A portable magnetic resonance imaging device-based brain disease remote diagnosis system

Abstract

The application provides a portable magnetic resonance imaging-based brain disease remote diagnosis system, and belongs to the cross field of medical imaging and remote medical treatment.The system comprises a portable magnetic resonance imaging device, a remote data transmission and cooperation network and a remote diagnosis and AI auxiliary workstation; wherein the portable magnetic resonance imaging device is arranged in a primary medical institution, is used for performing standardized scanning on a patient's brain disease under a low-field magnetic field, and acquires brain image data of the patient; the remote data transmission and cooperation network is used for connecting the device end and the diagnosis end, and is used for transmitting the brain image data to the remote diagnosis and AI auxiliary workstation; and the remote diagnosis and AI auxiliary workstation is used for performing brain disease analysis based on the brain image data and generating a structured report, and the report is returned to the primary medical institution.The application combines remote medical treatment and on-site magnetic resonance scanning, optimizes medical resources, and improves diagnosis effect.

Description

Technical Field

[0001] This invention belongs to the interdisciplinary field of medical imaging and telemedicine, and specifically relates to a remote diagnostic system for brain diseases based on a portable magnetic resonance imaging device. Background Technology

[0002] The brain is the core of the human nervous system, and brain lesions have a serious impact on human movement and consciousness. Monitoring brain diseases has always been a priority. Due to the unique structure of the brain, the diagnosis of brain diseases relies heavily on expert experience and the ever-evolving medical imaging technology. However, expert experience demands a high level of personal expertise and skill, while medical imaging relies on expensive and bulky equipment and requires highly skilled operators. In some economically underdeveloped areas, there is a severe lack of corresponding medical resources, failing to meet the needs of patients. Magnetic Resonance Imaging (MRI) is an indispensable tool for diagnosing brain diseases. Its superior soft tissue contrast and lack of ionizing radiation make it crucial in the diagnosis and evaluation of diseases such as stroke, brain tumors, and neurodegenerative diseases. However, traditional high-field (typically 1.5T or 3.0T) superconducting MRI systems used in clinical practice have many inherent limitations: the equipment is bulky, often weighing over 5 tons, and must be permanently installed in a dedicated radiology room with strict electromagnetic shielding and special power supply and cooling facilities. The high cost of purchasing a single MRI unit and the substantial examination fees result in a high concentration of medical resources, making it difficult for primary healthcare institutions to adopt them. With advancements in permanent magnet technology, low-field imaging algorithms, and electronic technology, mobile, low-field (typically below 0.5T) MRI equipment has become an important development direction.

[0003] In modern healthcare systems, remote diagnosis can, to some extent, address the shortage of medical resources, allowing them to transcend geographical limitations and integrate resources across multiple regions. This is particularly beneficial for remote areas, underdeveloped healthcare systems, and patients with mobility issues, improving both the quality of care and service capabilities. However, due to the on-site nature of medical imaging equipment, such as MRI, which requires patient presence, large-scale equipment cannot be used in remote diagnosis. Furthermore, for simpler image transmissions, the lack of standardized procedures and deeply integrated diagnostic tools hinders efficiency. For time-sensitive emergencies like stroke, there is a "golden window" for treatment (e.g., thrombolytic therapy for ischemic stroke typically needs to be performed within 6 hours of symptom onset). The process of transferring critically ill patients to the radiology department is not only time-consuming but also carries significant life-threatening risks. Although there have been some attempts at mobile MRI and remote diagnosis, for example, Chinese patent CN107296588B provides a remote medical consultation system integrating vehicle-mounted MRI, which places the MRI equipment in a special vehicle and achieves interaction with remote units through wireless communication. However, this system is essentially still a "large mobile platform", which has limited deployment flexibility and scenario universality, and has not deeply solved the inherent image quality and diagnostic efficiency problems of low-field MRI. Summary of the Invention

[0004] In view of the above-mentioned defects or deficiencies in the prior art, the present invention aims to provide a remote diagnostic system for brain diseases based on portable magnetic resonance imaging (MRI). This system uses a portable MRI device to perform bedside scanning on patients, rapidly and securely transmitting the bedside images to remote experts. It also integrates intelligent analysis tools to assist in rapid decision-making, forming a complete system. The low-field brain images obtained from the bedside scan are used for remote diagnosis of patients. By integrating resources from multiple parties through the remote diagnostic system, the system improves the efficiency of medical resource utilization and enhances the diagnostic effectiveness for brain diseases. It is particularly suitable for rapid bedside imaging, real-time remote consultation, and intelligent diagnostic analysis for acute and critical neurological conditions such as stroke and brain injury, thereby achieving the decentralization and efficient utilization of medical imaging resources.

[0005] To achieve the above objectives, the embodiments of the present invention adopt the following technical solutions:

[0006] This invention provides a remote diagnostic system for brain diseases based on portable magnetic resonance imaging (MRI). The system includes: a portable MRI device, a remote data transmission and collaboration network, and a remote diagnostic and AI-assisted workstation; wherein,

[0007] The portable magnetic resonance imaging device is deployed in primary healthcare institutions to perform standardized scans of patients' brains under low-field magnetic fields and obtain brain imaging data.

[0008] The remote data transmission and collaboration network is used to connect the device end and the diagnostic end, and transmit brain image data to the remote diagnostic and AI-assisted workstation;

[0009] The remote diagnostic and AI-assisted workstation is used to analyze brain diseases based on brain imaging data and generate structured reports, which are then transmitted back to primary healthcare institutions.

[0010] In a preferred embodiment of the present invention, the portable magnetic resonance imaging device is designed as an integrated, mobile cart-type or compact floor-standing device, including a low-field permanent magnet subsystem, an integrated spectrometer and gradient subsystem, and an electron power supply system; wherein...

[0011] The low-field permanent magnet subsystem uses high-performance rare-earth permanent magnet materials to form a magnet array, generating a low-field main magnetic field B0 of 0.1T~0.2T within a preset volume and weight.

[0012] The integrated spectrometer and gradient subsystem employs an integrated digital spectrometer and gradient coil to generate precisely controlled gradient magnetic fields and radio frequency pulses.

[0013] The power supply system is used to power the device via an external power source or battery.

[0014] In a preferred embodiment of the present invention, the preset weight is less than 900 kg.

[0015] As a preferred embodiment of the present invention, the remote data transmission and collaboration network includes a multimodal transmission channel and a cloud platform service module;

[0016] The multimodal transmission channel has a built-in intelligent sensing and scheduling engine, which is used to integrate the communication links of 5G cellular network, wired broadband and satellite communication, and monitor and adjust the signal quality, bandwidth and latency of each link in real time.

[0017] The cloud platform service module is used to intelligently receive and standardize brain imaging data after it is uploaded to the cloud, control access and collaborative workflow, and centrally monitor and manage system resources.

[0018] In a preferred embodiment of the present invention, when the cloud platform service module performs intelligent reception and standardized processing of brain image data, it first performs integrity verification and automatic reassembly on the uploaded brain image data. After ensuring that there are no errors, it adds a globally unique identifier and stores it in a secure cloud storage pool. At the same time, it automatically extracts and indexes key metadata, laying the foundation for subsequent rapid retrieval and access.

[0019] In a preferred embodiment of the present invention, when the cloud platform service module is used to control access and collaborative workflow, once the image is ready, it automatically or semi-automatically assigns the inspection task to the corresponding remote expert or AI-assisted diagnostic workstation according to preset rules; after the expert passes the security authentication, he can access the image through a webpage or a dedicated client within the authorized time. All access, annotation and diagnostic operations are recorded in detail by the audit log.

[0020] In a preferred embodiment of the present invention, the remote diagnostic and AI-assisted workstation includes an image noise reduction module, an image segmentation module, an expert interaction module, an AI-assisted analysis module, and a structured report generation module; wherein...

[0021] The image noise reduction module is used to suppress the interference of external environmental noise on magnetic resonance data at the original signal level, and to perform noise cancellation on a voxel-by-voxel and echo-by echo basis on the original MRI images; the image segmentation module has a model for image segmentation, which segments the uploaded image, delineates the lesion boundary, and provides quantitative volume measurement.

[0022] The expert interaction module adopts an integrated medical digital imaging and communication system browser, which supports multi-plane reconstruction, window width and window level adjustment, basic measurement functions, as well as professional functions such as MPR analysis, MIP analysis, and AI tool call. After interacting with experts, it generates expert interaction records.

[0023] The AI-assisted analysis module is used to perform preliminary analysis of brain imaging data based on the invoked AI tools and generate AI analysis results;

[0024] The structured report generation module is used to automatically generate a structured draft preliminary diagnostic report based on expert interaction records and AI analysis results, and supports the review and issuance process. After expert review and confirmation, the report is sent back to the on-site doctor at the portable magnetic resonance imaging device and simultaneously archived to the relevant hospital's image archiving and communication system PACS.

[0025] In a preferred embodiment of the present invention, the image denoising module includes: a multi-channel environmental noise acquisition submodule, a noise field time-varying model construction submodule, and a noise suppression submodule; wherein,

[0026] The multi-channel environmental noise acquisition submodule is used to acquire environmental noise signals in real time during the scanning process, including but not limited to low-frequency to mid-frequency noise such as vibration of the equipment during the scanning process, external electromagnetic interference, mechanical operation noise, and personnel activities in the operating room.

[0027] The noise field time-varying model construction submodule is used to construct a time-varying noise field model based on multi-channel environmental noise. By analyzing the correlation between noise and magnetic resonance echo signal, it dynamically estimates the noise components mixed into the data channel under the current scanning environment.

[0028] The noise suppression submodule is used to perform voxel-by-voxel and echo-by echo noise cancellation on the original MRI image based on the noise estimation results of the time-varying noise field model.

[0029] In a preferred embodiment of the present invention, the image segmentation model is built based on the U-Net or Transformer architecture. In another preferred embodiment of the present invention, the interactive interface, after expert login, is divided into three panels: the left side displays a list of tasks to be processed, the middle side displays an image browsing and operation area, and the right side displays a report editing area and an AI tool access area.

[0030] As a preferred embodiment of the present invention, the system further includes an edge computing module deployed in primary healthcare institutions and used in conjunction with the portable magnetic resonance imaging device. The edge computing module has a built-in lightweight AI model for real-time quality assessment and control of brain imaging data before data upload.

[0031] The technical solutions provided in the embodiments of the present invention have the following beneficial effects:

[0032] It has achieved true bedside real-time diagnosis: transforming heavy, fixed MRI equipment into lightweight, portable examination tools that can be pushed directly to the patient's bedside, greatly reducing the risks and time delays in the transfer of critically ill patients, and making it a reality for equipment to find patients;

[0033] Diagnostic-grade image quality is ensured in open environments: cloud-based AI image enhancement effectively overcomes the imaging disadvantages in low-field and unshielded environments, and the obtained image quality meets the clinical diagnostic needs of acute brain diseases (such as stroke and trauma).

[0034] An efficient closed-loop remote diagnostic process has been constructed: it seamlessly integrates rapid imaging, intelligent transmission, AI pre-analysis, multi-party collaborative consultation and report generation, which greatly shortens the entire process time from the completion of the scan to obtaining an authoritative diagnostic opinion;

[0035] It promotes the widespread accessibility of high-quality medical resources: the system reduces the barriers to MRI examinations in terms of location, cost, and operation, enabling primary hospitals, community health centers, and even pre-hospital emergency units to conduct high-quality brain MRI examinations and obtain technical support from higher-level hospitals through remote connections. It is an effective technical means to promote hierarchical medical treatment.

[0036] Of course, implementing any product or method of the present invention does not necessarily require achieving all of the advantages described above at the same time. Attached Figure Description

[0037] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0038] Figure 1 This is a schematic diagram of the structure of the remote diagnosis system for brain diseases based on portable magnetic resonance imaging as described in an embodiment of the present invention;

[0039] Figure 2 This is a schematic diagram of the structure of a portable magnetic resonance imaging device according to an embodiment of the present invention;

[0040] Figure 3 yes Figure 2 The diagram shows the operation of a portable magnetic resonance imaging device. Detailed Implementation

[0041] After discovering the aforementioned problems, the inventors of this application conducted a detailed study on existing brain magnetic resonance imaging (MRI) scanning and remote diagnostic methods. The study found that traditional high-field MRI systems cannot be quickly deployed to emergency rooms, intensive care units (ICUs), primary hospitals, or special sites (such as disaster relief), failing to meet the need for immediate bedside imaging within the critical care "time window," resulting in insufficient accessibility and flexibility. Meanwhile, existing portable low-field MRI systems are susceptible to electromagnetic interference in complex environments and have an inherently low signal-to-noise ratio, potentially leading to numerous image artifacts and unclear details, affecting the diagnostic accuracy and confidence of remote physicians. Furthermore, existing remote consultation systems lack deep integration in terms of imaging equipment, real-time data transmission, multi-party collaborative image reading, and intelligent auxiliary analysis, resulting in a lengthy process from imaging to obtaining authoritative diagnostic opinions, failing to fully leverage the synergistic advantages of "portability + remoteness."

[0042] It should be noted that the defects in the above-mentioned prior art solutions are all the result of the inventors' practice and careful research. Therefore, the discovery process of the above problems and the solutions proposed by the embodiments of the present invention in the following text should be the inventors' contributions to the present invention.

[0043] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. It should be noted that, without conflict, the embodiments and features in the embodiments of the present invention can also be combined with each other.

[0044] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. In the description of this invention, the terms "first," "second," "third," "fourth," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0045] To address the integration issues of existing medical imaging and remote diagnosis, this invention provides a remote diagnostic system for brain diseases based on a portable magnetic resonance imaging (MRI) device. This system overcomes the shortcomings of existing MRI systems in terms of accessibility and uneven distribution of diagnostic resources. The system integrates a high-performance portable low-field MRI device, anti-interference imaging technology, a high-speed and secure data transmission network, and a cloud-based artificial intelligence (AI)-assisted diagnostic platform. It enables rapid, high-quality brain imaging at the patient's location and generates diagnostic reports instantly through remote expert consultation and AI analysis. This significantly shortens the diagnosis time for acute and critical brain diseases and improves medical service capabilities in primary care and special scenarios.

[0046] like Figure 1 As shown, the remote diagnosis system for brain diseases based on portable magnetic resonance imaging equipment includes: a portable magnetic resonance imaging device, a remote data transmission and collaboration network, and a remote diagnosis and AI-assisted workstation;

[0047] The portable magnetic resonance imaging device is deployed in primary healthcare institutions to perform standardized scans of patients' brains under low-field magnetic fields, acquiring brain imaging data. It is the core of the system's front-end data acquisition, breaking through the site limitations of traditional MRI.

[0048] like Figure 2As shown, the portable magnetic resonance imaging (MRI) device is designed as an integrated, mobile cart-type or compact floor-standing unit, comprising a low-field permanent magnet subsystem, an integrated spectrometer and gradient subsystem, and a power supply system. The low-field permanent magnet subsystem uses high-performance rare-earth permanent magnet materials (such as neodymium iron boron) to construct an optimized magnet array, generating a low-field main magnetic field (B0) of 0.1T~0.2T within a preset volume and weight (target ≤900 kg). Through magnet geometry design and passive / active shimming technology, the uniformity of the magnetic field within the imaging area is ensured to meet clinical diagnostic requirements. The integrated spectrometer and gradient subsystem employs a highly integrated digital spectrometer and lightweight, high-efficiency gradient coils to generate precisely controlled gradient magnetic fields and radio frequency pulses. The power supply system can be driven by standard 220V / 10A AC mains power, with a total power consumption of less than 3kW, requiring no liquid helium cooling or dedicated power distribution. Preferably, the device integrates a local storage and data transmission module for a Digital Imaging and Communications in Medicine (DICOM) system. For example, it features a pre-installed 4G / 5G slot and gigabit Ethernet port, along with a built-in algorithm encryption chip to ensure secure data transmission. The power supply system is used to power the device via an external power source or battery.

[0049] The remote data transmission and collaboration network is used to securely and efficiently connect portable magnetic resonance imaging equipment and remote diagnostic and AI-assisted workstations, enabling communication between the equipment and the diagnostic end. The network structure of the remote data transmission and collaboration network includes a multimodal transmission channel and a cloud platform service module.

[0050] The multimodal transmission channel incorporates an intelligent sensing and scheduling engine to integrate communication links from 5G cellular networks, wired broadband, and satellite communications, monitoring and adjusting the signal quality, bandwidth, and latency of each link in real time. For example, in complex environments such as remote areas, the system can automatically perform seamless switching or multi-link aggregation transmission based on preset strategies and real-time network conditions. When uploading large-capacity medical images, it can simultaneously aggregate broadband and 5G networks to accelerate transmission, thereby ensuring reliable, continuous, and efficient transmission of critical data.

[0051] The cloud platform service module is used for the intelligent management, processing, and collaborative scheduling of brain imaging data throughout its entire lifecycle after it is uploaded to the cloud. Specifically, the cloud platform service module is used for intelligent data reception and standardized processing, controlled access and collaborative workflows, and centralized monitoring and policy management of system resources.

[0052] During the intelligent reception and standardized processing of data, the uploaded brain imaging data is first verified for integrity and automatically reassembled. After ensuring that it is error-free, a globally unique identifier is added and stored in a secure cloud storage pool. At the same time, key metadata (such as examination type, patient ID, and scan parameters) is automatically extracted and indexed to lay the foundation for subsequent rapid retrieval and access.

[0053] The cloud platform service module implements refined role and permission management to control access and collaborative workflows. Once images are ready, the examination task is automatically or semi-automatically assigned to the appropriate remote experts or AI-assisted diagnostic workstations based on preset rules (such as patient location and disease specialty). After security authentication, experts can access images via a webpage or dedicated client within authorized timeframes. All access, annotation, and diagnostic operations are recorded in detailed audit logs.

[0054] When centrally monitoring and managing system resources, the cloud platform service module provides system administrators with a unified dashboard that can monitor the status of all online devices, network transmission quality, cloud resource utilization, and diagnostic task queues in real time. Administrators can dynamically adjust global parameters such as network transmission policies and user access permissions to achieve centralized and intelligent operation and maintenance of the entire system.

[0055] The remote diagnostic and AI-assisted workstation is used to connect experts and perform brain disease analysis and generate structured reports based on brain imaging data, which are then transmitted back to primary healthcare institutions.

[0056] Specifically, the remote diagnostic and AI-assisted workstation includes an image noise reduction module, an image segmentation module, an expert interaction module, an AI-assisted analysis module, and a structured report generation module.

[0057] The image denoising module is used to suppress the interference of external environmental noise on magnetic resonance data at the original signal level, thereby improving the inherent signal-to-noise ratio of low-field-strength equipment. Since portable magnetic resonance imaging equipment (0.1 T–0.2 T) is often transported between different clinical environments (emergency room, ICU, mobile screening vehicle, etc.), the environmental noise characteristics change drastically, so it is necessary to perform denoising processing on the acquired images.

[0058] The image denoising module includes a multi-channel environmental noise acquisition submodule, a noise field time-varying model construction submodule, and a noise suppression submodule. The multi-channel environmental noise acquisition submodule is used to acquire environmental noise signals in real time during the scanning process, including but not limited to low- to mid-frequency noise such as vibrations during equipment scanning, external electromagnetic interference, mechanical operating noise, and personnel activity in the operating room. This can be achieved by integrating dedicated noise acquisition coils at key locations of the MRI equipment (such as gradient coils, radio frequency shielding layers, patient beds, and equipment casings), working synchronously with MRI image acquisition. The noise field time-varying model construction submodule is used to construct a time-varying noise field model based on the multi-channel environmental noise. By analyzing the correlation (including linear and nonlinear relationships) between noise and MRI echo signals, it dynamically estimates the noise components mixed into the data channels under the current scanning environment. The noise suppression submodule is used to perform voxel-by-voxel and echo-by-echo noise cancellation on the original MRI image based on the noise estimation results of the noise field time-varying model. Adaptive filtering algorithms (such as Normalized Least Mean Squares (NLMS) or Recursive Least Squares (RLS)) or lightweight deep learning models (such as Convolutional Recurrent Networks (CRN)) can be used for noise suppression. The image denoising module reduces the contribution of environmental noise by 10–20 dB without significantly attenuating the real magnetic resonance signal, significantly improving the signal-to-noise ratio of the original data and outputting "preliminary cleaned" data. At this point, the environmental noise in the data has been greatly suppressed, and the deep learning model can focus on improving resolution, contrast and structural details, laying a clean and reliable data foundation for subsequent image segmentation and AI-assisted analysis modules.

[0059] The image segmentation module receives the denoised image. This method achieves pixel-level precision segmentation of brain lesions (such as tumors and stroke lesions) through adversarial training. The network architecture employs a modified Seg-GLGAN architecture, which includes a segmentation generator G. seg and a context discriminator D context Generator G seg Based on 3D U-Net, an Atrous Spatial Pyramid Pooling (ASPP) module is embedded between the encoder and decoder to capture multi-scale contextual information and generate a segmentation probability map S. pred = G seg (I enhanced ) Context discriminator D context It includes parallel global feature extractors and local feature extractors. The global feature extractor uses the entire ground truth label map S. gt Or predict segmentation map S pred As input, the output is the global feature vector f. globalThe local feature extractor takes the region of interest (ROI) ϕ(S) as input and outputs a local feature vector f. global This forces the network to focus on the fine structure of lesion boundaries. The total loss function combines pixel-level loss with adversarial loss.

[0060] (1)

[0061] In equation (1), the pixel-level loss L pixel The combination of Dice loss and weighted cross-entropy loss is defined as:

[0062] (2)

[0063] In equation (2), p i To predict the probability that the i-th pixel belongs to a lesion, g i The true label is 0 or 1, N is the total number of pixels, and w pos These are positive sample weights, used to alleviate class imbalance problems.

[0064] The expert interaction module employs an integrated professional DICOM browser to support basic functions such as multi-planar reconstruction, window width and level adjustment, and measurement, as well as professional functions such as MPR, MIP analysis, and AI tool invocation. It generates expert interaction records after interaction with experts. In one executable embodiment, the interactive interface, after expert login, is divided into three panels: a list of tasks to be processed on the left, an image browsing and operation area in the middle, and a report editing area and AI tool invocation area on the right. The AI-assisted analysis module is used to perform preliminary analysis of brain image data based on the invoked AI tools and generate AI analysis results.

[0065] The structured report generation module is based on a large language model with feature projection. This module transforms the structured data (lesion location, volume, signal features) output by the segmentation network into a diagnostic report described in natural language. First, it converts the quantified features output by the segmentation network (including anatomical localization L based on standard brain atlas, volume measurement value V) into a diagnostic report. ml And the signal intensity ratio R with the contralateral normal tissue signal The feature vector sequence T is obtained by projecting it onto the word embedding space of a large language model (LLM) through a lightweight multilayer perceptron mapping network P. feat =P(L,V ml ,R signalThe prompt word constraint generation adopts a template-based structured prompt strategy, concatenating feature vectors with system prompt words and inputting them into the LLM (Limited Language Management). Under this constraint, the LLM generates text, ensuring the report includes key image findings and generating a preliminary diagnostic prompt such as "Consider...possible". The structured report generator automatically fills the analysis results into standardized templates, generating screening reports containing key image findings, lesion descriptions (location, size), and preliminary prompts, providing direct decision support for primary care physicians. The screening report templates provide multiple styles and selectable standardized templates. Based on expert interaction records and AI analysis results, this module automatically generates a structured draft preliminary diagnostic report, supporting review and issuance processes. After expert review and confirmation, it is transmitted back to the on-site physician within seconds and simultaneously archived to the relevant hospital's Picture Archiving and Communication System (PACS).

[0066] The workstation can be developed using technologies such as HTML5 and WebGL, and can run in mainstream browsers without the need for plugins.

[0067] The remote diagnostic and AI-assisted workstation is a web-based collaborative diagnostic workstation that provides remote experts with an interface for image analysis, AI tool access, and report writing. The web interface offers remote experts a client interface directly accessible via a browser, providing advanced post-processing functions such as multi-dimensional image browsing, 3D reconstruction, and diffusion analysis. It also deeply integrates various AI-assisted analysis plugins, such as automatic lesion delineation and automatic hematoma volume calculation. From a device perspective, the remote diagnostic and AI-assisted workstation is a cloud platform. Experts can securely log in to the cloud platform via personal computers, workstations, or other devices to access their dedicated task lists and collaborative diagnostic workstations, enabling efficient remote diagnostic work.

[0068] In one executable embodiment, the system further includes an edge computing module deployed in primary healthcare institutions and used in conjunction with the portable magnetic resonance imaging device. The edge computing module has a built-in lightweight AI model for real-time quality assessment and control of the brain imaging data before data upload. For example, it can detect whether there are serious motion artifacts in the brain imaging data in real time before upload and immediately alert the operator to ensure data quality from the source.

[0069] In one executable embodiment, the system may further include a blockchain evidence storage module, used to generate immutable evidence for key operations (such as data upload, report generation, and modification) using blockchain technology, for auditing and dispute resolution. When a report is issued on the platform, the module packages the report's hash value, issuer, issuance timestamp, and other information into a block and broadcasts it to other nodes on the chain (such as regulatory agencies and third-party certification centers) for evidence storage.

[0070] When using the remote diagnosis system for brain diseases based on the portable magnetic resonance imaging device for remote diagnosis, the portable magnetic resonance imaging device is first deployed in primary medical institutions based on a cloud platform, and an edge computing module is built into the device.

[0071] like Figure 3 As shown, in primary healthcare institutions, patients are scanned using a portable magnetic resonance imaging (MRI) device according to a standard protocol to acquire brain imaging data. For example, according to a preset "rapid screening for brain tumors and stroke" protocol, the relevant T1WI, T2WI, and FLAIR sequence scans are automatically executed, with a total time of approximately 5 minutes. When an edge computing module is built-in, the edge computing module performs real-time quality assessment and control of the brain imaging data based on a built-in lightweight AI model, evaluating the signal-to-noise ratio, detecting the presence of severe motion artifacts in the images, and providing immediate prompts to the operator. Images that meet the quality assessment and control requirements are preprocessed, such as encryption and desensitization (hiding the patient's name and ID number, retaining only the scan ID, age, gender, and other necessary clinical information), forming a secure data packet, which is then sent to a remote data transmission and collaboration network.

[0072] The remote data transmission and collaboration network receives and stores the data packet, creates a diagnostic task based on the brain imaging data, and assigns the diagnostic task to at least one remote diagnostic and AI-assisted workstation.

[0073] At each remote diagnostic and AI-assisted workstation, the image segmentation module segments the received brain image data and sends it to the expert interaction module. Experts logged into the workstation, upon receiving the diagnostic task, review and analyze the images based on the task and the segmented brain image data. They then use the AI-assisted analysis module integrated into the workstation to identify and label lesions. In the structured report template, they select the corresponding imaging features and input diagnostic opinions and suggestions. The structured report generation module generates a diagnostic report based on the expert's feedback and the AI ​​analysis results. The workstation then sends the diagnostic report back to the corresponding primary healthcare institution and updates the task status to "issued." Upon receiving the diagnostic results, the primary healthcare institution's community doctor explains the situation to the patient and completes the referral procedures based on the remote diagnostic report.

[0074] In this embodiment, each module is implemented using a processor, with additional memory added as needed for storage. The processor can be, but is not limited to, a microprocessor (MPU), a central processing unit (CPU), a network processor (NP), a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), other programmable logic devices, discrete gates, transistor logic devices, discrete hardware components, etc. The memory can include random access memory (RAM) or non-volatile memory (NVM), such as at least one disk storage device. Optionally, the memory can also be at least one storage device located remotely from the aforementioned processor.

[0075] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the 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.

[0076] As can be seen from the above technical solutions, the remote diagnosis system for brain diseases based on portable magnetic resonance imaging provided by this invention effectively extends MRI diagnostic capabilities to the grassroots level, which cannot be covered by traditional fixed equipment, through a portable and mobile acquisition terminal. This truly constructs a new model of "mobile scanning + cloud diagnosis," achieving flexible full-domain coverage of diagnostic capabilities. The intelligent task scheduling engine accurately matches suitable cases with suitable experts, optimizing the allocation of expert resources and solving the information asymmetry problem of "experts not finding cases, and cases not finding experts," thus improving the utilization efficiency of high-quality medical resources. The powerful AI-assisted tools integrated in the cloud act as an "external brain" for experts, effectively reducing missed diagnoses and providing quantitative analysis support. Standardized processes and reports also improve the consistency and speed of diagnosis, enhancing the accuracy and efficiency of remote diagnosis. Through technologies such as data anonymization, encrypted transmission, blockchain notarization, and federated learning, while realizing the flow and mining of data value, patient privacy and data security are maximized, complying with medical regulations and constructing a secure and reliable medical data collaboration environment.

[0077] The above description is merely a preferred embodiment of the present invention and an explanation of the technical principles employed, and is not intended to limit the scope of the claimed invention, but merely to illustrate preferred embodiments of the invention. Those skilled in the art should understand that the scope of the invention is not limited to the specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the inventive concept. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

Claims

1. A remote diagnostic system for brain diseases based on portable magnetic resonance imaging, characterized in that, The system includes: a portable magnetic resonance imaging device, a remote data transmission and collaboration network, and a remote diagnostic and AI-assisted workstation; wherein... The portable magnetic resonance imaging device is deployed in primary healthcare institutions to perform standardized scans of patients' brains under low-field magnetic fields and obtain brain imaging data. The remote data transmission and collaboration network is used to connect the device end and the diagnostic end, and transmit brain image data to the remote diagnostic and AI-assisted workstation; The remote diagnostic and AI-assisted workstation is used to analyze brain diseases based on brain imaging data and generate structured reports, which are then transmitted back to primary healthcare institutions.

2. The system according to claim 1, characterized in that, The portable magnetic resonance imaging device is designed as an integrated, mobile cart-type or compact floor-standing unit, comprising a low-field permanent magnet subsystem, an integrated spectrometer and gradient subsystem, and an electron power supply system; wherein... The low-field permanent magnet subsystem uses high-performance rare-earth permanent magnet materials to form a magnet array, generating a low-field main magnetic field B0 of 0.1T~0.2T within a preset volume and weight. The integrated spectrometer and gradient subsystem employs an integrated digital spectrometer and gradient coil to generate precisely controlled gradient magnetic fields and radio frequency pulses. The power supply system is used to power the device via an external power source or battery.

3. The system according to claim 2, characterized in that, The preset weight is less than 900 kg.

4. The system according to claim 1, characterized in that, The remote data transmission and collaboration network includes a multimodal transmission channel and a cloud platform service module; The multimodal transmission channel has a built-in intelligent sensing and scheduling engine, which is used to integrate the communication links of 5G cellular network, wired broadband and satellite communication, and monitor and adjust the signal quality, bandwidth and latency of each link in real time. The cloud platform service module is used to intelligently receive and standardize brain imaging data after it is uploaded to the cloud, control access and collaborative workflow, and centrally monitor and manage system resources.

5. The system according to claim 4, characterized in that, When the cloud platform service module intelligently receives and standardizes brain imaging data, it first performs integrity verification and automatic reconstruction on the uploaded brain imaging data to ensure that it is error-free. Then, it adds a globally unique identifier and stores it in a secure cloud storage pool. At the same time, it automatically extracts and indexes key metadata, laying the foundation for subsequent rapid retrieval and access.

6. The system according to claim 4, characterized in that, When the cloud platform service module is used to control access and collaborative workflow, once the images are ready, it automatically or semi-automatically assigns inspection tasks to the corresponding remote experts or AI-assisted diagnostic workstations according to preset rules. After the experts pass the security authentication, they can access the images through a webpage or a dedicated client within the authorized time. All access, annotation, and diagnostic operations are recorded in detailed audit logs.

7. The system according to claim 1, characterized in that, The remote diagnostic and AI-assisted workstation includes an image noise reduction module, an image segmentation module, an expert interaction module, an AI-assisted analysis module, and a structured report generation module; among which, The image noise reduction module is used to suppress the interference of external environmental noise on magnetic resonance data at the original signal level, and to perform noise cancellation on a voxel-by-voxel and echo-by echo basis on the original MRI images. The image segmentation module contains a model for image segmentation, which segments the uploaded image, delineates the lesion boundary, and provides quantitative volume measurement. The expert interaction module adopts an integrated medical digital imaging and communication system browser, which supports multi-plane reconstruction, window width and window level adjustment, basic measurement functions, as well as professional functions such as MPR analysis, MIP analysis, and AI tool call. After interacting with experts, it generates expert interaction records. The AI-assisted analysis module is used to perform preliminary analysis of brain imaging data based on the invoked AI tools and generate AI analysis results; The structured report generation module is used to automatically generate a structured draft preliminary diagnostic report based on expert interaction records and AI analysis results, and supports the review and issuance process. After expert review and confirmation, the report is sent back to the on-site doctor at the portable magnetic resonance imaging device and simultaneously archived to the relevant hospital's image archiving and communication system PACS.

8. The system according to claim 7, characterized in that, The image denoising module includes: a multi-channel environmental noise acquisition submodule, a noise field time-varying model construction submodule, and a noise suppression submodule; wherein... The multi-channel environmental noise acquisition submodule is used to acquire environmental noise signals in real time during the scanning process, including but not limited to low-frequency to mid-frequency noise such as vibration of the equipment during the scanning process, external electromagnetic interference, mechanical operation noise, and personnel activities in the operating room. The noise field time-varying model construction submodule is used to construct a time-varying noise field model based on multi-channel environmental noise. By analyzing the correlation between noise and magnetic resonance echo signal, it dynamically estimates the noise components mixed into the data channel under the current scanning environment. The noise suppression submodule is used to perform voxel-by-voxel and echo-by echo noise cancellation on the original MRI image based on the noise estimation results of the time-varying noise field model.

9. The system according to claim 7, characterized in that, After logging in, the interactive interface is divided into three panels: the left side is the list of tasks to be processed, the middle is the image browsing and operation area, and the right side is the report editing area and the AI ​​tool access area.

10. The system according to claim 1, characterized in that, The system also includes an edge computing module deployed in primary healthcare institutions and used in conjunction with the portable magnetic resonance imaging device. The edge computing module has a built-in lightweight AI model for real-time quality assessment and control of brain imaging data before data upload.

Images