Ultrasonic-assisted diagnosis method and system

Abstract

The present invention is applicable to the field of ultrasonic diagnostic technology, and provides an ultrasonic assisted diagnosis method and system. The method includes: collecting ultrasonic data by using an ultrasonic probe; transmitting the ultrasonic data from the ultrasonic probe to a millimeter wave communication module physically separated from the ultrasonic probe through a wired link; the millimeter wave communication module uses millimeter wave wireless communication technology to send the ultrasonic data to one of multiple millimeter wave access points deployed in the operating environment; transmitting the ultrasonic data received by the millimeter wave access point to a remote computing platform through a wired bearer network configured to support deterministic network functions; receiving the ultrasonic data transmitted through the deterministic network functions in the remote computing platform, and performing ultrasonic data processing on the received ultrasonic data according to a preset processing strategy to generate diagnostic auxiliary data for ultrasonic assisted diagnosis, and providing it to a display device for display. The present invention solves the problem of insufficient flexibility in existing ultrasonic diagnosis.

Description

Technical Field

[0001] The present invention relates to the field of ultrasonic diagnosis technology, and particularly to an ultrasonic assisted diagnosis method and system. Background Art

[0002] As a non-invasive, real-time, portable and relatively low-cost medical imaging technology, ultrasonic diagnosis plays an indispensable role in clinical diagnosis and is widely used in multiple fields such as the heart, abdomen, obstetrics and gynecology, blood vessels, superficial organs, etc. Traditional ultrasonic diagnosis devices usually adopt an integrated design, that is, integrating an ultrasonic probe, a signal processing unit, an image reconstruction unit, a control system, a display, a power supply, etc. in an independent mainframe system.

[0003] However, with the development of medical imaging technology and the continuous improvement of clinical requirements, some limitations of this traditional ultrasonic device architecture have gradually emerged:

[0004] High cost and resource waste: A large number of dedicated hardware (such as high-performance FPGA, ASIC, GPU) are integrated inside the traditional ultrasonic mainframe for real-time signal processing and image reconstruction, and the cost of this part of the hardware occupies the vast majority of the total cost of the device. For medical institutions with multiple consulting rooms, each consulting room needs to be equipped with a complete mainframe, resulting in a large amount of computing resources being idle during non-examination periods, causing significant resource waste and high purchase costs.

[0005] Performance bottleneck and difficulty in upgrading: Once the computing power inside the mainframe is determined, it is usually difficult to upgrade online. With the continuous emergence of new imaging algorithms with larger computational requirements (such as high-resolution three-dimensional / four-dimensional imaging, advanced elastography, super-resolution imaging, complex analysis based on artificial intelligence, etc.), the fixed computing power of the traditional mainframe often cannot support the real-time operation requirements of these computationally intensive tasks, restricting the expansion of clinical applications and the continuous improvement of device performance. The replacement of equipment usually means replacing the entire expensive mainframe, with high costs and a long cycle.

[0006] Large volume and limited portability: Ultrasonic mainframes with comprehensive functions are usually large in volume and heavy in weight. Although there are portable devices, their performance and functions are often compromised. This restricts the flexible application of ultrasonic examinations in scenarios such as beside the bed, emergency rooms, operating rooms, and remote areas.

[0007] Data silos and insufficient intelligence: The data generated by traditional devices is usually stored locally or in an independent PACS system, making it difficult to achieve large-scale data aggregation, management, and in-depth mining. Although some high-end devices integrate preliminary AI-assisted functions, limited by local computing power, they are difficult to run complex large AI models, and the level of intelligence needs to be improved. At the same time, the data and algorithms between devices of different manufacturers are usually incompatible, forming data and technology silos.

[0008] Challenges of Wireless Transmission: To improve portability and usability, wireless ultrasound probe technology has been explored. However, existing wireless transmission technologies (such as Wi-Fi and early Bluetooth) are difficult to meet the real-time and reliable transmission requirements of a large amount of raw ultrasound data (especially raw RF data) or high-channel beam data (usually requiring a bandwidth of several Gbps or even more than 10 Gbps) needed for high-quality diagnosis in terms of bandwidth, latency, and stability. Transmitting compressed images or low-channel data will result in loss of diagnostic information, limiting its application in high-end diagnosis. In addition, ordinary wireless networks cannot guarantee low latency and extremely low jitter, which are crucial for ultrasonic beam synthesis and real-time imaging that require precise timing control, and network jitter will lead to a decline in image quality or even artifacts.

[0009] Therefore, there is an urgent need for a new type of ultrasound diagnostic system architecture that can overcome the above drawbacks of traditional devices, achieve on-demand and flexible use of high-performance computing resources, support the deep integration of advanced imaging technologies and artificial intelligence, improve the flexibility and accessibility of devices, and reduce the overall cost of ownership. Summary of the Invention

[0010] Based on this, the object of the present invention is to provide an ultrasound-assisted diagnosis method and system to fundamentally solve the problem of insufficient flexibility in existing ultrasound diagnosis.

[0011] An ultrasound-assisted diagnosis method according to an embodiment of the present invention, the method includes:

[0012] Collect ultrasound data using at least one ultrasound probe;

[0013] Transmit the ultrasound data from the ultrasound probe to a millimeter-wave communication module physically separated from the ultrasound probe through a wired link;

[0014] The millimeter-wave communication module uses millimeter-wave wireless communication technology to send the ultrasound data to one of multiple millimeter-wave access points deployed in the operating environment;

[0015] Transmit the ultrasound data received by the millimeter-wave access point to a remote computing platform through a wired bearer network configured to support deterministic network functions to provide data transmission with a predetermined latency upper limit and a predetermined jitter upper limit;

[0016] Receive the ultrasound data transmitted through deterministic network functions in the remote computing platform, and perform ultrasound data processing on the received ultrasound data according to a preset processing strategy to generate diagnostic auxiliary data for ultrasound-assisted diagnosis, and provide it to a display device for display.

[0017] In addition, an ultrasonic assisted diagnosis method according to the above embodiments of the present invention may further have the following additional technical features:

[0018] Further, the step of the millimeter wave communication module using millimeter wave wireless communication technology to send the ultrasonic data to one of a plurality of millimeter wave access points deployed in the operating environment includes:

[0019] The millimeter wave communication module scans a preset millimeter wave channel to discover beacon frames or probe responses sent by a plurality of available millimeter wave access points in the operating environment;

[0020] The millimeter wave communication module selects a target millimeter wave access point from the discovered available millimeter wave access points for connection based on a preset selection criterion;

[0021] The millimeter wave communication module performs an authentication and association process with the selected target millimeter wave access point to establish a wireless link;

[0022] The millimeter wave communication module performs baseband processing on the ultrasonic data received from the wired link, and up-converts the signal after baseband processing to a predetermined millimeter wave operating frequency band. The baseband processing includes channel coding and digital modulation;

[0023] The millimeter wave communication module sends the up-converted millimeter wave signal to the target millimeter wave access point through the established wireless link.

[0024] Further, the step of the millimeter wave communication module performing an authentication and association process with the selected target millimeter wave access point to establish a wireless link includes:

[0025] The millimeter wave communication module performs a predefined authentication protocol with the selected target millimeter wave access point to verify the identities of both parties and establish a security context;

[0026] The millimeter wave communication module sends an association request to the target millimeter wave access point and establishes a logical connection after receiving an association response;

[0027] The millimeter wave communication module exchanges beamforming training information including antenna direction information with the target millimeter wave access point, and each calculates and adjusts the weight coefficients of the signal phase and / or amplitude of its own millimeter wave phased array antenna according to the received beamforming training information to form a directional wireless link pointing to each other.

[0028] Further, the step of the millimeter wave communication module sending the up-converted millimeter wave signal to the target millimeter wave access point through the established wireless link further includes:

[0029] The millimeter-wave communication module continuously monitors the wireless link quality parameters of the current connected target millimeter-wave access point, and periodically or based on a trigger event scans the signal quality parameters of other neighboring available millimeter-wave access points;

[0030] When the wireless link quality parameters of the millimeter-wave communication module with the current connected target millimeter-wave access point are lower than a preset handover threshold and the signal quality parameters of at least one neighboring available millimeter-wave access point are better than the current wireless link quality parameters, the millimeter-wave communication module selects the one with the best signal quality from the neighboring available millimeter-wave access points as the new target millimeter-wave access point;

[0031] The millimeter-wave communication module and the new target millimeter-wave access point execute a fast handover protocol to establish a new wireless link, and the fast handover protocol includes a fast authentication and association process.

[0032] Further, the step of transmitting the ultrasonic data received by the millimeter-wave access point to the remote computing platform through a wired bearer network configured to support deterministic network functions includes:

[0033] Synchronize the time of network devices and the remote computing platform in the wired bearer network using the Precision Time Protocol;

[0034] Calculate a fixed forwarding path for the ultrasonic data stream through the wired bearer network by the network controller, and calculate and determine the network resources to be reserved for the ultrasonic data stream, where the network resources include bandwidth resources and buffer resources;

[0035] The network controller configures the reserved network resources in the network devices on the fixed forwarding path, and configures the parameters of a periodic scheduling mechanism aligned with the synchronized time, and the periodic scheduling mechanism is selected from time-based gated scheduling or cyclic queue forwarding;

[0036] The network devices on the fixed forwarding path control the ultrasonic data stream to be forwarded along the fixed forwarding path according to the configured reserved network resources and the parameters of the periodic scheduling mechanism at a predetermined time rule to provide data transmission with a predetermined delay upper limit and a predetermined jitter upper limit.

[0037] Further, when the periodic scheduling mechanism is selected from time-based gated scheduling, the step of configuring the parameters of the periodic scheduling mechanism aligned with the synchronized time includes:

[0038] The network controller calculates and issues a gated list for the corresponding output port of each network device on the fixed forwarding path for the ultrasonic data stream, and the gated list specifies the exact time points when the gates of the queue corresponding to the ultrasonic data stream are opened and closed in each scheduling period;

[0039] The steps of controlling the ultrasonic data stream to be forwarded along a fixed forwarding path on a network device according to a predetermined time rule include:

[0040] The network device only opens the gate of the queue within the time window specified by the gating list to allow the ultrasonic data stream to be forwarded along the fixed forwarding path.

[0041] Further, when the periodic scheduling mechanism is selected from cyclic queue forwarding, the steps of configuring the periodic scheduling mechanism parameters aligned with the synchronization time include:

[0042] Configure a forwarding period for the network device;

[0043] The steps of controlling the ultrasonic data stream to be forwarded along a fixed forwarding path on a network device according to a predetermined time rule include:

[0044] The network device caches the received ultrasonic data stream and delays it to a specific time point within the next or subsequent predetermined forwarding period to send the cached ultrasonic data stream along the fixed forwarding path.

[0045] Further, the remote computing platform includes a distributed computing power cluster composed of multiple edge computing nodes and a central cloud node;

[0046] The steps of receiving ultrasonic data transmitted through a deterministic network function within the remote computing platform, performing ultrasonic data processing on the received ultrasonic data according to a preset processing strategy, and generating diagnostic auxiliary data for ultrasonic-assisted diagnosis include:

[0047] The distributed computing power cluster receives ultrasonic data from different ultrasonic probes transmitted through the deterministic network function;

[0048] The hybrid load computing power scheduler within the distributed computing power cluster receives processing task requests corresponding to each ultrasonic data, and the processing task requests include task types, required computing resources, and QoS requirements;

[0049] The hybrid load computing power scheduler queries the real-time available status of heterogeneous computing resources of each edge computing node within the distributed computing power cluster;

[0050] The hybrid load computing power scheduler distributes and schedules each ultrasonic data processing task to the corresponding specific edge computing nodes within the distributed computing power cluster for execution according to the processing task requests corresponding to each ultrasonic data, the real-time available status of heterogeneous computing resources of each edge computing node, and a preset scheduling strategy;

[0051] Each specific edge computing node performs ultrasonic data processing on the ultrasonic data corresponding to the ultrasonic data processing task according to a preset processing strategy.

[0052] Further, the step of performing ultrasonic data processing on the received ultrasonic data according to a preset processing strategy to generate diagnostic auxiliary data for ultrasonic assisted diagnosis and providing it to a display device for display includes:

[0053] The edge computing node cluster performs all required ultrasonic data processing on the received ultrasonic data according to a preset processing strategy, and sends the generated diagnostic auxiliary data for ultrasonic assisted diagnosis to the central cloud node. The central cloud node manages the received diagnostic auxiliary data and provides it to the display device for display; or the edge computing node cluster performs preliminary ultrasonic data processing on the received ultrasonic data according to a preset processing strategy, generates intermediate processing result data and sends it to the central cloud node. The central cloud node generates diagnostic auxiliary data for ultrasonic assisted diagnosis based on the received intermediate processing results and provides it to the display device for display.

[0054] Another object of an embodiment of the present invention is to provide an ultrasonic assisted diagnosis system, and the system includes:

[0055] At least one ultrasonic probe for collecting ultrasonic data;

[0056] A millimeter-wave communication module physically separated from the ultrasonic probe and connected by a wired link, configured to send the ultrasonic data collected by the ultrasonic probe using millimeter-wave wireless communication technology;

[0057] Multiple millimeter-wave access points deployed in the operating environment, configured to receive the ultrasonic data sent by the millimeter-wave communication module;

[0058] A wired bearer network configured to support deterministic network functions, for connecting the millimeter-wave access points to a remote computing platform and transmitting the received ultrasonic data to the remote computing platform to provide data transmission with a predetermined latency upper limit and a predetermined jitter upper limit; and the remote computing platform, configured to receive the ultrasonic data transmitted by the wired bearer network through the deterministic network functions, perform ultrasonic data processing on the received ultrasonic data according to a preset processing strategy, generate diagnostic auxiliary data for ultrasonic assisted diagnosis, and provide it to the display device for display.

[0059] The ultrasonic assisted diagnosis method provided by the embodiment of the present invention significantly reduces the complexity and cost of the front-end device by decoupling the core computing function from the probe end and transferring it to a shared remote computing platform, improves the utilization rate through pooling and sharing of computing resources, and reduces the total cost of ownership of deploying a high-performance ultrasonic system in a hospital or department; by using a millimeter-wave communication module physically separated from the ultrasonic probe and millimeter-wave wireless communication technology, combined with multi-millimeter-wave access point deployment and a fast handover mechanism, the bandwidth limitation and occlusion problems are successfully overcome, and stable and high-bandwidth wireless transmission of high-fidelity original ultrasonic data can be achieved while ensuring the portability of the ultrasonic probe; by applying deterministic network technology to the wired bearer network of ultrasonic data, through precise time synchronization, periodic scheduling, resource reservation and path binding, it ensures predictable low latency and extremely low jitter transmission of the ultrasonic data stream from the millimeter-wave access point to the remote computing platform, laying a solid network foundation for high-quality and artifact-free real-time ultrasonic image reconstruction; by means of a distributed computing power resource pool architecture based on edge computing, combined with a hybrid load computing power scheduler, heterogeneous computing resources can be elastically scheduled according to real-time requirements, and high-concurrency scenarios where multiple ultrasonic probes work simultaneously can be handled. At the same time, the computing power resources can be expanded on demand, avoiding the computing power bottleneck of traditional hosts; by aggregating ultrasonic data to the remote computing platform, data islands are broken, which is convenient for unified management, big data analysis, multi-modal fusion research and training of better ultrasonic diagnosis models; the problem of insufficient flexibility in existing ultrasonic diagnosis is solved. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 It is a schematic flowchart of the ultrasonic assisted diagnosis method in the first embodiment of the present invention;

[0061] Figure 2 It is a schematic structural diagram of the ultrasonic assisted diagnosis system in the second embodiment of the present invention;

[0062] The following specific embodiments will further illustrate the present invention in conjunction with the above-mentioned drawings. SPECIFIC EMBODIMENTS

[0063] To facilitate the understanding of the present invention, the present invention will be described more comprehensively below with reference to the relevant drawings. Several embodiments of the present invention are given in the drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. On the contrary, these embodiments are provided to make the disclosure of the present invention more thorough and comprehensive.

[0064] It should be noted that when an element is referred to as being "fixed to" another element, it may be directly on the other element or there may be a central element. When an element is considered to be "connected to" another element, it may be directly connected to the other element or there may be a central element at the same time. The terms "vertical", "horizontal", "left", "right" and similar expressions used herein are for illustrative purposes only.

[0065] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which the present invention belongs. The terms used herein in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more of the related listed items.

[0066] Embodiment 1

[0067] See also Figure 1 , which shows the ultrasound-assisted diagnosis method in the first embodiment of the present invention. For the convenience of description, only the part related to the embodiment of the present invention is shown. The ultrasound-assisted diagnosis method provided by the embodiment of the present invention includes:

[0068] Step S10, collecting ultrasound data using at least one ultrasound probe;

[0069] In one embodiment of the present invention, a specific part of a patient is scanned by an ultrasonic probe. At this time, the piezoelectric transducer array built into the ultrasonic probe transmits ultrasonic pulses and receives echo signals. The analog front end (AFE) circuit inside the ultrasonic probe performs low-noise amplification, bandpass filtering, and high-speed, high-precision (e.g., 12-16-bit) analog-to-digital conversion (ADC) on the received analog echo signals to generate a multi-channel digitized raw RF data stream.

[0070] Specifically, it mainly includes that the transmitting circuit (usually controlled by a main control unit such as an FPGA) generates a series of high-voltage and short-duration electrical pulse signals according to a preset imaging mode (such as B-mode, color Doppler, etc.) and parameters (such as transmit frequency, pulse width, focusing delay, etc.). For a phased array probe, excitation pulses with specific delays are generated for each (or a selected subset) of the piezoelectric transducer elements in the array. These delays are precisely calculated to focus the ultrasonic energy at a specific depth and direction within the tissue during transmission. At this time, these high-voltage electrical pulses are transmitted through a cable to the piezoelectric transducer array inside the ultrasonic probe. Each transducer element that receives the electrical pulse uses the inverse piezoelectric effect to convert electrical energy into mechanical vibration, thereby generating high-frequency (usually 1 - 20 MHz) ultrasonic pulses. Due to the time difference of the pulses emitted by each element (controlled by the transmit delay), these individually emitted sound waves interfere and superimpose within the tissue, forming a focused ultrasonic beam that is transmitted. Among them, the transmitted ultrasonic beam propagates in the human tissue. When encountering an interface with different acoustic impedances (the product of tissue density and sound velocity) (such as organ boundaries, blood vessel walls, internal tissue structures), a part of the ultrasonic energy will be reflected back. The inhomogeneity within the tissue (such as red blood cells, microstructures) will also cause the ultrasonic waves to scatter and propagate in all directions, and a part of them will also scatter back towards the probe direction. For the Doppler mode, moving objects (such as red blood cells in blood flow) will cause a Doppler frequency shift in the frequency of the reflected / scattered echo. Among them, the weak ultrasonic echo signal returning from the tissue reaches the transducer array on the probe surface. Each transducer element uses the direct piezoelectric effect to convert the received mechanical vibration (acoustic wave pressure) into a weak analog voltage signal. At this time, each channel (corresponding to one or a small group of transducer elements) outputs an analog voltage signal that changes with time, representing the echo intensity information returning from different depths. Since the received echo signal is very weak (from microvolts to millivolts) and contains a large amount of noise and interference, analog front-end processing (Analog Front-End, AFE) is required, specifically including a low-noise amplifier, time gain compensation, band-pass filtering, or demodulation. Among them, the low-noise amplifier amplifies the weak signal of each channel with a high multiple and low noise. Since ultrasonic waves attenuate during propagation in the tissue, the echo signal from deep tissue is much weaker than that from shallow tissue. Among them, time gain compensation dynamically increases the gain of the amplifier with the change of time (i.e., depth) to compensate for this attenuation, so that the echo signals at different depths have a similar amplitude range. Among them, band-pass filtering filters out noise and interference signals outside the ultrasonic working frequency range. In some architectures (especially traditional architectures), demodulation may be performed in the analog domain to extract the envelope information of the echo signal. However, in modern high-performance systems, it is more inclined to directly digitize the RF signal. Among them, the analog signal after analog front-end processing (usually the amplified and filtered RF signal) is sent into a high-speed and high-precision ADC.Among them, the ADC samples and quantizes the analog signals of each channel at a sampling rate much higher than the Nyquist frequency (for example, dozens of MHz to hundreds of MHz). The quantization bits are usually 10 bits, 12 bits, 14 bits or 16 bits to ensure sufficient dynamic range and signal accuracy. The output of the ADC is a multi-channel, digital ultrasonic data stream. If the RF signal is sampled directly, the output of the ADC is the raw RF data, where each sampling point represents the instantaneous amplitude of the echo signal received by a specific channel at a specific time (corresponding to a specific depth). At this time, the final output is multi-channel, digital ultrasonic data (usually raw RF data), which contains all the original information required for reconstructing ultrasonic images and making diagnoses.

[0071] Step S20, transmitting the ultrasonic data from the ultrasonic probe to a millimeter-wave communication module physically separated from the ultrasonic probe through a wired link;

[0072] Among them, in an embodiment of the present invention, during ultrasonic examination, good operational convenience is required. Considering that the size and weight of the millimeter-wave communication module may still be relatively large in a short period of time, if integrated with the ultrasonic probe, it will inevitably increase the weight and size of the probe, which will affect the operational convenience of doctors. On the other hand, when the ultrasonic probe is held in the doctor's hand, the possibility of millimeter waves being blocked by the arm and body is relatively large. For the above reasons, in the embodiment of the present invention, the millimeter-wave communication module is decoupled from the ultrasonic probe, and the ultrasonic probe is connected to the millimeter-wave communication module through a wired link, which can avoid the inconvenience caused by the extra weight brought by the millimeter-wave communication module and reduce the risk of signal occlusion.

[0073] Specifically, the digital ultrasonic data stream is transmitted through a high-speed, flexible hybrid cable connecting the ultrasonic probe and the millimeter-wave communication module. The cable uses differential signal transmission (such as LVDS or a higher-speed interface standard) to ensure signal integrity and may simultaneously provide low-voltage DC power for the ultrasonic probe. At the same time, the millimeter-wave communication module is physically separated from the ultrasonic probe, for example, worn on the waist of the operating doctor or placed on a trolley beside. Therefore, even if the ultrasonic probe is physically blocked in a certain direction (for example, the doctor holds it tightly), as long as the cable is okay, the ultrasonic data can reach the separately arranged millimeter-wave communication module, thus effectively improving the problem of signal attenuation or interruption of the wireless path caused by the existing users blocking the millimeter-wave communication module integrated on the ultrasonic probe.

[0074] Step S30, the millimeter-wave communication module uses millimeter-wave wireless communication technology to send the ultrasonic data to one of multiple millimeter-wave access points deployed in the operating environment;

[0075] Among them, in an embodiment of the present invention, the step of the millimeter-wave communication module using millimeter-wave wireless communication technology to send ultrasonic data to one of a plurality of millimeter-wave access points deployed in the operating environment includes:

[0076] The millimeter-wave communication module scans the preset millimeter-wave channels to discover beacon frames or probe responses sent by a plurality of available millimeter-wave access points in the operating environment;

[0077] The millimeter-wave communication module selects a target millimeter-wave access point from the discovered available millimeter-wave access points for connection based on a preset selection criterion;

[0078] The millimeter-wave communication module executes an authentication and association process with the selected target millimeter-wave access point to establish a wireless link;

[0079] The millimeter-wave communication module performs baseband processing on the ultrasonic data received from the wired link, and up-converts the signal after baseband processing to a predetermined millimeter-wave operating frequency band. The baseband processing includes channel coding and digital modulation;

[0080] The millimeter-wave communication module sends the up-converted millimeter-wave signal to the target millimeter-wave access point through the established wireless link.

[0081] Specifically, after the millimeter-wave communication module is started, its control unit will instruct the millimeter-wave transceiver to passively or actively scan the configured millimeter-wave channel list on a preset group of millimeter-wave channels according to a preset strategy (such as when idle, periodically, or according to a network-side instruction). Among them, passive scanning is to listen to the beacon frames (Beacon Frames) periodically broadcast by the millimeter-wave access point. The beacon frame contains information such as the identifier (BSSID) of the millimeter-wave access point (AP, Access Point), network name (SSID), supported rate, security policy, and channel information. Among them, active scanning is that the millimeter-wave communication module broadcasts probe request frames on each channel. At this time, the received millimeter-wave access point will reply with a probe response frame, providing information similar to the beacon frame. Finally, the millimeter-wave communication module constructs a list containing the surrounding available millimeter-wave access points and their basic information. The millimeter-wave access point serves as a bridge to connect the millimeter-wave communication module and the wired network infrastructure. The millimeter-wave access point receives data from the millimeter-wave communication module through millimeter-wave signals, and then forwards the data to the remote computing platform through its wired interface (usually an Ethernet interface or a fiber optic interface). The millimeter-wave access point itself usually does not perform complex millimeter-wave signal processing. Its main responsibilities are to establish and manage wireless connections, and forward data packets between wireless (millimeter-wave communication module to millimeter-wave access point) and wired (millimeter-wave access point to remote computing platform) networks.

[0082] Further, after the control unit of the millimeter-wave communication module collects information on all available millimeter-wave access points, it evaluates the millimeter-wave access points in the discovered millimeter-wave channel list according to one or more preset selection criteria (algorithms) and determines a target millimeter-wave access point for connection attempts. These selection criteria may include signal strength (RSSI), signal quality (SNR), AP load, or specific network requirements, etc. Among them, the signal strength selection criterion is to select the millimeter-wave access point with the strongest signal. The signal quality selection criterion is to select the millimeter-wave access point with the highest signal-to-noise ratio. The AP load selection criterion is that if the millimeter-wave access point broadcasts load information (such as the number of connected devices, channel utilization rate), then select the millimeter-wave access point with a lighter load. The specific network requirement criterion is to preferentially select the millimeter-wave access point that supports specific QoS capabilities or belongs to a specific network.

[0083] Further, the millimeter-wave communication module performs an authentication and association process with the selected target millimeter-wave access point to establish a wireless link. Further, the baseband processor in the millimeter-wave communication module receives a high-bit-rate ultrasonic data stream from the wired link and performs channel coding (such as LDPC), adding redundant check bits for detecting and correcting bit errors that may occur in wireless transmission at the receiving end, improving transmission reliability. Further, digital modulation (such as 64-QAM, 256-QAM) is also performed to map the encoded bit stream to the amplitude / phase state points of the millimeter-wave carrier. Further, the modulated digital baseband signal (I / Q signal) is sent to the millimeter-wave radio frequency transceiver. At this time, the millimeter-wave radio frequency transceiver mixes the modulated baseband signal with a high-frequency carrier (such as 60 GHz), shifting the signal spectrum to the predetermined millimeter-wave operating frequency band. Therefore, through baseband processing and upconversion, the digital ultrasonic data is converted into a high-frequency analog signal suitable for wireless transmission and capable of being transmitted in the millimeter-wave frequency band to resist channel noise and interference, finally generating a high-frequency millimeter-wave analog signal carrying ultrasonic data to be transmitted.

[0084] Further, the upconverted millimeter-wave signal is amplified by a power amplifier. At this time, the amplified signal is fed to the millimeter-wave phased array antenna of the millimeter-wave communication module, and the phased array antenna radiates the amplified millimeter-wave signal in the form of a directional beam towards the target millimeter-wave access point. The data transmission follows the MAC layer (Media Access Control) rules of the selected millimeter-wave wireless communication protocol, such as how to access the channel, frame format, etc., which will not be elaborated here.

[0085] Further, in an embodiment of the present invention, the steps of the millimeter-wave communication module performing an authentication and association process with the selected target millimeter-wave access point to establish a wireless link include:

[0086] The millimeter-wave communication module executes a predefined authentication protocol with the selected target millimeter-wave access point to verify the identities of both parties and establish a secure context;

[0087] The millimeter-wave communication module sends an association request to the target millimeter-wave access point and establishes a logical connection after receiving an association response;

[0088] The millimeter-wave communication module exchanges beamforming training information containing antenna direction information with the target millimeter-wave access point, and each calculates and adjusts the weight coefficients of the signal phase and / or amplitude of its own millimeter-wave phased array antenna according to the received beamforming training information to form a directional wireless link pointing to the other party.

[0089] Specifically, the millimeter-wave communication module executes a predefined authentication protocol agreed upon by both parties with the selected target millimeter-wave access point. In a medical environment, usually stronger security is required, so a protocol based on EEAP (such as EAP-TLS, EAP-PEAP) combined with a RADIUS server may be used for authentication, or standards such as certificate-based WPA3-Enterprise may be used to ensure that only authorized millimeter-wave communication modules can access the network and the communication is encrypted. The authentication process involves message exchange to verify identities and generate a session key. At this time, both parties exchange certificates or credentials, verify each other's identities, and negotiate a session key for subsequent data encryption, and finally establish a secure context. After successful authentication, the millimeter-wave communication module sends an association request frame to the target millimeter-wave access point, requesting to join the network served by this millimeter-wave access point. If the millimeter-wave access point accepts the request, it will reply with an association response frame and assign an association ID to the millimeter-wave communication module. At this time, a logical connection is established between the millimeter-wave communication module and the millimeter-wave access point, and the millimeter-wave communication module logically joins the network served by the millimeter-wave access point, ensuring that only legitimate millimeter-wave communication modules can access the network and establishing an encrypted channel and a logical link for subsequent data transmission.

[0090] Further, after completing the above-mentioned authentication and logical connection, beamforming training is performed. Specifically, a series of predefined beamforming training information containing antenna direction information is exchanged between the millimeter-wave communication module and the target millimeter-wave access point. The beamforming training information is some special signal sequences or training frames known to both parties in advance, which are used to detect the signal propagation effects in different directions. This process is usually two-way, that is, the millimeter-wave communication module sends training signals to the millimeter-wave access point, and the millimeter-wave access point also sends training signals to the millimeter-wave communication module. The training mechanism may include the sender alternately transmitting training signals in different preset directions (sectors), and the receiver measuring the signal strength in each direction to find the best receiving sector. After finding the best sector, more refined beam adjustment can be performed to further optimize the signal quality. The training signals may directly include the antenna configuration information of the sender or the proposed receiving direction information to accelerate the training process. Further, both the millimeter-wave communication module and the target millimeter-wave access point calculate the optimal phased array antenna weight coefficients (the complex gains of each antenna unit, that is, amplitude and phase) according to the received training information (for example, which direction has the strongest signal and the highest signal-to-noise ratio) using internal algorithms (such as least mean square error LMS, recursive least squares RLS, etc.). And the calculated weight coefficients are loaded into their respective phased array antenna controllers to adjust the phase and / or amplitude of the actual transmitted / received signals, thereby forming a high-gain, narrow-beam, and precisely directed wireless link to the other party. Compared with omnidirectional antennas, higher signal gains (link budgets) can be obtained, supporting higher data rates and longer transmission distances, and reducing interference in other directions.

[0091] Further, the step of the millimeter-wave communication module sending the up-converted millimeter-wave signal to the target millimeter-wave access point through the established wireless link further includes:

[0092] The millimeter-wave communication module continuously monitors the wireless link quality parameters of the currently connected target millimeter-wave access point and periodically or based on trigger events scans the signal quality parameters of other neighboring available millimeter-wave access points;

[0093] When the wireless link quality parameters of the millimeter-wave communication module and the currently connected target millimeter-wave access point are lower than the preset handover threshold and the signal quality parameters of at least one neighboring available millimeter-wave access point are better than the current wireless link quality parameters, the millimeter-wave communication module selects the one with the best signal quality from the neighboring available millimeter-wave access points as the new target millimeter-wave access point;

[0094] The millimeter-wave communication module and the new target millimeter-wave access point execute a fast handover protocol to establish a new wireless link, and the fast handover protocol includes a fast authentication and association process.

[0095] Specifically, the control unit of the millimeter-wave communication module continuously monitors the quality parameters of the currently connected wireless link, such as RSSI, SNR, PER / BER, etc. Among them, RSSI is the signal strength, SNR is the ratio of signal to noise, and PER / BER is the packet / bit error rate. At the same time, the millimeter-wave communication module will use the transmission gap or dedicated time window to periodically (for example, every few hundred milliseconds) or based on trigger events (for example, when the current link quality is lower than a certain warning threshold), briefly scan other channels and measure the signal quality of neighboring available millimeter-wave access points. Then, compare the current wireless link quality parameters with one or more preset handover thresholds, and at the same time compare the current wireless link quality parameters with the signal quality parameters of other neighboring available millimeter-wave access points scanned. If the current wireless link quality parameters are lower than the preset handover threshold, and there is a neighboring millimeter-wave access point whose signal quality parameters are better than the current link (usually, it is also required to be better than a specific threshold to avoid ping-pong handover), then select the neighbor with the best signal quality as the new target millimeter-wave access point from all neighboring millimeter-wave access points that meet the handover conditions. Once the handover is decided, the millimeter-wave communication module immediately initiates the fast handover protocol with the new target millimeter-wave access point. The fast handover protocol includes fast authentication and association processes. Specifically, use the partial security context or cached key information established with the new target millimeter-wave access point before (for example, during the initial scan or with the assistance of the network side) to perform a simplified and faster authentication and association process. Specifically, if the network supports, the millimeter-wave communication module and the new target millimeter-wave access point may skip the complete EAP authentication using the master key or cached key information established with the authentication server before and complete some authentication steps with the potential target millimeter-wave access point before the handover. At the same time, use the information obtained from the previous scan or the fast beam training mechanism defined in the protocol to quickly establish an optimized directional beam with the new target millimeter-wave access point. After the handover is successful, the millimeter-wave communication module will update its transmission target and send subsequent data packets to the new target millimeter-wave access point. Therefore, even in the case of deterioration of the original link caused by movement or occlusion, the millimeter-wave communication module can quickly and automatically switch to a better millimeter-wave access point to maintain the continuous and reliable transmission of the ultrasonic data stream.

[0096] Step S40: Transmit the ultrasonic data received by the millimeter-wave access point to the remote computing platform through a wired bearer network configured to support deterministic network functions;

[0097] Among them, in an embodiment of the present invention, the step of transmitting the ultrasonic data received by the millimeter-wave access point to the remote computing platform through a wired bearer network configured to support deterministic network functions includes:

[0098] Synchronize the network devices and remote computing platforms in the wired bearer network using the Precision Time Protocol;

[0099] Calculate a fixed forwarding path for the ultrasonic data stream through the wired bearer network by the network controller, and calculate and determine the network resources to be reserved for the ultrasonic data stream. The network resources include bandwidth resources and buffer resources;

[0100] The network controller configures the reserved network resources in the network devices on the fixed forwarding path, and configures the parameters of the periodic scheduling mechanism aligned with the synchronization time. The periodic scheduling mechanism is selected from time-based gating scheduling or cyclic queue forwarding;

[0101] The network devices on the fixed forwarding path control the ultrasonic data stream to be forwarded along the fixed forwarding path according to the configured reserved network resources and the parameters of the periodic scheduling mechanism, in accordance with a predetermined time rule, so as to provide data transmission with a predetermined delay upper limit and a predetermined jitter upper limit.

[0102] Specifically, deploy the Precision Time Protocol (PTP-IEEE 1588) in the entire wired bearer network, and specify one or more master clocks in the network, usually high-precision and stable clock devices. All network devices (switches, routers) participating in deterministic forwarding and remote processing systems run the PTP protocol stack, serving as PTP slave clocks or boundary clocks / transparent clocks. The slave clocks exchange PTP protocol messages with precise timestamps with the master clock, and use the path delay measurement mechanism to continuously adjust their local clocks to keep them highly consistent with the master clock time, achieving a synchronization accuracy at the nanosecond or microsecond level, providing a unified and precise time reference for all subsequent time-based scheduling operations.

[0103] Among them, when a deterministic connection needs to be established for the ultrasonic data stream, its network controller receives a request to establish a deterministic connection for the ultrasonic data stream (such as from a target millimeter-wave access point to a specific edge computing node), where the request includes a flow identifier, source / destination addresses, and QoS requirements (such as maximum latency, maximum jitter, required bandwidth). The network controller needs to know the real-time topology of the network and the current resource usage of network devices (link bandwidth occupancy, queue status, scheduling arrangements of existing deterministic flows, etc.). At this time, the network controller runs a constraint-based path calculation algorithm (such as an extended Dijkstra algorithm considering latency, hop count, and link bandwidth) to calculate a fixed forwarding path in the network topology that meets the QoS requirements and has available resources. The network controller checks the available resources of each node and link along the calculated fixed forwarding path, and calculates the network resources that need to be reserved for each node on the fixed forwarding path according to the characteristics of the data stream (such as average rate, maximum burst) and the characteristics of the selected path. The network resources specifically include bandwidth resources and buffer resources. The bandwidth resources ensure that each link on the path has sufficient available bandwidth (for example, peak rate); the buffer resources are how much queue buffer needs to be allocated on each network device to temporarily store the data packets of the ultrasonic data stream to prevent packet loss caused by instantaneous rate fluctuations or scheduling waits.

[0104] Furthermore, the network controller distributes the calculated fixed forwarding path information to each network device on the path through a control protocol, which usually manifests as configuring specific flow table rules (matching the ultrasonic data stream, with the action of forwarding to the specified next hop) or Segment Routing information. Furthermore, the network controller converts the calculated bandwidth and buffer resource requirements into specific configuration commands (such as queue size settings, bandwidth limit parameters), and distributes them to the corresponding ports and queues of the corresponding network devices. Furthermore, the network controller distributes the calculated periodic scheduling mechanism parameters (such as GCL list or CQF queue allocation / forwarding timing) for the ultrasonic data stream to the hardware scheduler of the network device. At this time, by configuring the calculated path, resources, and scheduling rules into the network devices, the network devices are prepared to perform deterministic forwarding.

[0105] Furthermore, when an ultrasonic data packet from a millimeter-wave access point arrives at a certain network device on the path, the network device first identifies the ultrasonic data stream that requires deterministic guarantee according to the packet header information (such as five-tuple, VLAN Tag, MPLS label, or specific field), and puts it into a designated queue with pre-configured resources. The forwarding engine of the network device operates strictly according to the configured periodic scheduling mechanism parameters and the synchronized clock. Since the bandwidth and buffer of the relevant queue have been reserved, and the transmission time window is exclusive or protected, the packet transmission will not be interfered by other (low-priority) traffic and will not be discarded due to insufficient resources. At this time, by precisely controlling the forwarding time and path of each packet at each network device node and guaranteeing the required resources, the end-to-end predetermined delay upper limit (the sum of fixed delays of each segment) and the predetermined jitter upper limit (the queuing jitter is eliminated or controlled within a very small range) are finally achieved.

[0106] It should be noted that the network controller is a logically independent (physically may be independently deployed or may run as software on a server or cloud platform) control plane entity. For example, the network controller can be an independent software application running on a dedicated server or virtual machine, and this server can be located in the hospital's data center or in the cloud. If the hospital's network adopts a software-defined network (SDN) architecture, the network controller can be integrated into the SDN controller, where the SDN controller itself is a platform for centralized network control. If the entire system is uniformly managed by the cloud platform, the network controller can also exist as a service module of the cloud platform. At this time, the network controller is a logically centralized control point, capable of communicating with relevant devices in the network (the wired interface side of the millimeter-wave access point, network devices (switches and routers), DPU / network cards of the remote computing platform) to obtain network status, calculation paths, and resources, and issue configuration instructions. The network controller is not located on the data forwarding path but interacts with the data plane devices through control protocols.

[0107] Among them, in an embodiment of the present invention, when the periodic scheduling mechanism is selected from time-based gated scheduling, the steps of configuring the periodic scheduling mechanism parameters aligned with the synchronization time include: the network controller calculates and distributes a gating list for the corresponding output port of each network device on the fixed forwarding path of the ultrasonic data stream, and the gating list specifies the exact time points when the gate of the queue corresponding to the ultrasonic data stream is opened and closed within each scheduling period. Among them, the network controller calculates a gating list (GCL) for the relevant output port of each network device on the path (that is, the ultrasonic stream will be sent from this port), and distributes and programs the gating list into the hardware scheduler of the network device through the control protocol. Among them, the gating list is a schedule that precisely defines at which time points the gate of the hardware queue assigned to the ultrasonic data stream should be opened within each repeated scheduling period, how long it should be opened, and at which time points it should be closed. Among them, the above steps of controlling the ultrasonic data stream to be forwarded along the fixed forwarding path on the network device according to the predetermined time rule include: the network device only opens the gate of the queue within the time window specified by the gating list to allow the ultrasonic data stream to be forwarded along the fixed forwarding path. Specifically, the hardware scheduler of the network device runs strictly according to the synchronization clock and the loaded gating list. When the time reaches the opening time point specified in the gating list for the queue, the hardware opens the gate of the queue. Within the time window when the gate is open, if there are packets in the queue, they are sent out in order (usually FIFO) and follow the fixed path rule. When the time reaches the closing time point specified by the gating list, the hardware scheduler closes the gate of the queue, and even if there are still packets in the queue, the transmission is paused until the next opening time window assigned to it.

[0108] Among them, in one embodiment of the present invention, when the periodic scheduling mechanism is selected as cyclic queue forwarding, the steps of configuring the periodic scheduling mechanism parameters aligned with the synchronization time include: configuring a forwarding period for the network device. The network controller defines or configures a forwarding period for each network device on the path uniformly. At the same time, the controller may also need to configure in which (or which) cyclic queue the data packet should be processed. Among them, the above steps of controlling the ultrasonic data stream to be forwarded along a fixed forwarding path on the network device according to a predetermined time rule include: the network device caches the received ultrasonic data stream and delays it until a specific time point within the next or subsequent predetermined forwarding period to send the cached ultrasonic data stream along the fixed forwarding path. Specifically, when a data packet belonging to the ultrasonic data stream arrives at the network device, the network device puts it into a specific cache queue (usually associated with the next forwarding period) according to its arrival time and the configured forwarding period. The network device holds this data packet without forwarding it immediately until a predetermined time point in the next forwarding period (such as the start of the period or a fixed offset within the period), and then the network device takes out the data packet from the cache queue and sends it out according to the fixed path rule. This process ensures that the data packet experiences a fixed delay approximately equal to the forwarding period (or its multiple) at each node, thereby smoothing the jitter of the arrival time.

[0109] Further, the specific workflow for the above millimeter-wave access point to transmit ultrasonic data through a wired bearer network is as follows: The target millimeter-wave access point receives the up-converted millimeter-wave signal from the millimeter-wave communication module through its millimeter-wave antenna. Further, the radio frequency front-end inside the target millimeter-wave access point amplifies, filters, and down-converts the signal to restore it to the baseband. Further, the baseband processing unit of the target millimeter-wave access point demodulates (such as QAM demodulation) and channel decodes (such as LDPC decoding) the baseband signal to restore the original digitized ultrasonic data stream. Further, the restored ultrasonic data stream is sent to the wired network interface of the target millimeter-wave access point (such as an SFP+ / QSFP port supporting Gigabit Ethernet or optical fiber), where if necessary, the target millimeter-wave access point may encapsulate the original ultrasonic data stream and package it into a standard network data packet (such as encapsulated in UDP / IP or a specific tunneling protocol), and the above encapsulation step can also be completed in the millimeter-wave communication module, in which case the target millimeter-wave access point only needs to forward it. Necessary header information may be added during encapsulation, including identifiers for identifying the data stream (such as VLAN ID, MPLS label, Flow Label, etc.), and possible deterministic network-related metadata (if the protocol requires it to be carried in the data packet). The target millimeter-wave access point, as the boundary node between the wireless domain and the wired deterministic network, may act as the entrance to the deterministic network, or the first switch adjacent to the target millimeter-wave access point acts as the entrance to the deterministic network. If entrance processing is performed here, the target millimeter-wave access point (or the first-hop switch) needs to perform flow classification, marking, and traffic shaping. Flow classification is used to identify that this is an ultrasonic data stream that requires deterministic guarantee. Marking is used to mark the data packet with a specific priority mark or flow identifier. Traffic shaping is used to shape the data stream that may still have a small amount of bursts from the wireless link to make it smoother and meet the requirements of subsequent deterministic network scheduling. Further, the wired interface side of the target millimeter-wave access point (and all switches and routers on the subsequent path) are synchronized with the network master clock through PTP. When the data packet arrives at the first switch on the path, the switch determines the next hop of the data packet and the outgoing port to be used according to the fixed forwarding path pre-distributed by the network controller. The data packet is placed in the high-priority queue of this outgoing port. The switch processes it according to the PTP synchronized clock and the periodic scheduling mechanism parameters configured by the network controller. If it is time-based gated scheduling, the switch waits until the time window corresponding to this queue opens before allowing the data packet to be sent out from the queue. If it is circular queue forwarding, the switch caches the data packet and delays it until a specific time point in the next predetermined period before sending it out.Among them, the data packet follows the fixed forwarding path specified by the network controller. On each switch / router that supports deterministic network functions on the path, the above-mentioned scheduling and forwarding processes are repeated. At each node, the data packet is processed and forwarded according to a predetermined time rule. Since the network controller has reserved bandwidth and buffer resources, the data packet will not be lost or experience excessive queuing delays due to network congestion during transmission (except for the predictable delays introduced by the scheduling mechanism itself). After being transmitted through the deterministic wired bearer network, the ultrasonic data stream finally reaches the remote computing platform. Due to the guarantee of the deterministic network, the arriving data stream has very low jitter and predictable, bounded end-to-end delay. The receiving node (such as the DPU or network card of the edge computing node) in the remote computing platform receives the data packet, performs decapsulation (if necessary), and sends the recovered ultrasonic data to the upper-layer application or memory for processing. The DPU may also participate in the deterministic scheduling at the receiving end (such as ensuring that the data is delivered to the application buffer on time).

[0110] Step S50: Receive the ultrasonic data transmitted through the deterministic network function within the remote computing platform, perform ultrasonic data processing on the received ultrasonic data according to a preset processing strategy, generate diagnostic auxiliary data for ultrasonic-assisted diagnosis, and provide it to a display device for display;

[0111] Among them, in an embodiment of the present invention, the remote computing platform includes a distributed computing power cluster composed of multiple edge computing nodes and a central cloud node;

[0112] The steps of receiving the ultrasonic data transmitted through the deterministic network function within the remote computing platform, performing ultrasonic data processing on the received ultrasonic data according to a preset processing strategy, and generating diagnostic auxiliary data for ultrasonic-assisted diagnosis include:

[0113] The distributed computing power cluster receives the ultrasonic data from different ultrasonic probes transmitted through the deterministic network function;

[0114] The hybrid load computing power scheduler within the distributed computing power cluster receives the processing task requests corresponding to each ultrasonic data, and the processing task requests include task types, required computing resources, and QoS requirements;

[0115] The hybrid load computing power scheduler queries the real-time available status of the heterogeneous computing resources of each edge computing node within the distributed computing power cluster;

[0116] The hybrid load computing power scheduler distributes and schedules each ultrasonic data processing task to the corresponding specific edge computing nodes within the distributed computing power cluster for execution according to the processing task requests corresponding to each ultrasonic data, the real-time available status of the heterogeneous computing resources of each edge computing node, and a preset scheduling strategy;

[0117] Each specific edge computing node processes the ultrasonic data corresponding to the ultrasonic data processing task according to a preset processing strategy.

[0118] Specifically, a network interface (usually a high-speed Ethernet) located at the boundary of the edge computing node receives an ultrasound data stream that has been deterministically transmitted from the wired bearer network. The network interface or DPU needs to be able to identify different data streams (e.g., based on the source AP, ultrasound probe ID, or specific network flow identifier) and direct them to the correct processing pipeline or initial buffer. The distributed computing power cluster can simultaneously receive multiple concurrent data streams from different ultrasound probes (corresponding to different examination sessions). The hybrid workload computing power scheduler (which may be an independent cluster management service or integrated into an orchestration platform such as Kubernetes) receives a processing task request associated with each ultrasound data stream. This request may be automatically triggered by the system (e.g., once a new data stream is detected), or initiated by an upper-layer application (such as an electronic medical record system or an examination appointment system). The processing task request includes the task type, the required computing resources, and QoS requirements; among them, the task type is standard real-time imaging (B-mode, color Doppler, etc.), or specific ultrasound diagnostic analysis (lesion detection, segmentation, measurement) is required, or other special processing. The required computing resources are how many CPU cores are needed, how much memory is needed, whether a GPU is required (and the specific model or computing power requirements), whether DPU acceleration is required, etc. The QoS requirements are requirements for latency, throughput requirements, task priorities, etc. The hybrid workload computing power scheduler continuously or when scheduling is needed, queries the real-time resource availability status of all edge computing nodes within the distributed computing power cluster. Usually, a monitoring agent runs on each edge computing node, responsible for collecting heterogeneous computing resources of the edge computing node, such as CPU usage, memory usage, GPU usage and video memory occupancy, DPU load, network bandwidth occupancy, disk I / O, etc., and reporting them to the hybrid workload computing power scheduler. The hybrid workload computing power scheduler obtains the current availability and load conditions of the heterogeneous computing resources of each node by communicating with the monitoring agents deployed on each node. The hybrid workload computing power scheduler matches the resource requirements in the received task request with the available resources of each node queried. It filters out candidate nodes that can meet the task resource requirements, scores the candidate nodes according to a series of preset policies (algorithms), and selects the optimal node to execute the task. These policies may include resource sufficiency, load balancing, priority and preemption, data locality, affinity / anti-affinity, functional specialization, where resource sufficiency is to select the node with the most abundant resources. Load balancing is to select the node with the lowest current load to avoid hotspots. Priority and preemption mean that if it is a high-priority real-time task and there are no idle resources, it may trigger a preemption mechanism to pause or evict low-priority tasks to release resources. Data locality means that if a task needs to access a large amount of data, it preferentially selects the storage node where the data is located or the computing node closest to the data.Affinity / anti-affinity schedules a specific task to (or avoids scheduling to) a specific type of node based on the tags of the task or node (e.g., scheduling an AI task to a node with a GPU). Functional specialization schedules a matching task to a node if the node has a functional label (such as "heart processing node"). Finally, the hybrid workload computing power scheduler decides to allocate the processing task to a specific computing unit on a selected specific edge computing node. The execution agent (such as kubelet or container runtime) on the selected edge computing node receives the instruction from the hybrid workload computing power scheduler. The agent starts the corresponding handler or container on the server according to the instruction. The handler / container loads the ultrasound data to be processed (which may be read from local cache, distributed storage, or directly from the network stream). According to the processing strategy specified in the task request or preset, the corresponding ultrasound data processing algorithm (such as beamforming, image reconstruction, AI inference, etc.) is executed. After processing is completed, the generated diagnostic auxiliary data is output to a predetermined location (such as memory, local storage, distributed storage, or directly sent to the network).

[0119] Further, the steps of performing ultrasound data processing on the received ultrasound data according to a preset processing strategy, generating diagnostic auxiliary data for ultrasound-assisted diagnosis, and providing it to a display device for display include:

[0120] The edge computing node cluster performs all required ultrasound data processing on the received ultrasound data according to a preset processing strategy, and sends the generated diagnostic auxiliary data for ultrasound-assisted diagnosis to the central cloud node. The central cloud node manages the received diagnostic auxiliary data and provides it to the display device for display; or the edge computing node cluster performs preliminary ultrasound data processing on the received ultrasound data according to a preset processing strategy, generates intermediate processing result data and sends it to the central cloud node. The central cloud node generates diagnostic auxiliary data for ultrasound-assisted diagnosis based on the received intermediate processing results and provides it to the display device for display.

[0121] Specifically, after the ultrasound data has been transmitted to the edge computing cluster through a deterministic network and the hybrid workload computing power scheduler has allocated the processing task to a specific edge computing node, it includes two different ways of task allocation and data flow between the edge computing cluster and the central cloud node.

[0122] The first method is that the edge computing node assigned with tasks loads the received ultrasonic data. The edge computing node sequentially executes all necessary processing algorithms according to a preset processing strategy (for example, for the complete process of cardiac examination). It includes real-time signal / image processing, advanced imaging processing, and diagnostic analysis, where real-time signal / image processing includes beamforming, B-mode image reconstruction, color / spectral Doppler processing, M-mode generation, or basic measurements (such as distance, area, speed), etc. Advanced imaging processing includes elastography calculation, contrast-enhanced analysis, or three-dimensional / four-dimensional reconstruction, etc. If the diagnostic analysis includes loading a diagnostic model deployed on the shared storage within the edge computing node or distributed computing power cluster to perform real-time inference on the generated images or data, for example: automatically identifying anatomical structures, detecting suspicious lesions, automatically performing standardized section evaluation, providing preliminary quantitative indicators or classification suggestions. The edge computing node integrates the results of all processing steps into a complete set of diagnostic auxiliary data that can be directly used by doctors for ultrasonic assisted diagnosis. The diagnostic auxiliary data specifically includes, for example, the finally optimized two-dimensional / three-dimensional / four-dimensional image sequences, Doppler spectrograms and related measurement values, diagnostic analysis results (such as lesion markings, measurement values, risk scores), and structured measurement data. Then the edge computing node packages the generated diagnostic auxiliary data and sends it to the central cloud node through the network. The central cloud node receives the diagnostic auxiliary data from different edge computing nodes, uniformly stores and archives these diagnostic auxiliary data, and can also associate the diagnostic results or key indicators determined by the diagnostic auxiliary data with the hospital information system (HIS) or radiology information system (RIS). At this time, the central cloud platform serves as a unified outlet, responsible for securely providing these diagnostic auxiliary data to authorized display devices (such as doctor workstations, mobile terminals) for display, which can be achieved through web services, API interfaces, or specific streaming media services. At this time, doctors can view the complete examination results and reports by accessing the cloud platform.

[0123] In the second approach, the edge computing nodes assigned tasks load the received ultrasound data. According to the preset processing strategy, the edge computing nodes only execute some of the preprocessing or real-time processing steps. The preliminary processing may include beamforming to generate beam domain data, basic B-mode image reconstruction to generate a grayscale image sequence without advanced post-processing, basic color / spectral Doppler data processing, or extraction of some key intermediate features or metadata. The edge computing nodes package the outputs of these preliminary processes into intermediate processing result data, which may be smaller in volume than the original ultrasound data but still contains the core information required for subsequent analysis. Then, the edge computing nodes send the generated intermediate processing result data to the central cloud node via the network. The central cloud node receives the intermediate processing result data from the edge computing nodes and utilizes its more powerful computing resources (which may include large-scale GPU clusters, dedicated AI chips, big data platforms) to execute subsequent and more complex processing steps, which may specifically include: 1. Advanced image post-processing: such as applying complex filtering algorithms, spatial / frequency multiplexing techniques, texture analysis, tissue characterization, etc. 2. Complex quantitative analysis: such as advanced quantitative analysis of cardiac function, hemodynamic simulation, calculation of elastic modulus, etc. 3. Complex diagnostic model inference: such as running diagnostic models with huge computational requirements, requiring access to massive parameters or combining historical / multi-modal data for more refined lesion detection, segmentation, benign / malignant judgment, prognosis prediction, etc. 4. Multi-modal fusion: such as fusing the ultrasound intermediate results with other imaging data such as CT / MRI stored in the cloud for analysis. The central cloud node finally generates complete diagnostic assistance data and then provides it to the display device for display or stores it in the cloud database.

[0124] That is to say, the first approach focuses on the edge computing nodes, aiming for the lowest latency and the autonomy of the edge computing nodes. The second approach makes better use of the powerful capabilities of the central cloud node and is suitable for more complex analysis tasks. Which specific approach to adopt, or a hybrid use of these two approaches in different examinations, depends on the preset processing strategy, which determines the specific division of tasks and the data flow mode between the edge computing nodes and the central cloud node according to clinical requirements, available resources, and performance requirements. Specifically, the preset processing strategy is the key to determining which approach to adopt and which specific algorithms to execute, and its implementation is usually:

[0125] Based on the examination type / protocol: Different ultrasound examination items (such as routine abdominal scan, cardiac structure measurement, fetal NT examination, AI analysis of thyroid nodules) will have corresponding processing flow templates predefined.

[0126] User selection or system automatic recognition: The doctor selects the examination protocol at the start of the examination, or the system automatically matches the processing strategy based on information such as the examination site and probe type.

[0127] Policy Configuration: These processing policies (including the algorithm call order, parameter settings, whether to adopt Method 1 or Method 2, etc.) are usually configured and managed at the central cloud node and then sent to the hybrid load computing power scheduler or the edge computing node for execution.

[0128] Dynamic Adjustment: Dynamically adjust the processing policies according to the real-time network conditions, the load of the distributed computing power cluster, and even the image content (for example, tend to use Method 1 when the network is congested, and adopt Method 2 when complex diagnostic processing is required).

[0129] In summary, in the above embodiments of the present invention, the ultrasonic assisted diagnosis method decouples the core computing function from the probe end and transfers it to a shareable remote computing platform, significantly reducing the complexity and cost of the front-end device, improving the utilization rate through the pooling and sharing of computing resources, and reducing the total cost of ownership of deploying a high-performance ultrasonic system in a hospital or department; by using a millimeter-wave communication module and millimeter-wave wireless communication technology physically separated from the ultrasonic probe, combined with the deployment of multiple millimeter-wave access points and the fast handover mechanism, the bandwidth limitation and occlusion problems are successfully overcome, and while ensuring the portability of the ultrasonic probe, stable and high-bandwidth wireless transmission of high-fidelity original ultrasonic data can be achieved; by applying deterministic network technology to the wired bearer network of ultrasonic data, through precise time synchronization, periodic scheduling, resource reservation, and path binding, it ensures predictable low latency and extremely low jitter transmission of ultrasonic data streams from the millimeter-wave access point to the remote computing platform, laying a solid network foundation for high-quality and artifact-free real-time ultrasonic image reconstruction; through the distributed computing power resource pool architecture based on edge computing, combined with the hybrid load computing power scheduler, it can flexibly schedule heterogeneous computing resources according to real-time needs, can handle the high-concurrency scenario of multiple ultrasonic probes working simultaneously, and at the same time the computing power resources can be expanded on demand, avoiding the computing power bottleneck of traditional hosts; by aggregating ultrasonic data to the remote computing platform, data islands are broken, facilitating unified management, big data analysis, multi-modal fusion research, and the training of better ultrasonic diagnosis models; it solves the problem of insufficient flexibility in existing ultrasonic diagnosis.

[0130] Embodiment 2

[0131] Please refer to Figure 2 , which is a schematic structural diagram of an ultrasonic assisted diagnosis system provided by the second embodiment of the present invention. For the convenience of description, only the parts related to the embodiments of the present invention are shown. The system includes:

[0132] At least one ultrasonic probe 10 for collecting ultrasonic data;

[0133] A millimeter-wave communication module 20 physically separated from the ultrasonic probe 10 and connected by a wired link, which is used to send the ultrasonic data collected by the ultrasonic probe 10 by using millimeter-wave wireless communication technology;

[0134] Multiple millimeter-wave access points 30 deployed in an operating environment for receiving ultrasonic data transmitted by a millimeter-wave communication module 20;

[0135] A wired bearer network 30 configured to support deterministic network functions, for connecting the millimeter-wave access points 30 to a remote computing platform 40 and transmitting the received ultrasonic data to the remote computing platform 40 to provide data transmission with a predetermined upper limit of latency and a predetermined upper limit of jitter; and a remote computing platform 40, for receiving the ultrasonic data transmitted by the wired bearer network 30 through deterministic network functions, performing ultrasonic data processing on the received ultrasonic data according to a preset processing strategy, generating diagnostic auxiliary data for ultrasonic-assisted diagnosis, and providing it to a display device for display.

[0136] The ultrasonic-assisted diagnosis system provided by the embodiments of the present invention has the same implementation principle and the same technical effects as those of the foregoing method embodiments. For the sake of brief description, for the parts not mentioned in the device embodiments, reference may be made to the corresponding content in the foregoing method embodiments.

[0137] In the description of this specification, the description with reference to terms such as "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0138] The above-described embodiments only represent several implementation manners of the present invention. Their descriptions are relatively specific and detailed, but should not be construed as limiting the scope of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, several modifications and improvements can still be made, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the appended claims.

Claims

1. An ultrasound-assisted diagnosis method, characterized in that: The method comprises: collecting ultrasound data using at least one ultrasound probe; transmitting the ultrasound data from the ultrasound probe to a millimeter wave communication module physically separated from the ultrasound probe via a wired link; The millimeter wave communication module uses millimeter wave wireless communication technology to send the ultrasonic data to one of a plurality of millimeter wave access points deployed in the operating environment; Transmitting the ultrasonic data received by the millimeter wave access point to a remote computing platform through a wired bearer network configured to support a deterministic network function, so as to provide data transmission with a predetermined upper limit of delay and a predetermined upper limit of jitter; Receiving ultrasound data transmitted via the deterministic network function in the remote computing platform, performing ultrasound data processing on the received ultrasound data according to a preset processing strategy, generating diagnosis auxiliary data for ultrasound-assisted diagnosis, and providing the data to a display device for display; The step of using the millimeter wave communication module to send the ultrasonic data to one of a plurality of millimeter wave access points deployed in the operating environment using millimeter wave wireless communication technology comprises: The millimeter wave communication module scans a preset millimeter wave channel to find beacon frames or probe responses sent by a plurality of millimeter wave access points available in the operating environment; The millimeter wave communication module selects a target millimeter wave access point from the available millimeter wave access points found for connection based on a preset selection criterion; The millimeter wave communication module performs authentication and association procedures with the selected target millimeter wave access point to establish a wireless link; The millimeter wave communication module performs baseband processing on the ultrasonic data received from the wired link and up-converts the baseband processed signal to a predetermined millimeter wave operating frequency band, wherein the baseband processing includes channel coding and digital modulation; The millimeter wave communication module sends the up-converted millimeter wave signal to the target millimeter wave access point through the established wireless link; The step of transmitting the ultrasonic data received by the millimeter wave access point to a remote computing platform via a wired bearer network configured to support deterministic network functions comprises: Using the precise time protocol to synchronize the time between the network equipment in the wired bearer network and the remote computing platform; Calculating a fixed forwarding path through a wired bearer network for the ultrasonic data stream through a network controller, and calculating and determining network resources required to be reserved for the ultrasonic data stream, wherein the network resources include bandwidth resources and buffer resources; The network controller configures reserved network resources in the network devices on the fixed forwarding path, and configures periodic scheduling mechanism parameters aligned with the synchronization time, wherein the periodic scheduling mechanism is selected from time-based gated scheduling or circular queue forwarding; The network device on the fixed forwarding path controls the ultrasonic data stream to be forwarded along the fixed forwarding path on the network device according to a predetermined time rule according to the configured reserved network resources and periodic scheduling mechanism parameters, so as to provide data transmission with a predetermined delay upper limit and a predetermined jitter upper limit; The remote computing platform includes a distributed computing cluster consisting of multiple edge computing nodes and a central cloud node; The step of receiving the ultrasound data transmitted via the deterministic network function in the remote computing platform, performing ultrasound data processing on the received ultrasound data according to a preset processing strategy, and generating diagnosis auxiliary data for ultrasound-assisted diagnosis comprises: The distributed computing power cluster receives ultrasound data from different ultrasound probes transmitted via deterministic network functions; The mixed load computing power scheduler in the distributed computing power cluster receives a processing task request corresponding to each ultrasonic data, wherein the processing task request includes a task type, required computing resources, and QoS requirements; The hybrid load computing scheduler queries the real-time availability status of heterogeneous computing resources of each edge computing node in the distributed computing cluster; The hybrid load computing power scheduler allocates and schedules each ultrasonic data processing task to each specific edge computing node in the distributed computing power cluster for execution according to the processing task request corresponding to each ultrasonic data, the real-time available status of the heterogeneous computing resources of each edge computing node, and the preset scheduling strategy; Each specific edge computing node performs ultrasonic data processing on the ultrasonic data corresponding to the ultrasonic data processing task according to a preset processing strategy.

2. The ultrasound-assisted diagnosis method according to claim 1, characterized in that: The steps of the millimeter wave communication module performing authentication and association processes with the selected target millimeter wave access point to establish a wireless link include: The millimeter wave communication module performs a predefined authentication protocol with the selected target millimeter wave access point to verify the identities of both parties and establish a security context; The millimeter wave communication module sends an association request to the target millimeter wave access point, and establishes a logical connection after receiving an association response; The millimeter wave communication module exchanges beamforming training information containing antenna direction information with the target millimeter wave access point, and each calculates and adjusts the weight coefficient of the signal phase and / or amplitude of its respective millimeter wave phased array antenna according to the received beamforming training information to form a directional wireless link pointing to the other party.

3. The ultrasound-assisted diagnosis method according to claim 1, characterized in that: The step of sending the up-converted millimeter wave signal to the target millimeter wave access point through the established wireless link by the millimeter wave communication module further includes: The millimeter wave communication module continuously monitors the wireless link quality parameters with the target millimeter wave access point currently connected, and periodically or based on a trigger event scans the signal quality parameters of other nearby available millimeter wave access points; When a radio link quality parameter between the millimeter wave communication module and the target millimeter wave access point currently connected is lower than a preset switching threshold and a signal quality parameter of at least one adjacent available millimeter wave access point is better than a current radio link quality parameter, the millimeter wave communication module selects one with the best signal quality from the adjacent available millimeter wave access points as a new target millimeter wave access point; The millimeter wave communication module executes a fast handover protocol with the new target millimeter wave access point to establish a new wireless link, wherein the fast handover protocol includes a fast authentication and association process.

4. The ultrasound-assisted diagnosis method according to claim 1, characterized in that: When the periodic scheduling mechanism is selected from time-based gating scheduling, the step of configuring periodic scheduling mechanism parameters aligned with synchronization time includes: The network controller calculates and issues a gating list for the corresponding egress port of each network device on the fixed forwarding path of the ultrasonic data flow, wherein the gating list specifies the precise time points at which the gates of the queues corresponding to the ultrasonic data flow are opened and closed in each scheduling cycle; The step of controlling the ultrasonic data stream to be forwarded along a fixed forwarding path on the network device according to a predetermined time rule comprises: The network device opens the gate of the queue only within a time window specified by the gating list to allow the ultrasound data flow to be forwarded along a fixed forwarding path.

5. The ultrasound-assisted diagnosis method according to claim 1, characterized in that: When the periodic scheduling mechanism is selected from circular queue forwarding, the step of configuring periodic scheduling mechanism parameters aligned with the synchronization time includes: Configuring a forwarding period for the network device; The step of controlling the ultrasonic data stream to be forwarded along a fixed forwarding path on the network device according to a predetermined time rule comprises: The network device caches the received ultrasonic data stream, and delays sending the cached ultrasonic data stream along the fixed forwarding path until a specific time point in the next or subsequent predetermined forwarding period.

6. The ultrasound-assisted diagnosis method according to claim 1, characterized in that: The step of performing ultrasound data processing on the received ultrasound data according to a preset processing strategy to generate diagnosis auxiliary data for ultrasound-assisted diagnosis and providing the data to a display device for display comprises: The edge computing node cluster performs all required ultrasound data processing on the received ultrasound data according to a preset processing strategy, and sends the generated diagnostic auxiliary data for ultrasound-assisted diagnosis to the central cloud node, and the central cloud node manages the received diagnostic auxiliary data and provides it to a display device for display; or The edge computing node cluster performs preliminary ultrasound data processing on the received ultrasound data according to a preset processing strategy, generates intermediate processing result data and sends it to the central cloud node. The central cloud node generates diagnostic auxiliary data for ultrasound-assisted diagnosis based on the received intermediate processing results, and provides it to the display device for display.

7. An ultrasound-assisted diagnosis system applied to the ultrasound-assisted diagnosis method according to any one of claims 1 to 6, characterized in that: The system comprises: at least one ultrasound probe for acquiring ultrasound data; a millimeter wave communication module physically separated from the ultrasound probe and connected via a wired link, for transmitting ultrasound data collected by the ultrasound probe using millimeter wave wireless communication technology; A plurality of millimeter wave access points deployed in the operating environment, for receiving ultrasonic data sent by the millimeter wave communication module; a wired bearer network configured to support deterministic network functions, for connecting the millimeter wave access point with a remote computing platform and transmitting the received ultrasonic data to the remote computing platform to provide data transmission with a predetermined upper limit of delay and a predetermined upper limit of jitter; and The remote computing platform is used to receive ultrasound data transmitted by the wired bearer network through the deterministic network function, and perform ultrasound data processing on the received ultrasound data according to a preset processing strategy, generate diagnostic auxiliary data for ultrasound-assisted diagnosis, and provide it to a display device for display.

Images