Medical image low-delay transmission method under 5G network

Abstract

The invention provides a medical image low-delay transmission method under a 5G network, the method is applied to a medical image transmission system, and the method comprises the following steps: an image data acquisition terminal acquires medical image data, and preprocesses the medical image data to obtain data to be transmitted; the image data acquisition terminal transmits to-be-transmitted data to the heterogeneous 5G cellular network; based on the 5G network slicing technology, the heterogeneous 5G cellular network allocates network resources to obtain an allocation result, and the allocation result is used for representing a transmission channel of the medical image data; the transmission channel is used for performing low-delay transmission on the medical image data; and the heterogeneous 5G cellular network transmits the to-be-transmitted data to the remote terminal through the transmission channel based on the distribution result. And a 5G network slicing technology is used for customizing a channel for medical image transmission, so that real-time transmission of medical images can be realized. In a remote surgery scene, a doctor can quickly make an operation decision according to a real-time image, so that the timeliness and the accuracy of the surgery are greatly improved.

Description

Technical Field

[0001] The present invention relates to the field of medical image processing, and in particular to a low-latency transmission method for medical images under a 5G network. Background Art

[0002] With the development and emergence of wireless technology, medical staff can treat patients remotely. However, for remote surgery, the transmission process of medical images cannot meet the requirements of larger capacity connections, higher data transmission speeds and most importantly, low latency. Summary of the invention

[0003] Based on this, the purpose of the present invention is to provide a low-latency transmission method for medical images under a 5G network. By utilizing 5G network slicing technology to allocate network resources, the transmission delay of medical imaging data can be reduced.

[0004] In a first aspect, the present invention provides a low-latency transmission method for medical images under a 5G network, which is applied to a medical image transmission system, the medical image transmission system comprising: an image data acquisition terminal, a remote terminal and a heterogeneous 5G cellular network; the image data acquisition terminal is connected to the heterogeneous 5G cellular network through a first communication module connected to the 5G network; the remote terminal is connected to the heterogeneous 5G cellular network through a second communication module connected to the 5G network;

[0005] The method includes:

[0006] The image data acquisition terminal acquires medical image data, pre-processes the medical image data, and obtains data to be transmitted;

[0007] The image data acquisition terminal transmits the data to be transmitted to the heterogeneous 5G cellular network;

[0008] Based on 5G network slicing technology, heterogeneous 5G cellular networks allocate network resources and obtain allocation results, which are used to characterize the transmission channels of medical imaging data; the transmission channels are used to transmit medical imaging data with low latency;

[0009] The heterogeneous 5G cellular network transmits the data to be transmitted to the remote terminal through the transmission channel based on the allocation result.

[0010] The low-latency transmission method for medical images under the 5G network described in the present invention uses 5G network slicing technology to customize channels for medical image transmission, reducing the time for data queuing and processing in the network, and can achieve real-time transmission of medical images. In remote surgery scenarios, doctors can make quick operational decisions based on real-time images, greatly improving the timeliness and accuracy of surgery.

[0011] In a second aspect, the present invention provides a medical image transmission system, the system comprising:

[0012] Image data acquisition terminal, used to acquire medical image data, pre-process the medical image data, obtain data to be transmitted; and transmit the data to be transmitted to the heterogeneous 5G cellular network;

[0013] A heterogeneous 5G cellular network is used to allocate network resources based on 5G network slicing technology to obtain an allocation result, and the allocation result is used to characterize a transmission channel for medical imaging data; the transmission channel is used to transmit the medical imaging data with low latency; based on the allocation result, the data to be transmitted is transmitted to a remote terminal through the transmission channel;

[0014] A remote terminal for receiving data to be transmitted;

[0015] Among them, the image data acquisition terminal is connected to the heterogeneous 5G cellular network through its first communication module connected to the 5G network; the remote terminal is connected to the heterogeneous 5G cellular network through its second communication module connected to the 5G network.

[0016] For better understanding and implementation, the present invention is described in detail below with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 The present invention provides a flow chart of a low-latency transmission method for medical images under a 5G network;

[0018] Figure 2 A schematic diagram of a method for allocating resources in an eMBB slice according to an embodiment of the present invention;

[0019] Figure 3 The figure is a flow chart of a resource allocation method in a uRLLC slice according to an embodiment of the present invention. DETAILED DESCRIPTION

[0020] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0021] refer to Figure 1, which shows a flow chart of a low-latency transmission method for medical images under a 5G network provided by an embodiment of the present invention. The method is applied to a medical image transmission system, which includes: an image data acquisition terminal, a remote terminal and a heterogeneous 5G cellular network; the image data acquisition terminal is connected to the heterogeneous 5G cellular network through a first communication module connected to the 5G network; the remote terminal is connected to the heterogeneous 5G cellular network through a second communication module connected to the 5G network; the method includes the following steps:

[0022] S101. An image data acquisition terminal acquires medical image data, pre-processes the medical image data, and obtains data to be transmitted.

[0023] Specifically, the image data acquisition terminal here can be an imaging device used for emergency surgery far away from the hospital, such as a portable ultrasound device, a portable X-ray machine, an electrocardiograph, an endoscope, a high-definition image transmission device installed on a surgical robot, etc. After acquiring these medical image data, they are preprocessed, which may include the following aspects:

[0024] Image compression: Reduce the amount of data through appropriate compression algorithms for more efficient transmission while ensuring that the image quality is within an acceptable range.

[0025] Format conversion: Convert medical imaging data into a format suitable for transmission over the network to ensure compatibility and stability.

[0026] Data encryption: In order to protect patient privacy and the security of medical data, the imaging data may be encrypted. After preprocessing, the data to be transmitted is obtained.

[0027] S102. The image data acquisition terminal transmits the data to be transmitted to the heterogeneous 5G cellular network.

[0028] Specifically, the first communication module of the image data acquisition terminal establishes a connection with the heterogeneous 5G cellular network. This communication module can adopt advanced 5G communication technology standards and has high-speed data transmission capabilities and low latency characteristics. It can ensure a stable connection with the heterogeneous 5G cellular network through a series of signal interactions and authentication processes. After the connection is established, the image data acquisition terminal begins to prepare to transmit data. It can group and encapsulate the pre-processed data to be transmitted so that it meets the format requirements of network transmission. These data packets can contain data identification information, checksums, etc., so as to detect and correct errors during transmission. Then, the image data acquisition terminal can select a suitable transmission method according to the network conditions and transmission protocols. The 5G network supports multiple transmission modes, such as unicast, multicast, and broadcast. For the transmission of medical image data, unicast is usually used to send data directly to a specific target address, that is, a node in the heterogeneous 5G cellular network.

[0029] S103. Based on 5G network slicing technology, the heterogeneous 5G cellular network allocates network resources to obtain allocation results, and the allocation results are used to characterize the transmission channel of medical imaging data; the transmission channel is used to transmit the medical imaging data with low latency.

[0030] Specifically, 5G network slicing is a key technology that allows network operators to create multiple virtual, independent logical networks on a physical 5G network infrastructure. It is like cutting a large physical network "cake" into multiple "slices" with different characteristics, each of which can be customized according to specific business needs.

[0031] Heterogeneous 5G cellular networks first need to identify the various network resources they have. These resources include but are not limited to wireless spectrum resources, base station processing capabilities, transmission link bandwidth, etc. For example, wireless spectrum in different frequency bands differs in propagation characteristics and available bandwidth, and the network needs to have a clear understanding of the distribution and status of these resources. At the same time, the current usage of these resources should be evaluated. This can be achieved through network monitoring tools and algorithms, for example, understanding the current load of each base station, including the number of connected users, the amount of data being transmitted, etc., in order to determine how many resources are available for allocation.

[0032] The demand for medical image transmission focuses on low latency and high bandwidth. Low latency requires the network to quickly transmit image data from the image data acquisition terminal to the remote terminal. This means that the network needs to allocate sufficient resources to reduce the queuing and processing time of data in the network. The high bandwidth requirement is because medical image data usually has a large amount of data, such as high-definition CT images, MRI images, etc. Take an ordinary chest CT image as an example, its data volume may reach tens of MB or even higher. Therefore, sufficient bandwidth is required to ensure that data can be transmitted quickly and avoid jamming during transmission.

[0033] Based on the needs of medical image transmission, heterogeneous 5G cellular networks have begun to create dedicated network slices. This slice will be given specific resources, such as allocating a certain frequency band of wireless spectrum for data transmission. For example, a higher frequency band is allocated to the medical image transmission slice. These frequency bands usually have higher transmission rates, but the propagation distance is relatively short. However, they are more suitable for telemedicine scenarios within hospitals or at short distances.

[0034] In terms of base station processing capabilities, certain computing resources will be reserved for this slice for data processing and forwarding. For example, when image data arrives at the base station, the base station can quickly perform necessary processing on the data, such as checking data integrity and performing simple decryption operations. In terms of transmission link bandwidth, sufficient bandwidth will be allocated to the medical image transmission slice to ensure that the data can pass through the network at a higher speed. This is like opening up a "highway" for medical image data, allowing data to be transmitted unimpeded.

[0035] S104. Based on the allocation result, the heterogeneous 5G cellular network transmits the data to be transmitted to the remote terminal through the transmission channel.

[0036] Specifically, the heterogeneous 5G cellular network encapsulates the acquired data to be transmitted and adds verification information based on the allocation results. Data is quickly relayed between base stations to ensure that the data reaches the remote terminal quickly. After the doctor's remote control terminal receives the data, it can perform remote surgery based on the data in real time. For example, when the surgical image data reaches the remote terminal used by the doctor to control the surgical robot through the transmission channel, the second communication module of the remote terminal will respond quickly and receive the data. The doctor can view the high-definition images sent back by the surgical robot in real time at the remote terminal, just like being at the surgical site. These images provide doctors with an accurate surgical field of view, allowing doctors to accurately control the surgical robot to perform various operations. For example, doctors can adjust the movements, angles, and strength of the surgical robot through the control interface of the remote terminal, and make timely decisions based on real-time image feedback to ensure the safety and success of the operation.

[0037] Using the above method, 5G network slicing technology is used to customize channels for medical image transmission, which reduces the time for data to queue and process in the network and enables real-time transmission of medical images. In remote surgery scenarios, doctors can make quick operational decisions based on real-time images, greatly improving the timeliness and accuracy of surgery.

[0038] In an optional embodiment, a heterogeneous 5G cellular network includes a cloud wireless access network, a plurality of virtual machines, and a network slicing module; the cloud wireless access network is grouped into a plurality of sub-cells by a distributed plurality of remote wireless head ends, the sub-cells are served by a macro base station, and the plurality of remote wireless head ends and the plurality of macro base stations are concentrated in a cloud single baseband unit pool through an optical fronthaul link; a virtual machine is created in the cloud single baseband unit pool, the virtual machine is used to accommodate network wireless resources, and each virtual machine serves as a virtual single baseband unit for each data flow; the network slicing module is located in the cloud single baseband unit pool, the network slicing module is connected to the plurality of virtual single baseband units, and the network slicing module is used to divide the network;

[0039] In an optional embodiment, step S103, based on the 5G network slicing technology, the heterogeneous 5G cellular network allocates network resources to obtain an allocation result, and the allocation result is used to characterize the transmission channel of the medical imaging data, specifically including the following steps:

[0040] The remote wireless head end in the cloud wireless access network receives the medical imaging data and sends the medical imaging data to the cloud single baseband unit pool.

[0041] The network slicing module in the cloud single baseband unit pool classifies the medical imaging data into 4K video data and other medical data as the classification result; according to the classification result, the type of medical imaging data is sliced ​​and processed; among them, the resource allocation method of maximizing video throughput is adopted to allocate 4K video data to eMBB slices, and the resource allocation method of high reliability and low latency transmission is adopted to allocate other medical data to uRLLC slices as the allocation result.

[0042] Specifically, the remote wireless head end in the cloud wireless access network receives medical imaging data, which may come from an imaging data acquisition terminal, such as an imaging acquisition device of a surgical robot. The medical imaging data is then sent to the cloud single baseband unit pool to prepare for subsequent processing and distribution. The network slicing module in the cloud single baseband unit pool classifies the medical imaging data into 4K video data and other medical data to obtain classification results. This is because different types of data have different requirements for the network and need to be processed in a targeted manner. According to the classification results, the resource allocation method that maximizes video throughput is used to allocate 4K video data to eMBB (enhanced mobile broadband) slices. This is because 4K video data requires high bandwidth to ensure smooth playback, and eMBB slices can provide high-speed data transmission capabilities to meet the needs of 4K video. The resource allocation method of high-reliability and low-latency transmission is used to allocate other medical data to uRLLC (ultra-reliable low-latency communication) slices. Other medical data may have higher requirements for latency and reliability, and uRLLC slices can provide low-latency and high-reliability network services to ensure timely transmission and accuracy of medical data. The advantage of this method is that it can perform targeted network resource allocation based on the characteristics of different types of medical imaging data, improve network efficiency and service quality, and meet the needs of low-latency transmission of medical images.

[0043] refer to Figure 2 In an optional embodiment, a resource allocation method for maximizing video throughput is used to allocate 4K video data to eMBB slices, including the following steps:

[0044] S201. Acquire 4K video data, and determine maximizing the video throughput of the 4K video data as an optimization problem with the goal.

[0045] S202. Construct an objective function based on user parameters related to the eMBB slice and network parameters of the heterogeneous 5G cellular network; the objective function represents an optimization goal; the optimization goal is to maximize the video throughput of users related to the eMBB slice.

[0046] S203: Solve the optimization problem using a dual decomposition method to obtain a first resource allocation solution.

[0047] S204. Allocate 4K video data to eMBB slices according to the optimal resource allocation solution.

[0048] The dual decomposition method includes the following steps: constructing a Lagrangian function according to the objective function; derivatizing the Lagrangian function to obtain the KKT condition; and determining the first resource allocation scheme according to the KKT condition.

[0049] Specifically, the heterogeneous 5G cellular network obtains a set L of remote wireless heads in uplink and downlink, a set U of permitted users associated with the remote wireless heads l and the set of subchannels K, and determine the fronthaul capacity C of the remote wireless head fh , virtual machine computing power C c,v , the user's minimum data rate requirement R min,u , the transmission power budget P of the remote wireless head end max,l and the tolerable interference threshold I on the subchannel th ;

[0050] For user u related to the eMBB slice, an optimization problem is constructed, and the objective function is:

[0051]

[0052] Where W(u) is the user weight function associated with the slice; w lku An indication indicating whether subchannel k is allocated to user u associated with a remote wireless head; p lku is the power transmitted by the remote wireless head end for user u on subchannel k; R lku is the achievable data rate between user u and the remote wireless head on subchannel k, R lku =B k log2(1+γ lku ); γ lku is the user's signal-to-interference-plus-noise ratio, B k is the subchannel bandwidth; g lku is the channel gain between the remote wireless head and the user, σ 2 is the noise power.

[0053] This optimization problem is subject to the following constraints:

[0054] The traffic of each remote radio head does not exceed its fronthaul capacity:

[0055]

[0056] The achievable data rate per user is limited by the computing power of the virtual machine:

[0057]

[0058] The achievable data rate for each user meets its minimum rate requirement:

[0059]

[0060] The transmit power of each remote radio head is limited:

[0061]

[0062] The interference on each subchannel is limited by a tolerable threshold:

[0063]

[0064] Each remote radio head end allocates no more than one subchannel to the same user:

[0065]

[0066] The transmission power of the remote radio head is specified to be non-negative and the subchannel allocation is binary:

[0067]

[0068] In a multi-carrier system with a sufficiently large number of subchannels, the duality gap for resource allocation is almost zero. Therefore, in order to solve the above optimization problem, we assume that the number of subchannels involved in our system is very large, and then we use the duality decomposition method with the duality gap equal to 0 to solve the above optimization problem, including the following steps:

[0069] According to the objective function, the Lagrangian function is constructed:

[0070]

[0071] Among them, α l , β lu , lu , μ l , τ lk and η lk is the Lagrange multiplier corresponding to the above constraints;

[0072] The corresponding terms in the above Lagrangian function can be omitted and expressed as:

[0073]

[0074] And the dual optimization problem is formulated as:

[0075]

[0076] subjecttoα,β,λ,μ,τ,η≥0

[0077] For u∈U l , first assume that w lku =1, find the constructed Lagrangian function with respect to p lku The derivative of , we get the following KKT condition:

[0078]

[0079] w lku p lku (ψ lku -μ l )=0

[0080] in,

[0081]

[0082] For all u∈U l , l∈L and k∈K, according to the KKT condition, the optimal power allocation is obtained:

[0083]

[0084] Next, find the constructed Lagrangian function with respect to w lku The derivative of , we get the following KKT condition:

[0085]

[0086] w lku (Γ lku -η lku )=0

[0087] in,

[0088]

[0089] According to the KKT condition, if w lku =1, we get Γ lku -η lku =0; if w lku =0, we get Γ lku -η lku ≤0. So we can deduce that all users satisfy u∈U l , l∈L and k∈K, with Γ lku The user corresponding to the maximum value of should be assigned a subchannel. Assuming that a subchannel can only be assigned to one of the users associated with the same remote wireless head at the same time, the optimal subchannel allocation can be obtained by the following formula:

[0090]

[0091] refer to Figure 3 In an optional embodiment, a resource allocation method of high-reliability and low-latency transmission is used to allocate other medical data to uRLLC slices, including the following steps:

[0092] S301. Obtain other medical data, transmission requirement parameters of users related to other medical data in uRLLC slices, and system parameters of heterogeneous 5G cellular networks.

[0093] S302. Based on the data transmission delay calculation formula and according to the system parameters, the data propagation delay is obtained; and the error probability of the heterogeneous 5G cellular network during data transmission is obtained.

[0094] S303. Based on the effective bandwidth calculation formula, according to the transmission demand parameters, data propagation delay and error probability, determine the user's minimum data packet service rate data; the minimum data packet service rate data is a quantitative indicator of the minimum number of data packets successfully transmitted per unit time.

[0095] S304: Determine a second resource allocation scheme according to the minimum data packet service rate data.

[0096] S305. Allocate other medical data to uRLLC slices according to the second resource allocation scheme.

[0097] Specifically, in the uRLLC traffic, consider the total delay of user u receiving data at the hospital end. For uRLLC user u, its total delay Should meet:

[0098]

[0099] in, is the transmission delay, Δf (i,u) is the subcarrier spacing allocated to a user to meet its delay requirement, is the number of symbols with a specific subcarrier spacing i in a specific transmission time interval at the remote wireless head; is the fronthaul delay, is the distance between the cloud single baseband unit pool and the remote wireless head end; is the propagation delay, is the distance between the remote wireless headend and user u; is the queuing delay, which is the delay caused by queuing packets in the cloud single baseband unit pool; is the processing delay, which is the delay caused by processing in the cloud single baseband unit pool, and the processing delay is a known constant. The above parameters can be obtained through the network system.

[0100] Reliability constraints also need to be considered. The overall packet loss probability (OPLP) should satisfy where ε u is the OPLP bound or overall error probability bound, is the queue delay violation probability, is the packet drop probability, is the decoding error probability.

[0101] In order to meet the above delay and reliability constraints, the effective bandwidth is used to determine the minimum data packet service rate. The effective bandwidth formula is:

[0102]

[0103] According to the above effective bandwidth formula, determine the minimum packet service rate of the user queue, where T u The user's transmission interval, X u is the packet arrival rate of user u in the cloud single baseband unit pool.

[0104] By analyzing EB u and The relationship between To minimize EB u , but ensure that the OPLP target is not violated while considering the decoding error probability Relationship with user packet length n u The relationship between the information m u Bit mapping to packet length n u , the decoder starts from n u The channel is used to guess that its maximum rate meets:

[0105]

[0106] Among them, m u is the number of information bits, γ u is the correlated signal-to-interference-plus-noise ratio, Q -1 It is the inverse function of the Q function, by increasing n u Can reduce

[0107] According to the minimum packet service rate of the user queue, an optimal resource allocation algorithm is proposed. The algorithm includes two stages: calculation and execution.

[0108] Calculation phase:

[0109] Step 1: is the starting point of resource allocation.

[0110] Step 2: Since the interrelationships of OPLP are mainly and Between, set:

[0111]

[0112] Step 3: Use the formula Finding Satisfaction The packet length n uSum-correlated signal-to-interference-plus-noise ratio γ u .

[0113] Step 4: Based on the found packet length n u Sum-correlated signal-to-interference-plus-noise ratio γ u renew and

[0114] Step 5: Choose the narrowest Δf (i,u) and the maximum number of And calculate the transmission delay

[0115] Step 6: Find the distance between the cloud single baseband unit pool and the selected cloud single baseband unit pool and the distance between the cloud single baseband unit pool and its associated users And calculate the forward delay separately and propagation delay

[0116] Step 7: Using the formula Finding the queue delay And using the formula Calculate the user's effective bandwidth EB.

[0117] Step 8: Make two conditional judgments:

[0118] First, the total number of bits in the single baseband unit pool from the cloud should not exceed the fronthaul capacity, i.e.

[0119]

[0120] Second, the total number of symbols in each RG in the current transmission time interval should be limited to the total number of available resource elements. Right now

[0121]

[0122] If both conditions in step 8 are met, proceed to step 9 to map the packet elements to the RBs of the virtual RGs and update

[0123] If the condition of step 8 is not met, go to step 10. Go to this step. This indicates that the current resource allocation scheme may not meet the restrictions on the fronthaul capacity and the number of resource elements, and needs to be adjusted.

[0124] Step 11: This step continues to describe the loop situation when the conditions are not met, that is, continue to check whether the resource allocation meets the system requirements, which may involve further evaluation and judgment of various parameters and conditions until a suitable adjustment plan is found to make the system meet the constraints.

[0125] Step 12: After determining in steps 10 and 11 that the users that have completed partial resource allocation need to be adjusted, consider all users k that have completed resource allocation (here k represents each individual user, 1 from u). These users are further analyzed and processed to readjust the resource allocation to meet the system's fronthaul capacity and total resource element number constraints.

[0126] Steps 13-16: Used to reduce the user's effective bandwidth EB k To EB kn , until the forwarding capacity C is met fh and the total number of resource elements By iteratively will be reduced to Will Increase to While not violating the overall ε k By increasing n k To the actual effective n kn To reduce Then reallocate resources to user k and update

[0127] Execution phase:

[0128] Step 20: For each user u (u = 1, ···, N), use a series of parameter values ​​obtained in the calculation phase, including L

[0129] (some intermediate result or parameter set related to resource allocation), n u (packet length), etc. According to the effective bandwidth EB u to serve the queue of user u. This means that according to EB u The packets in the queue are processed at the determined rate to ensure that the requirements set during the calculation phase regarding queuing delay and effective bandwidth are met.

[0130] Step 21: Precode the data packet. Precoding is a technology that processes data at the transmitting end. It can adjust the data according to the channel characteristics and system requirements to improve the reliability and efficiency of transmission. The precoded data packet will be applied to the corresponding resource grids in the manner determined in the calculation stage. These resource grids are defined based on OFDM (Orthogonal Frequency-Division Multiplexing) technology and are related to resource allocation such as subcarriers and symbols. Specifically, it is based on the calculation stage, such as (the number of symbols in a specific transmission time interval at the remote wireless head) and Δf (i,u) (subcarrier spacing allocated to users) and other parameters to place the data packet at the appropriate resource location.

[0131] Step 22: This step completes the entire execution phase. In this step, all data packets processed in the previous steps have been correctly placed in the corresponding resource grid and processed and transmitted according to the predetermined rules. This ensures that the medical data in the uRLLC slice can be transmitted in a manner that meets the delay and reliability constraints, thereby achieving the goal of the entire algorithm, that is, improving the transmission reliability of medical data in the uRLLC slice and reducing delay.

[0132] In an optional embodiment, preprocessing the medical image data to obtain data to be transmitted includes the following steps:

[0133] S401, dividing the medical images into single-frame images and multi-frame images, and marking the single-frame images and multi-frame images in chronological order.

[0134] S402 , extracting a random data frame from multiple frames of images, and if the extracted data frame is the first frame of data, performing image encoding on the data frame to obtain a decoded frame of the first frame of data.

[0135] S403: If the extracted data frame is not the first frame data, a decoding frame of the data frame is determined according to the video data based on the data compression network.

[0136] S404: The image data acquisition terminal compresses the compressed multi-frame images and the single-frame image respectively according to the decoded frame to obtain compressed data as data to be transmitted.

[0137] Specifically, the image data acquisition terminal divides the medical image data into single-frame images and multi-frame images, and marks them in time sequence. The purpose of this is to better organize and process different types of image data. A single-frame image may be a static image, while a multi-frame image may be a dynamic video sequence. By marking the time sequence, the temporal relationship of the image can be better maintained in subsequent processing. For multi-frame images, random data frames are extracted for processing. If the first frame data is extracted, image encoding is performed on it to obtain a decoded frame of the first frame data. The first frame data is usually of special importance because it can provide a reference for subsequent frames. By encoding the first frame data, the amount of data can be reduced, and it also provides a basis for subsequent decoding. If the extracted data frame is not the first frame data, the decoded frame of the data frame is determined based on the video data based on the data compression network. The data compression network can use the correlation and redundancy of the video data to compress non-first frame data more efficiently. By referring to the features of the previous frame and video data, the content of the current frame can be predicted, thereby reducing the amount of data. The image data acquisition terminal compresses the compressed multi-frame images and single-frame images according to the decoded frames to obtain compressed data as the data to be transmitted. This step will perform the final compression on the processed single-frame images and multi-frame images to reduce the amount of data and improve the transmission efficiency. The compressed data can be transmitted faster through the network, reducing delays and bandwidth usage.

[0138] In an optional embodiment, preprocessing the medical image data to obtain data to be transmitted includes the following steps:

[0139] S501. The image data acquisition terminal divides the medical image data into a plurality of data blocks of 128 bytes in size.

[0140] S502: The image data acquisition terminal selects an initial vector, and the length of the initial vector matches the data block size.

[0141] S503: The image data acquisition terminal performs an XOR operation on the initial vector and the first data block of the medical image data to obtain a first XOR result data block.

[0142] S504: The image data acquisition terminal uses the AES algorithm to encrypt the first XOR result data block to obtain a first encrypted data block.

[0143] S505. For the second and subsequent data blocks of the medical image data, before encryption, the image data acquisition terminal performs an XOR operation on the encrypted data block obtained after encrypting the previous data block to obtain an XOR result data block of the current data block.

[0144] S506. The image data acquisition terminal uses the AES algorithm to encrypt the XOR result data block of the current data block to obtain an encrypted data block of the current data block; and so on to obtain multiple encrypted data blocks as data to be transmitted.

[0145] Specifically, the image data acquisition terminal divides the medical image data into multiple data blocks of 128 bytes in size. The purpose of this is to divide the larger medical image data into smaller units for subsequent processing and encryption operations. The image data acquisition terminal selects an initial vector, the length of which matches the data block size, that is, also 128 bytes. The initial vector plays an important role in the encryption process, which can increase the randomness and security of encryption. Different initial vectors will lead to different encryption results, and even for the same medical image data, the encryption strength can be improved. The image data acquisition terminal performs an XOR operation on the initial vector and the first data block of the medical image data to obtain the first XOR result data block. The XOR operation is a simple and effective logical operation that can perform bit-level operations on two data of the same length to generate a new data block. In the encryption process, the XOR operation can increase the complexity of the data and make the data more difficult to crack. The image data acquisition terminal uses the AES (Advanced Encryption Standard) algorithm to encrypt the first XOR result data block to obtain the first encrypted data block. The AES algorithm is a widely used symmetric encryption algorithm with high security and efficient encryption speed. By using the AES algorithm to encrypt the XOR result data block, the security of the data can be further improved to ensure that the medical imaging data is not stolen or tampered with during the transmission process. For the second and subsequent data blocks of the medical imaging data, before encryption, the imaging data acquisition terminal performs an XOR operation on the encrypted data block obtained after the current data block is encrypted with the previous data block to obtain the XOR result data block of the current data block. Then, the XOR result data block of the current data block is encrypted using the AES algorithm to obtain the encrypted data block of the current data block. In this way, all data blocks are processed to obtain multiple encrypted data blocks. This process is a chain encryption method, and the encryption of each data block depends on the encryption result of the previous data block. Doing so can increase the complexity and security of encryption. Even if an attacker cracks a data block, it is difficult to infer the content of other data blocks. At the same time, this method can also ensure the integrity of the data, because if any data block is tampered with during the transmission process, the decryption of subsequent data blocks will fail. Finally, these encrypted data blocks are prepared as data to be transmitted through the heterogeneous 5G cellular network. Through this preprocessing method, the security of medical imaging data can be effectively protected, ensuring the privacy of patients and the integrity of medical data.

[0146] In an optional embodiment, obtaining medical image data includes the steps of obtaining an original medical image, extracting a key portion of the original medical image, and obtaining the medical image data.

[0147] In an optional embodiment, the medical image data is a picture, and extracting the key part of the original medical image to obtain the medical image data includes the following steps:

[0148] S601, based on a threshold segmentation method, a threshold is determined according to the gray value distribution of the image, and the image is divided into different regions; according to the pixel distribution in different gray value intervals, pixels whose gray value difference is less than or equal to 10 are divided into the same region.

[0149] S602, extracting areas whose grayscale values ​​differ from those of surrounding areas by more than 10, and integrating all extracted areas to obtain the medical imaging data.

[0150] Specifically, when processing medical imaging data in the form of images, the threshold-based segmentation method is first used. The core of this method is to determine a suitable threshold value based on the grayscale value distribution of the image. The grayscale value reflects the brightness of each pixel in the image, and different tissues or structures often present different grayscale value characteristics in the image. By analyzing the grayscale values ​​of all pixels in the entire image, a threshold value that can better divide the image into different regions can be determined. For example, if the image mainly contains two parts, brighter and darker, then a grayscale value can be selected so that the pixels with higher brightness are divided into one region, and the pixels with lower brightness are divided into another region. After determining the threshold, a more detailed regional division is further performed according to the pixel distribution of different grayscale value intervals. Specifically, pixels with a grayscale value difference of less than or equal to 10 are divided into the same region. This is because in actual medical images, pixels with similar grayscale values ​​usually have similar characteristics and may belong to different parts of the same tissue or structure. In this way, the image can be further subdivided into multiple relatively uniform regions to prepare for the subsequent extraction of key parts. After completing the regional division, the key part is extracted. The areas whose grayscale values ​​differ from those of the surrounding areas by more than 10 are extracted. These areas often have obvious characteristic differences and may be the key areas that doctors pay attention to in diagnosis, such as diseased tissues, abnormal structures, etc. After extracting these areas, they are integrated. The integration process may include operations such as merging pixels in these areas and adjusting boundaries to ensure that the obtained medical imaging data is a coherent whole. After integration, the processed medical imaging data is obtained, which highlights the key parts of the original image, removes some unnecessary background information and noise, and is more conducive to subsequent transmission, analysis and diagnosis.

[0151] In an optional embodiment, the medical image data is a video, and extracting a key part of the original medical image to obtain the medical image data includes the following steps:

[0152] S701, based on the shot boundary detection method, based on the image changes in the video, determining the shot switching point in the video, taking the frames near the shot switching point as key frames, and obtaining a plurality of key frames sorted in chronological order;

[0153] S702, based on the threshold segmentation method, the threshold is determined according to the gray value distribution of the key frame, and the key frame is divided into different regions; according to the pixel distribution in different gray value intervals, the pixels whose gray value difference is less than or equal to 10 are divided into the same region;

[0154] S703, extracting areas whose grayscale values ​​differ from those of surrounding areas by more than 10, and integrating all extracted areas to obtain a new key frame;

[0155] S704 . According to the time sequence, a linear interpolation algorithm is used to combine multiple key frames sorted in time sequence to obtain medical image data.

[0156] Specifically, when the medical imaging data is a video, a method based on shot boundary detection is first used. In a video, the shot switching point is usually accompanied by obvious changes in the image, such as a sudden change in the content of the picture, a drastic change in brightness, etc. By analyzing the image changes in the video, the shot switching point in the video can be accurately determined. The frames near the shot switching point are taken as key frames. These key frames can represent the important content of different shots in the video and are sorted in chronological order. The extraction of key frames can greatly reduce the amount of data for subsequent processing, while also retaining the key information of the video. For the extracted key frames, a threshold-based segmentation method similar to image processing is used. According to the gray value distribution of the key frame, a suitable threshold is determined to divide the key frame into different regions. At the same time, according to the pixel distribution of different gray value intervals, pixels with a gray value difference of less than or equal to 10 are divided into the same region. In this way, the regional division of the key frame can be further refined so that pixels with similar gray value characteristics are classified into one category. After the key frame is regionally divided, the region with a gray value difference of more than 10 from the surrounding area is extracted. These areas are often the key parts in the keyframes and may contain important medical information, such as lesion areas, abnormal structures, etc. All the extracted areas are integrated to obtain new keyframes. This new keyframe highlights the key parts in the original keyframe and removes some unnecessary background information and noise. According to the time sequence, a linear interpolation algorithm is used to combine multiple keyframes sorted in time sequence. The linear interpolation algorithm can generate intermediate frames between keyframes based on the time relationship and image features between adjacent keyframes, making the video smoother. In this way, the processed keyframes are combined into new medical imaging data. This medical imaging data retains the key information in the original video while reducing the amount of data, which is more conducive to low-latency transmission in the network and subsequent analysis and diagnosis.

[0157] The present invention also provides a medical image transmission system, the system comprising:

[0158] Image data acquisition terminal, used to acquire medical image data, pre-process the medical image data, obtain data to be transmitted; and transmit the data to be transmitted to the heterogeneous 5G cellular network;

[0159] A heterogeneous 5G cellular network is used to allocate network resources based on 5G network slicing technology to obtain an allocation result, and the allocation result is used to characterize a transmission channel for medical imaging data; the transmission channel is used to transmit the medical imaging data with low latency; based on the allocation result, the data to be transmitted is transmitted to a remote terminal through the transmission channel;

[0160] A remote terminal for receiving data to be transmitted;

[0161] Among them, the image data acquisition terminal is connected to the heterogeneous 5G cellular network through its first communication module connected to the 5G network; the remote terminal is connected to the heterogeneous 5G cellular network through its second communication module connected to the 5G network.

[0162] The above-mentioned embodiments only express several implementation methods of the present invention, and the description is relatively specific and detailed, but it cannot be understood as limiting the scope of the invention patent. It should be pointed out that for ordinary technicians in this field, several modifications and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention.

Claims

1. A low-latency transmission method for medical images under a 5G network, characterized in that: Applied to a medical image transmission system, the medical image transmission system comprises: an image data acquisition terminal, a remote terminal and a heterogeneous 5G cellular network; the image data acquisition terminal is connected to the heterogeneous 5G cellular network through a first communication module connected to the 5G network; the remote terminal is connected to the heterogeneous 5G cellular network through a second communication module connected to the 5G network; The method comprises: The image data acquisition terminal acquires medical image data, pre-processes the medical image data, and obtains data to be transmitted; The image data acquisition terminal transmits the data to be transmitted to the heterogeneous 5G cellular network; Based on the 5G network slicing technology, the heterogeneous 5G cellular network allocates network resources to obtain an allocation result, and the allocation result is used to characterize a transmission channel for the medical imaging data; the transmission channel is used to perform low-latency transmission of the medical imaging data; Based on the allocation result, the heterogeneous 5G cellular network transmits the data to be transmitted to the remote terminal through the transmission channel.

2. The method according to claim 1, characterized in that The heterogeneous 5G cellular network includes a cloud wireless access network, multiple virtual machines and a network slicing module; the cloud wireless access network is grouped into multiple sub-cells by multiple distributed remote wireless head ends, the sub-cells are served by macro base stations, and multiple remote wireless head ends and multiple macro base stations are concentrated in a cloud single baseband unit pool through optical fronthaul links; the virtual machine is created in the cloud single baseband unit pool, the virtual machine is used to accommodate network wireless resources, and each of the virtual machines serves as a virtual single baseband unit for each data flow; the network slicing module is located in the cloud single baseband unit pool, the network slicing module is connected to multiple virtual single baseband units, and the network slicing module is used to divide the network; Based on the 5G network slicing technology, the heterogeneous 5G cellular network allocates network resources to obtain an allocation result, and the allocation result is used to characterize the transmission channel of the medical imaging data, including: The remote wireless head end in the cloud wireless access network receives the medical imaging data and sends the medical imaging data to a cloud single baseband unit pool; The network slicing module in the cloud single baseband unit pool classifies the medical imaging data into 4K video data and other medical data as the classification result; according to the classification result, the type of the medical imaging data is sliced; wherein, the 4K video data is allocated to the eMBB slice by a resource allocation method that maximizes video throughput, and the other medical data is allocated to the uRLLC slice by a resource allocation method that uses high reliability and low latency transmission as the allocation result.

3. The method according to claim 2, characterized in that The resource allocation method for maximizing video throughput is used to allocate the 4K video data to the eMBB slice, including: Acquire the 4K video data, and determine as an optimization problem the maximization of the video throughput of the 4K video data; Constructing an objective function according to user parameters related to the eMBB slice and network parameters of the heterogeneous 5G cellular network; the objective function represents an optimization goal; the optimization goal is to maximize the video throughput of users related to the eMBB slice; Solving the optimization problem using a dual decomposition method to obtain a first resource allocation solution; Allocating the 4K video data to the eMBB slice according to the optimal resource allocation solution; Wherein, the dual decomposition method includes: According to the objective function, construct a Lagrangian function; Derivative the Lagrangian function to obtain the KKT condition; A first resource allocation scheme is determined according to the KKT condition.

4. The method according to claim 2, characterized in that: The resource allocation method using high-reliability and low-latency transmission allocates the other medical data to the uRLLC slice, including: Obtaining the other medical data, transmission requirement parameters of users related to the other medical data in the uRLLC slice, and system parameters of the heterogeneous 5G cellular network; Based on the data transmission delay calculation formula and according to the system parameters, the data propagation delay is obtained; and the error probability of the heterogeneous 5G cellular network during data transmission is obtained; Based on the effective bandwidth calculation formula, according to the transmission demand parameter, the data propagation delay and the error probability, the minimum data packet service rate data of the user is determined; the minimum data packet service rate data is a quantitative indicator of the minimum number of data packets successfully transmitted per unit time; Determining a second resource allocation scheme according to the minimum data packet service rate data; The other medical data is allocated to the uRLLC slice according to the second resource allocation scheme.

5. The method according to claim 1, characterized in that The preprocessing of the medical image data to obtain data to be transmitted includes: Dividing the medical images into single-frame images and multi-frame images, and marking the single-frame images and multi-frame images in chronological order; For the multiple frames of images, a random data frame is extracted, and if the extracted data frame is the first frame of data, image encoding is performed on the data frame to obtain a decoded frame of the first frame of data; If the extracted data frame is not the first frame data, determining a decoding frame of the data frame according to the video data based on the data compression network; The image data acquisition terminal determines according to the decoded frame that the compressed multiple frames of images and the single frame of image are compressed separately to obtain compressed data as the data to be transmitted.

6. The method according to claim 1, characterized in that The preprocessing of the medical image data to obtain data to be transmitted includes: The image data acquisition terminal divides the medical image data into a plurality of data blocks of 128 bytes in size; The image data acquisition terminal selects an initial vector, the length of which matches the size of the data block; The image data acquisition terminal performs an XOR operation on the initial vector and the first data block of the medical image data to obtain a first XOR result data block; The image data acquisition terminal uses the AES algorithm to encrypt the first XOR result data block to obtain a first encrypted data block; For the second and subsequent data blocks of the medical image data, before encryption, the image data acquisition terminal performs an XOR operation on the current data block and the encrypted data block obtained after encrypting the previous data block to obtain an XOR result data block of the current data block; The image data acquisition terminal uses the AES algorithm to encrypt the XOR result data block of the current data block to obtain an encrypted data block of the current data block; and so on, to obtain multiple encrypted data blocks as the data to be transmitted.

7. The method according to claim 5 or 6, characterized in that: The obtaining of medical imaging data includes: Acquire the original medical image, extract the key part of the original medical image, and obtain the medical image data.

8. The method according to claim 1, characterized in that The medical image data is a picture, and the extracting of the key part of the original medical image to obtain the medical image data includes: A threshold-based segmentation method is used to determine a threshold according to the grayscale value distribution of the image, and divide the image into different areas; based on the pixel distribution in different grayscale value intervals, pixels with a grayscale value difference of less than or equal to 10 are divided into the same area; The area whose grayscale value difference with the surrounding area is greater than 10 is extracted, and all the extracted areas are integrated to obtain the medical imaging data.

9. The method according to claim 1, characterized in that: The medical image data is a video, and the extracting of the key part of the original medical image to obtain the medical image data includes: Based on the shot boundary detection method, based on the image changes in the video, the shot switching point in the video is determined, and the frames near the shot switching point are used as key frames to obtain a plurality of key frames sorted in chronological order; A threshold-based segmentation method is used to determine a threshold value according to the gray value distribution of the key frame, and divide the key frame into different regions; according to the pixel distribution in different gray value intervals, pixels with a gray value difference of less than or equal to 10 are divided into the same region; Extract the area whose gray value difference with the surrounding area is greater than 10, and integrate all the extracted areas to obtain a new key frame; According to the time sequence, a linear interpolation algorithm is used to combine the multiple key frames sorted in time sequence to obtain the medical image data.

10. A medical image transmission system, characterized in that: The system comprises: An image data acquisition terminal is used to acquire medical image data, pre-process the medical image data to obtain data to be transmitted; and transmit the data to be transmitted to a heterogeneous 5G cellular network; A heterogeneous 5G cellular network, for allocating network resources based on 5G network slicing technology to obtain an allocation result, wherein the allocation result is used to characterize a transmission channel for the medical imaging data; the transmission channel is used to perform low-latency transmission of the medical imaging data; based on the allocation result, the data to be transmitted is transmitted to a remote terminal through the transmission channel; A remote terminal, used for receiving the data to be transmitted; Among them, the image data acquisition terminal is connected to the heterogeneous 5G cellular network through its first communication module connected to the 5G network; the remote terminal is connected to the heterogeneous 5G cellular network through its second communication module connected to the 5G network.

Images