaging. Its core function is to provide an anatomical reference benchmark for subsequent formal scans, helping to determine key parameters such as scan range, slice thickness, and angles, replacing the traditional manual selection of the scan area. In the embodiments of this application, the scout image serves as the input data source for a deep learning model, providing basic image information for anatomical structure recognition.

[0045] The pre-trained deep learning model mentioned above refers to an artificial intelligence model that has undergone initial training on a large-scale labeled medical image dataset before performing this anatomical structure recognition task. In the embodiments of this application, the model architecture of this artificial intelligence model can be selected from segmentation networks or object detection networks suitable for medical images, including but not limited to U-Net, Faster R-CNN, ResNet, etc. The segmentation network or object detection network suitable for medical images has been specifically optimized for the features of medical images (such as grayscale distribution, anatomical structure morphology, tissue contrast, etc.). The pre-training process requires the use of massive amounts of labeled medical image data (covering normal / lesion sites of patients of different genders, ages, and body types), and the model parameters are continuously adjusted through the backpropagation algorithm so that the model learns the common features of different anatomical structures in the images (such as shape, positional relationship, grayscale range, etc.), and has a preliminary ability to recognize anatomical structures. In practical applications, this pre-trained model can be fine-tuned based on a small amount of specific scene data to further improve the recognition accuracy of target anatomical structures.

[0046] The anatomical information mentioned above refers to core human organs, tissues, or limbs directly related to the patient's examination needs (such as the head, liver, and knee joint listed in the protocol). These structures are the focus of clinical diagnosis, and their accurate identification in the localization image is a core prerequisite for automatically determining the scanning range and matching the scanning protocol. For example, when a patient undergoes an MRI examination due to knee pain, the knee joint and surrounding related tissues constitute the anatomical information. The model needs to accurately identify the contour, location, and anatomical boundaries of this structure to ensure that the subsequent scanning range can completely cover the target area while avoiding invalid scans of irrelevant areas. The entire process utilizes a deep learning model for the automatic identification of anatomical information in the localization image.

[0047] In some preferred embodiments, the intelligent positioning and protocol recommendation method further includes: acquiring the patient's clinical information and physiological parameters corresponding to the patient's positioning image to form multimodal data with the patient's positioning image, and standardizing the multimodal data; inputting the standardized positioning image features, clinical information semantic features, and physiological parameter quantification features into a pre-trained deep learning model, and obtaining fused features through the model's built-in feature fusion module; the pre-trained deep learning model adjusts the recognition strategy according to the fused features and outputs accurate recognition results containing lesion-related regions. This is because, in actual medical image recognition, it is also necessary to consider that positioning images can only capture the outline of anatomical structures and cannot be combined with the patient's specific condition, resulting in insufficient targeting of the scanning range (such as missing key diagnostic areas such as cartilage wear and lesion tissue boundaries). In addition, different patients have different clinical needs (such as lesion type, postoperative follow-up, etc.) and physiological characteristics (such as age, body type, etc.), and general recognition strategies are difficult to adapt to personalized scanning needs, easily causing invalid scans or incomplete coverage of lesion areas, thereby affecting the accuracy of subsequent diagnosis.

[0048] Specifically, the first step is to acquire the patient's multimodal data, including localization images, clinical information, and physiological parameters. Localization images are anatomical contour images generated by low-dose, rapid scanning, as explained above and will not be repeated here. Clinical information refers to structured data extracted from electronic medical records, such as disease diagnosis, examination purpose, and surgical history, like "knee osteoarthritis" or "post-liver cancer follow-up." Physiological parameters refer to quantitative data such as weight, height, and age retrieved from the patient information system. This multimodal data is then standardized, including formatting and noise suppression of localization images, keyword extraction and encoding of clinical information (e.g., classifying diseases as pathological, follow-up, or health screening types), and normalization of physiological parameters, ensuring that multi-source data can be collaboratively analyzed by the model. Finally, the standardized localization image features, clinical information semantic features, and physiological parameter quantitative features are input into a pre-trained deep learning model. Through the model's built-in feature fusion module (such as a cross-modal attention mechanism), deep association of different types of features is achieved; for example, the clinical label "knee osteoarthritis" is bound to the grayscale features of the joint space in the localization image. The pre-trained deep learning model automatically adjusts the recognition strategy based on fusion features, adjusting the focus of key anatomical details according to clinical needs. For example, it focuses on recognizing articular cartilage in patients with lesions and on recognizing the tissues surrounding the lesion in patients after tumor surgery, rather than just outputting the basic anatomical structure outline, and finally generating accurate recognition results that include the lesion-related areas.

[0049] In some preferred embodiments, the intelligent localization and protocol recommendation method further includes: adding an image enhancement module to the input of the pre-trained deep learning model to improve the clarity of the anatomical structures in the localization image through super-resolution reconstruction and noise suppression algorithms. This is because, although conventional pre-trained deep learning models can also perform image recognition on localization images, they do not consider the low-dose characteristics of the images, and the recognition accuracy is easily affected. In this preferred embodiment, by adding an image enhancement module to the deep learning model, it can be ensured that the model can still accurately identify anatomical structures without increasing the patient's radiation dose, thus resolving the contradiction between low dose and high recognition accuracy. The specific execution process is as follows:

[0050] First, the original data of the localization image is input into the image enhancement module. The module first performs noise suppression processing on the image, including addressing interference such as salt-and-pepper noise and Gaussian noise caused by low-dose scanning of the localization image. It adopts adaptive median filtering, non-local mean denoising (NLM), or a deep learning-based denoising network (such as DnCNN). By analyzing the gray-level distribution characteristics and neighborhood correlation of image pixels, it accurately distinguishes noise from anatomical structure signals, filters out invalid noise while preserving structural details, and reduces the interference of noise on subsequent feature extraction.

[0051] It should be noted that salt-and-pepper noise is a common discrete random noise in medical imaging, manifesting as randomly distributed pure white bright spots (like salt grains) and pure dark spots (like pepper grains) in the image, named for its resemblance to sprinkled salt and pepper. Its causes are often related to sensor photoelectric conversion errors and photon statistical fluctuations during low-dose scanning. It interferes with the identification of anatomical structural details and requires targeted denoising algorithms for removal. Gaussian noise is a continuous noise that follows a normal distribution. In medical imaging, it appears as a uniform gray-level interference layer superimposed on the entire image, reducing image clarity but without obvious discrete noise. It mainly originates from thermal noise of electronic components in imaging equipment and signal transmission interference, blurring the edges and textures of anatomical structures. It is a type of noise that needs to be suppressed in image preprocessing. Adaptive median filtering is a nonlinear denoising algorithm that dynamically adjusts the filtering window. Its core principle is to automatically change the size and shape of the filtering window based on the noise density and pixel gray-level distribution of local image regions. Compared to traditional median filtering, it can effectively remove discrete noise such as salt-and-pepper noise while better preserving the edge details of anatomical structures, avoiding excessive image blurring. Nonlocal means denoising is a denoising method based on image redundancy information. Its core idea is to calculate the similarity between any two pixel blocks in an image and then use a weighted average of similar pixel blocks globally to denoise the target pixels. Deep learning-based denoising networks are a class of denoising models with deep neural networks at their core, with DnCNN being a typical example. By constructing multi-layer convolutional structures, it learns the mapping relationship between noise features and essential image features from massive pairs of noisy and clean images. It can adaptively process different types and intensities of noise, accurately restoring the anatomical details of medical images while removing noise.

[0052] Secondly, super-resolution reconstruction is performed on the basis of noise suppression, including using super-resolution algorithms based on generative adversarial networks (GAN) or convolutional neural networks (CNN) (such as ESRGAN, RCAN) to perform pixel-level optimization on the denoised low-resolution localization image. The aforementioned pixel-level optimization refers to learning the mapping relationship between low-resolution images and high-resolution images through the model, supplementing the missing details in the image (such as the edge texture of anatomical structures and the gray-level gradient of tissue boundaries), improving the spatial resolution and clarity of the image, and making the originally blurred anatomical structure outlines (such as small blood vessels and tissue gaps) sharper and more discernible.

[0053] It should be noted that Generative Adversarial Networks (GANs) are deep learning frameworks where two neural networks, a generator and a discriminator, compete against each other and train collaboratively. The generator learns the distribution characteristics of real data and generates simulated data, while the discriminator distinguishes between generated and real data. Performance is improved through continuous competition between the two. In the field of medical imaging, GANs are commonly used for tasks such as image denoising and super-resolution reconstruction, generating detailed and highly realistic results. Convolutional Neural Networks (CNNs) are a class of deep learning models specifically designed for processing grid-structured data (such as images). Their core consists of convolutional layers, pooling layers, and fully connected layers. They extract local features (such as edges, textures, and shapes) from images through convolutional operations and reduce data dimensionality while preserving key information through pooling layers. They can automatically learn deep features of anatomical structures in medical images and are widely used in medical image processing tasks such as image recognition, segmentation, and denoising. Super-resolution algorithms are a class of techniques used to improve the spatial resolution of images; ESRGAN and RCAN are representative algorithms. It analyzes the pixel distribution and feature mapping relationship of low-resolution images, uses deep learning models to supplement missing details, and transforms low-resolution, blurry medical images (such as localization images) into high-resolution, clear images, thereby highlighting the edge contours and subtle features of anatomical structures and providing support for subsequent identification and diagnosis.

[0054] Finally, the localization image optimized by noise suppression and super-resolution reconstruction is output to the core recognition module of the pre-trained deep learning model. At this point, the image has clearer anatomical features and lower noise interference, providing high-quality input data for the model to accurately identify anatomical information (such as the boundary of lesion tissue and organ contours), ensuring the accuracy and stability of the subsequent recognition process.

[0055] Step S22: Based on the identified anatomical structure information, generate at least one suggested scan range box on the localization image according to a preset clinical rule base.

[0056] The aforementioned pre-defined clinical rule base refers to a set of structured rules built upon clinical diagnosis and treatment guidelines, imaging examination guidelines, and extensive practical experience. Its aim is to transform the scanning range standards developed by doctors and users over long-term scanning work into quantifiable rules and logical instructions that the system can recognize. Typically, each rule in the rule base includes core elements such as the applicable scenario (e.g., examination site, disease type), anatomical boundary definition (e.g., starting / ending anatomical structures), and range redundancy (e.g., exceeding the target structure edge by 5mm to avoid omitting surrounding related tissues). It should be noted that these rules are not subjective judgments based on experience, but standardized requirements verified by evidence-based medicine and covering different examination sites, lesion types, and scanning purposes. For example, the rule that whole-brain scans should cover from the top of the corpus callosum to the cerebellar tonsils clearly defines the anatomical landmarks corresponding to the upper and lower boundaries of the whole-brain scan, ensuring that the scan range can completely include the core functional areas of the brain without omitting key areas such as the frontal lobe, occipital lobe, and cerebellum; the knee joint scan rules clearly define coverage from the distal 1 / 3 of the femur to the proximal 1 / 3 of the tibia, including the complete joint space and the medial and lateral menisci; the breast scan rules stipulate coverage from the lower edge of the clavicle to the lower edge of the breast, and from the left and right to the lateral chest wall.

[0057] The suggested scanning bounding box mentioned above is a visual virtual frame based on the preset clinical rule base, used to define the formal scanning area. Its function is to replace the traditional manually dragged frame lines, clearly defining the upper, lower, left, right, front, and back boundaries of the scan, as well as parameters such as slice thickness and interslice spacing. It should be noted that this scanning bounding box is not a fixed-size general box, but a personalized result based on the individual anatomical structure of the patient. For example, after identifying the specific pixel coordinates of the top of the corpus callosum and the cerebellar tonsils in the localization image, the system will automatically calculate and generate a scanning box adapted to the size and position of the patient's head according to the rule requirement in the rule base that "whole brain scan covers this anatomical region". This ensures that the frame accurately covers the target anatomical structure, while avoiding invalid scans caused by including too much irrelevant tissue (such as neck muscles).

[0058] In this embodiment, the generation of the scan range bounding box based on a preset clinical rule base is based on the mapping relationship between the anatomical structure recognition results and the rule base. Specifically, the deep learning model identifies the anatomical structure information in the localization image and outputs quantitative information such as the specific coordinates and contour morphology of these structures in the image. Subsequently, according to the patient's examination needs (such as whole brain scan, knee pain examination, etc.), the corresponding target rules in the rule base are matched to determine the anatomical interval that the scan range needs to cover. Next, based on the coordinates of the identified anatomical structures, combined with the coverage range, redundancy, and other requirements specified in the rule base, the scan range bounding box is automatically generated through coordinate calculation. For example, if the rule requires the whole brain scan to cover the top of the corpus callosum to the cerebellar tonsils, then the upper and lower edge coordinates of the identified anatomical structures of the top of the corpus callosum and the cerebellar tonsils are used as the basis, and a preset redundancy (such as 2mm) is added outward to finally determine the upper and lower boundaries of the scan frame, while the left and right boundaries are automatically adapted according to the widest part of the skull contour. The entire process combines standardized clinical rules with individual patient anatomical data, ensuring both the standardization of the scanning range (avoiding human differences) and personalized adaptation (fitting the anatomical characteristics of different patients).

[0059] To facilitate understanding by those skilled in the art, the following is combined with Figures 3A-3D This explains the process from the original localization image to contour recognition, then to fine-tuning the scanning bounding box, and finally to the system automatically adding the scanning bounding box. Figure 3A It is a raw MRI localization image of the skull, clearly showing the anatomical structures of the brain and eye sockets. Figure 3B It is an organ outline that has been automatically identified and annotated by AI. The identified target brain region is displayed intuitively through green areas and markers, realizing a visual presentation of intelligent recognition of anatomical structure from the original image. Figure 3C The AI ​​automatically identifies the outline of brain organs, presenting an adjustable range. A "Confirm" button is located in the lower right corner of the interface, allowing users to fine-tune the identified area and confirm the scan range by clicking the button, achieving precise and convenient interactive operation. Figure 3D Based on the brain organ outline regions automatically identified by AI, a clear scanning range bounding box is automatically added. This box accurately covers the target anatomical area, and the box and the markers inside it intuitively present the planned scanning boundary, providing a clear and standardized reference for defining the range of subsequent MRI scans.

[0060] Step S23: Display the generated scan range frame in the interactive interface, and adjust the scan range frame in response to user operation commands.

[0061] In some optional implementations, after the automatic generation of the scan range is completed, the generated scan range frame is highlighted in the visualization interface of the clinical imaging workstation. For example, the scan frame is clearly distinguished from the anatomical background of the localization image by using bright colors (such as red and blue) or thick borders. At the same time, key information such as the scan site, slice thickness, and coverage area corresponding to the frame are marked (such as whole brain scan: top of corpus callosum - cerebellar tonsils). This allows users to quickly and intuitively confirm the rationality of the automatic planning results.

[0062] Next, users interact with the system based on actual clinical needs (such as patient anatomical deformities, special lesions requiring expanded coverage, and minor automatic frame shifts). To adjust the range, users can directly drag the edge or corner control points of the scanning frame; the system responds in real time and updates the frame size synchronously, while dynamically displaying the adjusted covered anatomical area. To move the entire scanning frame (e.g., shifting the frame away from the target structure), users can click and drag inside the frame; the frame will move synchronously with the mouse while maintaining its original size. If the automatically generated scanning range perfectly meets the requirements, users can click the "One-Click Confirmation" button on the interface to lock the scanning range directly without additional operation. The entire interaction process requires no complex parameter settings; fine-tuning can be completed simply through basic dragging and clicking actions, significantly reducing operational complexity.

[0063] Finally, after the user completes the fine-tuning or confirmation, the system automatically saves the final scan range parameters and synchronizes them with subsequent scan protocols. At this point, the interface will redirect to the scan preparation completion page, prompting the user to start the formal scan. If the user needs to undo the operation during fine-tuning, the interface also supports a "one-click restore" function to quickly restore the original scan range frame automatically generated by the system, ensuring operational flexibility and error tolerance. The simplified interaction design meets the fine-tuning needs of special clinical scenarios, avoiding the rigidity issues caused by full automation.

[0064] In some preferred embodiments, after the scanning range frame is automatically generated, dedicated fine-tuning handles are provided at the boundaries and corners of the frame; the fine-tuning handles are configured as semantic controls associated with anatomical structures.

[0065] Specifically, after automatically generating the scanning area bounding box, dedicated fine-tuning handles are set at the top, bottom, left, right boundaries, and four corners of the box. It's worth noting that these handles are not ordinary drag points, but semantic controls associated with anatomical structures. Each handle corresponds to a specific anatomical region and can be distinguished from ordinary drag points by a clear icon or color, allowing users to quickly identify the semantic operation entry point. When the mouse pointer hovers over or clicks a fine-tuning handle, a real-time prompt function is immediately triggered. For example, when operating the upper boundary handle, the interface dynamically displays the corresponding anatomical description next to the handle, informing the user that dragging the handle will affect the coverage of the frontal cortex; when dragging the lower boundary handle, relevant prompts regarding cerebellum or medulla oblongata coverage are displayed simultaneously, allowing the user to understand the anatomical impact of the operation before making the adjustment. The prompt information updates in real time with the dragging of the handle; if the adjustment causes a change in coverage, the prompt content will change accordingly, always remaining consistent with the anatomical significance of the current operation. Finally, users can make precise fine-tuning based on the prompts, without relying on personal experience to judge the corresponding anatomical area for the adjustment direction. This effectively avoids the omission of key structures or excessive coverage of irrelevant areas due to operational errors. During the fine-tuning process, the prompts remain clearly visible without obscuring the core anatomical structures in the localization image, ensuring that users can receive semantic guidance and observe the adaptation of the adjusted scan frame to the anatomical structures. After adjustment, clicking the confirmation button locks the final scan range. The entire interaction process, through the combination of semantic controls and real-time prompts, intuitively presents the anatomical decision-making logic of the background to the user, achieving the dual goals of operational guidance and precise localization.

[0066] In some preferred embodiments, after achieving intelligent positioning of the magnetic resonance scan by performing the above steps S11-S13, the method also performs the steps recommended by the protocol:

[0067] First, structured clinical information and anatomical information identified from localization images are extracted from examination request forms to form a contextual data set. Specifically, by connecting to the hospital information system (HIS) and the picture archiving and communication system (PACS), structured clinical information is extracted from examination request forms, including key data such as the patient's clinical indications (e.g., knee osteoarthritis, postoperative follow-up for liver cancer), age, and gender, ensuring accurate capture of the core examination needs. Simultaneously, combined with the anatomical information identified from localization images using a deep learning model, the clinical information and anatomical information are integrated into a complete contextual data set.

[0068] Secondly, a pre-defined protocol library is invoked, and multi-dimensional matching and sorting are performed based on the aforementioned context data set, recommending several top-priority protocols. Specifically, a large pre-defined protocol library (covering standardized protocols for different body parts, diseases, and scanning purposes) is invoked, and multi-dimensional intelligent matching is performed based on context information. For example, anatomical structure and clinical indications are used as core matching dimensions (such as "knee joint + degenerative disease" corresponding to a specific protocol for articular cartilage imaging), while auxiliary dimensions such as patient age (e.g., matching low-dose protocols for pediatric patients) and gender (e.g., matching a specific protocol for female breast scans) are combined to select suitable protocols. Subsequently, a built-in sorting algorithm prioritizes the selected protocols. This algorithm comprehensively considers multiple key indicators, including the degree to which the protocol matches the current examination needs, the frequency of use of the protocol in actual clinical applications, and the efficiency of the corresponding scanning process. Through comprehensive evaluation and calculation of these indicators, 1-3 optimal recommended protocols are finally determined and output, ensuring that users do not need to sift through a massive number of protocols one by one.

[0069] Finally, the key parameters of several recommended protocols are displayed on the interactive interface for comparison. Specifically, after the protocol recommendations are sorted, the interface automatically and clearly displays 1-3 of the best selected protocols. The interface presents protocol information in an intuitive format such as tables or charts, arranging key parameters such as scan time, resolution, contrast, radiation dose, slice thickness, and slice spacing side by side. Clear category labels and numerical displays make the differences in parameters between different protocols immediately apparent, allowing users to quickly grasp the core information. Next, the interface enhances the parameter comparison effect through differentiated display methods, such as using different colors to highlight the advantageous parameters of each protocol, or presenting parameters in order of importance, helping users quickly determine the suitability of different protocols for their current examination needs. For differences in key parameters, the system also provides concise text prompts to supplement explanations, allowing users to understand the clinical significance behind the parameters without additional queries, such as the impact of scan time on patient compliance and the role of resolution in lesion identification. Finally, after determining the appropriate protocol based on the comparison results, users can directly click the "One-Click Application" button corresponding to that protocol. Upon receiving the instruction, the system automatically synchronizes all parameter configurations of the selected protocol to the scanning device, completing the automatic entry and activation of parameters without requiring manual adjustments by the user. If the user needs to change the recommended protocol, they can directly click the application button corresponding to other protocols to switch. The entire operation process is simple and efficient, significantly reducing the time required for protocol selection and configuration.

[0070] For ease of understanding, combined with Figure 4The interface diagram of the intelligent protocol recommendation function is shown below: This interface lists three scanning protocols for different clinical needs. Each protocol entry clearly displays the protocol name, estimated scan time, and core advantage tags. The first protocol is labeled "Fastest Scan Completion," suitable for scenarios requiring rapid examination. The second and third protocols have the same scan time, corresponding to the clinical needs of prioritizing vascular visualization (facilitating clear imaging of vascular structures) and optimal tumor visualization (optimizing tumor lesion identification), respectively. Each protocol has a selection box, allowing users to select the most suitable protocol based on the patient's examination purpose (e.g., emergency rapid scan, vascular disease screening, tumor follow-up, etc.). "Cancel" and "Apply" buttons are located at the bottom of the interface. After selecting, clicking "Apply" instantly configures the parameters of the selected protocol to the scanning device, eliminating the need for manual adjustment of complex parameters. This greatly simplifies the protocol selection and configuration process, allowing users to quickly and accurately match clinical task requirements and significantly improve scanning efficiency.

[0071] In some preferred embodiments, after acquiring the contextual data set and performing preliminary protocol screening, different clinical task objectives are displayed through a protocol accelerator. Each clinical task objective corresponds to a type of examination requirement. Specifically, after acquiring contextual information and performing preliminary protocol screening, instead of directly presenting a complex list of protocols to the user, the interface focuses on core clinical needs and introduces a dedicated protocol recommender module. This module displays different clinical task objectives in the form of clear category cards or buttons, each objective corresponding to a type of core examination requirement, such as optimal tumor visualization, fastest scan completion, and priority for vascular visualization, allowing users to quickly connect with clinical needs without focusing on professional parameters. Users select the most suitable clinical task objective from the protocol recommender based on the current patient's examination request and clinical indications. For example, for follow-up examinations of cancer patients, users can choose the optimal tumor visualization option; for emergency patients, the fastest scan completion option; and for vascular-related examinations, the priority for vascular visualization. During the selection process, the interface supplements the core advantages of each task objective with concise text descriptions, helping users quickly understand the corresponding clinical value. For example, the protocol corresponding to optimal tumor visualization optimizes resolution and contrast to improve lesion identification. Finally, once the user selects a target, a background matching mechanism is immediately triggered. Based on preset association rules, the system automatically retrieves the most suitable standardized protocol from the protocol library for the clinical task objective and pre-configures all parameters. Simultaneously, the interface displays the matching results to the user, including the protocol name and a summary of core parameters, allowing for final confirmation. If the user approves of the recommended protocol, they can directly click the "Apply" button to complete the configuration; if adjustments are needed, minor modifications can be made based on the recommended results. The entire process is driven by the clinical task, significantly reducing the user's cognitive threshold for protocol parameters and improving the accuracy and efficiency of protocol selection.

[0072] Based on the intelligent positioning and protocol recommendation method for magnetic resonance imaging provided above, the following section provides an overall explanation using two specific examples.

[0073] Example 1:

[0074] Step a1) After registering the patient information, the user can accurately select the specific anatomical area to be scanned, such as the head-orbital region.

[0075] Step a2) Enter the dedicated scanning interface, execute the preset scanning positioning protocol, and quickly obtain the positioning image of the part through the magnetic resonance imaging device.

[0076] Step a3) Analyze the localization image using AI algorithms, automatically calculate and output the precise location information of the eye socket in the image, and provide data support for subsequent scanning range planning.

[0077] Step a4) Users can manually fine-tune the scanning range on the interactive interface based on the location information output by the AI ​​algorithm to ensure the accuracy of the positioning.

[0078] Step a5) After confirming that the adjustment results meet clinical needs, click the [Confirm] button to complete the positioning. The scanning range will automatically mark the area to the center.

[0079] Step a6) Click the [Add Protocol] button. A protocol recommendation pop-up will appear. The recommendation pop-up contains protocol features. Since the patient's eye trauma area is large and cannot be maintained for a long time, the protocol with the feature "Complete Scan Fastest" is selected.

[0080] Step a7) Add the protocol to the scan list and perform subsequent scans.

[0081] Example 2:

[0082] Step b1) The user first registers the patient and accurately enters the patient's basic information. Then, based on the clinical examination needs, the user selects the head-cerebral blood vessels as the target scanning area and clarifies the core focus of the scan on the anatomical structures related to the cerebral blood vessels.

[0083] Step b2) Enter the scanning interface, start the preset positioning scanning protocol, and quickly acquire positioning images of the head-brain vascular region through the MRI equipment to provide raw image data for subsequent accurate positioning.

[0084] Step b3) The acquired localization image is automatically analyzed by AI algorithm. Based on the anatomical features of cerebral blood vessels, the key information such as their specific location and contour range in the localization image is accurately calculated and output.

[0085] Step b4) Users can combine actual clinical needs with the anatomical details presented in the positioning image, and make fine manual adjustments to the scanning range on the interactive interface based on the location information output by the algorithm, to ensure that the positioning covers the complete cerebral vascular region.

[0086] Step b5) After confirming that the adjustment is correct, the user clicks the [Confirm] button to complete the positioning, and the scanning range will automatically mark the area to the center.

[0087] Step b6) The user clicks the "Add Protocol" button on the interface. A protocol recommendation pop-up appears, clearly listing several protocols suitable for head-brain vascular scans and their core features. Since the patient's core requirement for the examination is a clear view of the cerebral vascular structure, protocols marked with "Vascular Display Priority" are selected first.

[0088] Step b7) After selecting the protocol, add it to the scan list and complete the parameter pre-configuration. The user can start the formal scan process to ensure that the scan results can accurately meet the clinical diagnostic needs related to cerebrovascular diseases.

[0089] It should be noted that in the embodiments of this application, words such as "exemplary" or "for example" indicate examples, illustrations, or descriptions. Any embodiment or design scheme described as "exemplary" or "for example" in this application should not be construed as being more preferred or advantageous than other embodiments or design schemes. Specifically, the use of words such as "exemplary" or "for example" is intended to present the relevant concepts in a specific manner. In addition, "at least one" means one or more, and "more than one" means two or more. "And / or" describes the relationship between related objects, indicating that there can be three relationships, for example, A and / or B, which can mean: A exists alone, A and B exist simultaneously, or B exists alone, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c can be represented as: a, b, c, ab, ac, bc, or abc, where a, b, and c can be a single or multiple.

[0090] Figure 5 This is a schematic diagram of the structure of an intelligent positioning and protocol recommendation system for magnetic resonance imaging (MRI) scanning according to an embodiment of this application. The intelligent positioning and protocol recommendation system 500 includes: an anatomical structure information extraction module 501, a scanning range box generation module 502, and a range box adjustment module 503.

[0091] The anatomical structure information extraction module 501 is used to acquire the patient's localization image and identify the anatomical structure information in the localization image using a pre-trained deep learning model. The scan bounding box generation module 502 is used to generate at least one suggested scan bounding box on the localization image based on the identified anatomical structure information and according to a preset clinical rule base. The bounding box adjustment module 503 is used to display the generated scan bounding box in the interactive interface and adjust the scan bounding box in response to user operation commands.

[0092] It should be understood that the specific processes by which each module performs the corresponding steps described above have been detailed in the above method embodiments, and will not be repeated here for the sake of brevity. It should also be understood that the module division in the embodiments of this application is illustrative and merely a logical functional division; other division methods may exist in actual implementation. Furthermore, the functional modules in the various embodiments of this application can be integrated into a single processor, exist as separate physical entities, or have two or more modules integrated into one module. The integrated modules described above can be implemented in hardware or as software functional modules.

[0093] Figure 6 This is a schematic block diagram of a computer device provided in an embodiment of this application. Figure 6 As shown, the computer device includes at least one processor 601, a memory 602, at least one network interface 603, and a user interface 605. The various components in the device are coupled together via a bus system 604. It is understood that the bus system 604 is used to implement communication between these components. In addition to a data bus, the bus system 604 also includes a power bus, a control bus, and a status signal bus. However, for clarity, in… Figure 6 The general will label all buses as bus systems.

[0094] The user interface 605 may include a monitor, keyboard, mouse, trackball, clicker, button, touchpad, or touch screen.

[0095] It is understood that memory 602 can be volatile memory or non-volatile memory, or both. Non-volatile memory can be read-only memory (ROM) or programmable read-only memory (PROM), used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as static random access memory (SRAM) and synchronous static random access memory (SSRAM). The memories described in the embodiments of this invention are intended to include, but are not limited to, these and any other suitable categories of memory.

[0096] In this embodiment of the invention, the memory 602 is used to store various types of data to support the operation of the electronic terminal 600. Examples of this data include: any executable program for operation on the electronic terminal 600, such as the operating system 6021 and application program 6022; the operating system 6021 contains various system programs, such as the framework layer, core library layer, driver layer, etc., for implementing various basic services and handling hardware-based tasks. The application program 6022 may contain various applications, such as a media player, browser, etc., for implementing various application services. The intelligent positioning and protocol recommendation method for magnetic resonance scanning provided in this embodiment of the invention can be included in the application program 6022.

[0097] The methods disclosed in the above embodiments of the present invention can be applied to processor 601, or implemented by processor 601. Processor 601 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the integrated logic circuit of the hardware in processor 601 or by instructions in the form of software. The processor 601 may be a general-purpose processor, a digital signal processor (DSP), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Processor 601 can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of the present invention. General-purpose processor 601 may be a microprocessor or any conventional processor, etc. The steps of the accessory optimization method provided in the embodiments of the present invention can be directly reflected as being executed by a hardware decoding processor, or being executed by a combination of hardware and software modules in the decoding processor. The software module may be located in a storage medium, which is located in memory. The processor reads the information in the memory and combines it with its hardware to complete the steps of the aforementioned method.

[0098] In an exemplary embodiment, the electronic terminal 600 may be used by one or more application-specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), or complex programmable logic devices (CPLDs) to perform the aforementioned method.

[0099] According to the method provided in the embodiments of this application, this application also provides a computer program product, which includes: computer program code, which, when run on a computer, causes the computer to execute an intelligent positioning and protocol recommendation method for magnetic resonance scanning.

[0100] According to the method provided in the embodiments of this application, this application also provides a computer-readable storage medium storing program code, which, when run on a computer, causes the computer to perform the above-described method.

[0101] As used in this specification, the terms "component," "module," "system," etc., are used to refer to computer-related entities, hardware, firmware, combinations of hardware and software, software, or software in execution. For example, a component can be, but is not limited to, a process running on a processor, a processor, an object, an executable file, an execution thread, a program, and / or a computer. As illustrated, applications running on computing devices and computing devices can both be components. One or more components may reside in a process and / or an execution thread, and components may be located on a single computer and / or distributed among two or more computers. Furthermore, these components can be executed from various computer-readable media on which various data structures are stored. Components can communicate, for example, via local and / or remote processes based on signals having one or more data packets (e.g., data from two components interacting with another component between a local system, a distributed system, and / or a network, such as the Internet interacting with other systems via signals).

[0102] Those skilled in the art will recognize that the various illustrative logical blocks and steps described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this application.

[0103] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0104] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0105] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0106] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0107] In the above embodiments, the functions of each functional unit can be implemented entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially in the form of a computer program product. A computer program product includes one or more computer instructions (programs). When the computer program instructions (programs) are loaded and executed on a computer, all or part of the flow or function according to the embodiments of this application is generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. Computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., high-density digital video discs, DVDs), or semiconductor media (e.g., solid-state disks, SSDs, etc.).

[0108] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0109] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0110] In summary, this application provides a method, apparatus, system, and medium for intelligent positioning and protocol recommendation in magnetic resonance imaging (MRI) scanning. By transforming traditional manual operation into a "one-click confirmation" or fine-tuning mode, it significantly shortens pre-scan preparation time and effectively improves equipment turnaround time. Simultaneously, relying on a standardized algorithm and rule system, it eliminates human differences between different operators and different equipment, ensuring consistent scan results that meet clinical standards. Its core advantages also lie in significantly lowering the operational threshold and reducing over-reliance on individual user experience, thus lowering training costs for new users and effectively avoiding basic operational errors such as selecting the wrong protocol. Furthermore, the intuitive, intelligent, and efficient interactive interface provided by the system further optimizes the user experience, effectively reducing the workload and psychological pressure on technicians. Therefore, this application effectively overcomes the various shortcomings of existing technologies and has high industrial applicability.

[0111] The above embodiments are merely illustrative of the principles and effects of this application and are not intended to limit this application. Any person skilled in the art can modify or alter the above embodiments without departing from the spirit and scope of this application. Therefore, all equivalent modifications or alterations made by those skilled in the art without departing from the spirit and technical concept disclosed in this application should still be covered by the claims of this application.

Claims

1. An intelligent positioning and protocol recommendation method for magnetic resonance scanning, characterized in that, The method comprises the following steps: obtain a positioning image of a patient, identify anatomical structure information in the positioning image through a pre-trained deep learning model; based on the identified anatomical structure information, generate at least one recommended scan range frame on the positioning image according to a pre-set clinical rule base; display the generated scan range frame in an interactive interface and adjust the scan range frame in response to user operation instructions.

2. The method for intelligent positioning and protocol recommendation for magnetic resonance scan as claimed in claim 1 wherein, The intelligent positioning and protocol recommendation method further comprises: obtain patient clinical information and physiological parameters corresponding to the positioning image of the patient to form multi-modal data with the positioning image of the patient, and perform standardization processing on the multi-modal data; input the standardized positioning image features, clinical information semantic features, and physiological parameter quantitative features into a pre-trained deep learning model to obtain fusion features through a feature fusion module built in the model; The pre-trained deep learning model adjusts the identification strategy according to the fusion features and outputs a precise identification result containing a lesion-related region.

3. The method for intelligent positioning and protocol recommendation for magnetic resonance scan as claimed in claim 1 wherein, The intelligent positioning and protocol recommendation method further comprises: An image enhancement module is added to the input end of the pre-trained deep learning model, which is used to improve the anatomical structure definition of the positioning image through super-resolution reconstruction and noise suppression algorithms.

4. The method for intelligent positioning and protocol recommendation for magnetic resonance scan as claimed in claim 3 wherein, The image enhancement module at the input end of the deep learning model is used to perform the following: perform noise suppression processing on the input positioning image raw data to suppress noise caused by low-dose scanning of the positioning image; use a super-resolution algorithm based on a generative adversarial network or a convolutional neural network to perform super-resolution reconstruction on the positioning image after noise suppression processing.

5. The method for intelligent positioning and protocol recommendation for magnetic resonance scan as claimed in claim 1 wherein, The intelligent positioning and protocol recommendation method further comprises: After automatically generating the scan range frame, exclusive fine-tuning handles are arranged at the boundaries and corner points of the frame body; the fine-tuning handles are configured as semanticized controls associated with anatomical structures.

6. The method for intelligent positioning and protocol recommendation for magnetic resonance scan as claimed in claim 1 wherein, After generating and adjusting the scan range frame, the intelligent positioning and protocol recommendation method further performs the steps of protocol recommendation, which comprises: extract structured clinical information from the examination application form and anatomical structure information identified from the positioning image to form a context data set; call a pre-set protocol library, perform multi-dimensional matching and sorting based on the context data set, and recommend a number of optimal protocols therefrom; display the key parameters of the recommended protocols in the interactive interface for comparison.

7. The method for intelligent positioning and protocol recommendation for magnetic resonance scan as claimed in claim 6 wherein, After completing the context data set acquisition and protocol screening, the intelligent positioning and protocol recommendation method further performs the following: display different clinical task targets through a protocol promoter, each clinical task target corresponding to a type of examination demand.

8. An intelligent positioning and protocol recommendation system for magnetic resonance scans, characterized in that The method comprises the following steps: an anatomical structure information extraction module for obtaining a positioning image of a patient, identifying anatomical structure information in the positioning image through a pre-trained deep learning model; a scan range frame generation module for generating at least one recommended scan range frame on the positioning image based on the identified anatomical structure information according to a pre-set clinical rule base; a range frame adjustment module for displaying the generated scan range frame in an interactive interface and adjusting the scan range frame in response to user operation instructions.

9. A computer-readable storage medium having stored thereon a computer program, characterized in that, When the computer program is executed by the processor, it implements the intelligent positioning and protocol recommendation method for magnetic resonance scanning as described in any one of claims 1 to 7.

10. A computer apparatus comprising a memory, a processor, and a computer program stored on the memory, wherein the computer program, when executed by the processor, causes the processor to perform the method of any one of claims 1 to 9. The processor executes the computer program to implement the intelligent positioning and protocol recommendation method for magnetic resonance scanning as described in any one of claims 1 to 7.

Images