Multi-terminal fidelity display adaptation method for remote ultrasonic image

Abstract

The invention discloses a multi-terminal fidelity display adaptation method for remote ultrasonic images, which relates to the technical field of image processing, and comprises the following steps of: firstly, acquiring heterogeneous original video streams of ultrasonic instruments of different brands / models at a clinical end, and performing standardized analysis to generate basic video streams in a unified format; then obtaining characteristic parameters such as screen resolution, proportion and decoding capability of the target terminal, and calculating a non-stretching target rendering size according to the characteristic parameters; then different areas in the video are identified through ultrasonic image feature optimization, and differential zooming processing is carried out by adopting a corresponding interpolation algorithm; and finally, calculating a structural similarity index of the core diagnosis area, if the structural similarity index is lower than a threshold value, performing feedback adjustment, and transmitting a video stream with the fidelity reaching the standard to a terminal for display. According to the method, the heterogeneous video source is standardized, and regional differentiation zooming and fidelity feedback adjustment are carried out according to the terminal features, so that compatible adaptation between various types of ultrasonic instruments and various types of display terminals is realized.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, and more specifically to a method for adapting remote ultrasound images to multi-terminal high-fidelity display. Background Technology

[0002] Remote ultrasound diagnostic technology, as an important means to address the uneven distribution of medical resources and improve the capacity of primary healthcare services, has been widely applied and developed in recent years. Currently, real-time remote ultrasound image transmission solutions based on telecommunications networks are relatively mature, enabling the transmission of clinical ultrasound images to remote experts' mobile devices for viewing at a low cost and with high reliability. However, in practical applications, as remote consultation scenarios continue to expand and deepen, the limitations of existing technologies are gradually becoming apparent.

[0003] On the one hand, there are many brands and models of ultrasound instruments used in clinical settings. The video sources output by different manufacturers, and even different series of devices from the same manufacturer, have significant differences in parameters such as resolution, frame rate, encoding format, and color space, resulting in a large number of heterogeneous video sources.

[0004] On the other hand, the types of terminal devices used by experts in remote consultations are becoming increasingly diversified, no longer limited to a single personal computer or mobile phone, but encompassing smartphones, tablets, personal computers, and even professional medical monitors with different screen sizes, display ratios, physical resolutions, and decoding capabilities. Under the current technological architecture, there is a lack of an effective standardized processing mechanism for heterogeneous front-end video sources, and it is also impossible to dynamically adapt to the display characteristics of various types of back-end terminals. This makes it difficult to simultaneously maintain aspect ratio, detail clarity, and decoding compatibility, posing a risk of image adaptation failure. In particular, critical diagnostic details in ultrasound images, such as minute lesions, tissue boundaries, and blood flow signals, are easily lost during improper scaling, failing to meet the stringent requirements of medical diagnosis for image fidelity. Therefore, overcoming the heterogeneity of ultrasound image sources and achieving high-definition, lossless, and adaptive display on various remote terminals has become one of the key problems urgently needing to be solved in the field of remote ultrasound technology. Summary of the Invention

[0005] To overcome the heterogeneity of ultrasound image sources and achieve high-definition, lossless, and adaptive display on various remote terminals, this invention proposes a multi-terminal fidelity display adaptation method for remote ultrasound images, comprising the following steps: S1: Acquire heterogeneous raw ultrasound video streams from multiple ultrasound instruments of different brands and / or models at the clinical end, which differ in one or more parameters such as resolution, frame rate, encoding format, color space and grayscale level. S2: Standardize and resample heterogeneous raw ultrasound image video streams through multi-protocol adaptive parsing to generate a basic video stream in a unified format containing complete diagnostic details; S3: Obtain terminal characteristic parameters of the target display terminal, including at least screen resolution, display ratio, and hardware decoding capability; S4: Calculate the scaling factor based on the original resolution of the base video stream and the screen resolution, and determine the stretch-free target rendering size adapted to the target display terminal based on the scaling factor and the display ratio. S5: Scaling the base video stream according to the unstretched target rendering size, the scaling process includes: identifying different image regions in the base video stream through a semantic segmentation model optimized by ultrasound image features, resampling pixels using different interpolation algorithms respectively, and generating a target-adapted video stream; S6: Calculate the structural similarity index between the core diagnostic area of ​​the target adapted video stream and the corresponding area of ​​the basic video stream, and perform parameter feedback adjustment when the structural similarity index is lower than the preset threshold. Then, transmit the adjusted target adapted video stream to the target display terminal for display through the communication network.

[0006] This invention achieves compatibility and adaptation between various types of ultrasound instruments and various display terminals by standardizing heterogeneous video sources and performing regionally differentiated scaling and fidelity feedback adjustment based on terminal characteristics, ensuring that diagnostic details can be presented without stretching and with high fidelity on different screens.

[0007] Furthermore, in step S2, the heterogeneous original ultrasound image video stream is standardized and analyzed, specifically as follows: The heterogeneous raw ultrasound video stream is decoded and uniformly converted into a preset resolution range, frame rate range, encoding format, and color space.

[0008] Further, in step S4, determining the stretch-free target rendering size adapted to the target display terminal specifically includes: Calculate the original aspect ratio of the base video stream and the screen aspect ratio of the target display terminal; If the original aspect ratio is greater than or equal to the screen aspect ratio, the screen width of the target display terminal is set as the target rendering width, and the target rendering height is calculated based on the original aspect ratio. At the same time, black borders are filled at the top and bottom of the screen to center the image, and the size of the black borders is non-negatively verified. If the original aspect ratio is less than the screen aspect ratio, the screen height of the target display terminal is set as the target rendering height, and the target rendering width is calculated based on the original aspect ratio. At the same time, black borders are filled on the left and right sides of the screen to center the image, and the size of the black borders is non-negatively verified.

[0009] Furthermore, step S4 also includes: When the calculated scaling factor is greater than 1, that is, when the original resolution of the base video stream is lower than the screen resolution of the target display terminal, before scaling the base video stream, a lightweight EDSR super-resolution network pre-trained for features of ultrasound images such as tissue boundaries, lesions, and blood flow signals is used to reconstruct details of the base video stream. The EDSR super-resolution network has 8 residual blocks and 3×3 convolution kernels.

[0010] Further, in step S5, different image regions in the basic video stream are identified using a semantic segmentation model optimized by ultrasound image features. Specifically, the image frames in the basic video stream are divided into a core diagnostic region, an information overlay region, and a blank background region using the semantic segmentation model, wherein: The core diagnostic area includes the lesion tissue, tissue boundaries, and blood flow signals; the information overlay area includes parameter text, measurement values, or scale markings; and the blank background area is the area other than the core diagnostic area and the information overlay area.

[0011] Furthermore, in step S5, different interpolation algorithms are used for pixel resampling, specifically including: The core diagnostic area is resampled using a bicubic interpolation algorithm with an ultrasound-optimized weighting function. The specific formula for the ultrasound-optimized weighting function is as follows: In the formula, This is used to control the smoothness or sharpness of the interpolation result. The distance from a pixel to the target interpolation point. The distance to the target interpolation point is The weight of each pixel; The information overlay area is resampled using a bilinear interpolation algorithm; The blank background area is resampled using the nearest neighbor interpolation algorithm.

[0012] Furthermore, after resampling the core diagnostic area using a bicubic interpolation algorithm, the process also includes: Calculate the structural similarity index between the resampled core diagnostic region and the original core diagnostic region before resampling, and determine whether the structural similarity index is greater than or equal to the preset fidelity threshold. If not, adjust the convolution kernel parameters or weight function parameters of the bicubic interpolation algorithm until the structural similarity index reaches the fidelity threshold.

[0013] Furthermore, in step S3, the terminal characteristic parameters also include real-time network bandwidth and / or screen physical size. When the real-time network bandwidth is lower than a preset threshold, the scaling process in step S5 further includes: Reduce the quantization parameters of the blank background area, or perform frame skipping processing on the blank background area.

[0014] Furthermore, in step S6, before transmitting the adjusted target adapted video stream to the target display terminal for display via the communication network after calculating the structural similarity index between the core diagnostic area of ​​the target adapted video stream and the corresponding area of ​​the basic video stream, the following steps are also included: Based on the hardware decoding capabilities of the target display terminal, the target adaptive video stream is encoded using bitrate adaptive encoding to generate a transport stream that matches the hardware decoding capabilities.

[0015] Compared with the prior art, the present invention has at least the following beneficial effects: (1) The present invention proposes a multi-terminal fidelity display adaptation method for remote ultrasound images. By standardizing and parsing heterogeneous video streams from ultrasound instruments of different brands and models, a base stream of a unified format is generated, thereby eliminating source differences. Then, based on the screen parameters of the target terminal, the unstretched rendering size is determined by proportional calculation, and combined with black border filling, the deformation of anatomical structures caused by improper display ratio is solved. (2) Introducing a region scaling process based on ultrasound image feature optimization, identifying and distinguishing the core diagnostic area, information overlay area and blank background area through a semantic segmentation model, and specifically using bicubic interpolation, bilinear interpolation and nearest neighbor interpolation algorithms to ensure that while reducing the overall computational load and transmission bit rate, the fidelity of high-value diagnostic details such as lesions and blood flow is preserved to the maximum extent. (3) The scaling quality of the core area is quantitatively fed back and dynamically adjusted by the structural similarity index, and the bit rate can be adaptively encoded according to the terminal decoding capability and real-time network status, so that clear, stable, distortion-free image display effect that meets the requirements of medical diagnosis can be achieved on various terminals. Attached Figure Description

[0016] Figure 1 A step diagram illustrating a method for adapting remote ultrasound images to multi-terminal high-fidelity display. Detailed Implementation

[0017] The following are specific embodiments of the present invention, which are described in conjunction with the accompanying drawings. However, the present invention is not limited to these embodiments.

[0018] The core idea of ​​this invention is to introduce an intelligent adaptation layer into the remote ultrasound transmission architecture. This layer standardizes heterogeneous video sources and, combined with the display characteristics of the target terminal and network status, dynamically performs refined and regionally differentiated scaling and enhancement processing on the video stream. Furthermore, a fidelity feedback mechanism ensures the lossless transmission of core diagnostic information. Figure 1 As shown, the multi-terminal high-fidelity display adaptation method for remote ultrasound images proposed in this invention mainly includes the following steps: S1: Acquire heterogeneous raw ultrasound video streams from multiple ultrasound instruments of different brands and / or models at the clinical end, which differ in one or more parameters such as resolution, frame rate, encoding format, color space and grayscale level. S2: Standardize and resample heterogeneous raw ultrasound image video streams through multi-protocol adaptive parsing to generate a basic video stream in a unified format containing complete diagnostic details; S3: Obtain terminal characteristic parameters of the target display terminal, including at least screen resolution, display ratio, and hardware decoding capability; S4: Calculate the scaling factor based on the original resolution of the base video stream and the screen resolution, and determine the stretch-free target rendering size adapted to the target display terminal based on the scaling factor and the display ratio. S5: Scaling the base video stream according to the unstretched target rendering size, the scaling process includes: identifying different image regions in the base video stream through a semantic segmentation model optimized by ultrasound image features, resampling pixels using different interpolation algorithms respectively, and generating a target-adapted video stream; S6: Calculate the structural similarity index (SSIM) between the core diagnostic area of ​​the target adapted video stream and the corresponding area of ​​the basic video stream, and perform parameter feedback adjustment when the structural similarity index is lower than a preset threshold. Then, transmit the adjusted target adapted video stream to the target display terminal for display through the communication network.

[0019] Specifically, the method described in this invention first deploys a remote consultation proxy device (hereinafter referred to as the "proxy device") integrating a core algorithm module at the clinical end. This device connects to one or more ultrasound instruments of different brands and models via wired or wireless means. These ultrasound instruments (e.g., Mindray Resona R9T, GE Logiq E9, or Philips EPIQ 5) may output raw ultrasound video streams with significant differences in one or more parameters such as resolution (e.g., 480P, 720P, 1080P, 4K), frame rate (e.g., 15fps, 25fps, 30fps, 60fps), encoding format (e.g., H.264, H.265, MJPEG), color space (e.g., RGB, YUV, grayscale), and grayscale levels (e.g., 8bit, 10bit), constituting heterogeneous raw ultrasound video streams. The proxy device, through its built-in multiple video acquisition interfaces and protocol stack, can automatically identify and access these heterogeneous video sources.

[0020] After acquiring the heterogeneous raw ultrasound video streams, a multi-protocol adaptive parsing mechanism is used to standardize and resample them. This process is not a simple format conversion, but aims to generate a unified basic video stream containing complete diagnostic details. Specifically, the proxy device first decodes the input video stream, identifying its original encoding format, resolution, frame rate, color space, bit depth, and other metadata. Subsequently, these video streams with varying parameters are uniformly converted into a preset internal format. In a preferred embodiment, this internal format is set as follows: resolution 1920×1080 (or intelligently selected based on the original resolution, such as downsampling from 4K to 1080P to balance detail and bandwidth), frame rate 30fps, encoding format H.265 Main Profile, color space BT.709 YUV 4:2:2, and bit depth 10bit. This standardization process provides a stable, high-quality input foundation for all subsequent downstream adaptation work, ensuring that core diagnostic information is completely preserved and transmitted to subsequent processing stages regardless of the quality of the original video source.

[0021] After standardizing the video source, the most crucial adaptation process begins. The first step is to obtain the characteristic parameters of the target display terminal. When the expert-side device (such as a mobile phone, tablet, or computer) initiates a consultation connection request to the clinical-side proxy device via a telecommunications network (Internet / 4G / 5G), the proxy device will conduct a capability negotiation with the terminal. During this process, the terminal reports its detailed hardware and status information, which includes at least screen resolution, aspect ratio, and hardware decoding capabilities (such as support for H.264 / H.265 hardware decoding, maximum supported resolution, etc.). In a more preferred embodiment, it also includes real-time network bandwidth, screen physical size, supported color spaces, and color depth.

[0022] Based on the acquired terminal feature parameters, the proxy device calculates a target rendering size that ensures the image is not stretched. First, it calculates the original aspect ratio of the normalized base video stream (…). For example, a 16:9 video stream. At the same time, the screen aspect ratio of the target display terminal is calculated. For example, a portrait-oriented mobile phone with a resolution of 1080×2400, its By comparison and The size relationship determines the adaptation strategy.

[0023] If the original aspect ratio is greater than or equal to the screen aspect ratio ( This means the base video stream is "wider" than the screen, and should be scaled proportionally to the screen width. Target rendering width ( Set directly to the screen width ( ), while the target rendering height ( The value is calculated by dividing the screen width by the original aspect ratio, and then rounded down to ensure that the pixel value is an integer. Since the image is based on width, its height often cannot fill the entire screen, therefore black borders are needed at the top and bottom of the screen. Total height of the vertical black borders. (in (where the screen height is the base height), this value is evenly divided into the top black border (...). ) and the black border at the bottom ( This ensures the image is displayed in the center. Simultaneously, the program performs non-negativity checks on all calculation results, for example... This prevents negative values ​​from being caused by abnormal terminal parameters.

[0024] Conversely, if the original aspect ratio is smaller than the screen aspect ratio ( This indicates that the image is "taller" than the screen, and in this case, the screen height should be used as the reference for adaptation. Target rendering height ( Set to screen height ( ), target rendering width Accordingly, the total width of the horizontal black border needs to be evenly distributed across the left and right sides of the screen. Similarly, non-negative verification is performed and the image is displayed in the center. Through this mathematical constraint based on aspect ratio, the possibility of non-proportional stretching of the image to fill the screen is eliminated, thus ensuring the geometric authenticity of anatomical structures in ultrasound images and avoiding the risk of clinical misdiagnosis.

[0025] After determining the target rendering size without stretching, the base video stream can be scaled according to this size. In this process, the present invention also introduces a differentiated interpolation strategy based on region semantics. During scaling, a semantic segmentation model optimized by ultrasound image features is first used to identify different image regions in the base video stream. This model can be a deep neural network pre-trained using a large amount of ultrasound image data, such as U-Net or DeepLab series models, whose training data includes pixel-level annotations to distinguish tissue, lesions, blood flow, text, and background. In actual processing, the model divides each frame of image into three key regions: the core diagnostic region, the information overlay region, and the blank background region. The core diagnostic region refers to the region containing lesions, tissue boundaries, blood flow signals, etc., which have the highest diagnostic value; the information overlay region refers to the auxiliary information region such as parameter text, scale markings, and positional markings superimposed on the image by the ultrasound instrument; the blank background region is a pure black or noisy background region other than the above two regions, which has no diagnostic value.

[0026] After identifying different regions, the agent device uses different interpolation algorithms to resample pixels. This region-based processing strategy maximizes the retention of the most diagnostically valuable information within limited computing resources and network bandwidth, achieving a balance between fidelity and efficiency. Specifically: For the core diagnostic area, due to its extremely high requirement for detail fidelity, a bicubic interpolation algorithm, which has higher computational complexity but optimal fidelity, is employed. This algorithm considers the grayscale values ​​of 16 pixels within a 4×4 neighborhood surrounding the target pixel and uses an ultrasound-optimized weighting function for weighted calculation. The weighting formula is as follows: , In the formula, This is an adjustable parameter used to control the smoothness or sharpness of the interpolation result; in this embodiment, it is preferably -0.8. The distance from a pixel to the target interpolation point. The distance to the target interpolation point is The pixel weights are determined. This pixel interpolation method preserves the texture and edge information of the original image to the greatest extent, ensuring that key features such as lesion boundaries and blood flow textures remain clearly discernible after scaling.

[0027] For information overlay areas, a bilinear interpolation algorithm is used to ensure both clear readability of text and numbers and processing efficiency. This algorithm performs linear weighting on four pixels in a 2×2 neighborhood around the target pixel, which can better maintain the sharpness of character edges and avoid text blurring.

[0028] For the calculated blank background area, the nearest neighbor interpolation algorithm with the least computational cost is used. This algorithm directly maps the target pixel to the nearest pixel in the original image. Although it will produce some jagged edges, since the background area itself does not contain diagnostic information and is usually a solid color, this processing method has almost no impact on diagnosis, but can greatly reduce the computational load of the entire scaling process.

[0029] In addition, after completing the scaling process and generating the target-adapted video stream, this invention introduces a safeguard mechanism: calculating the fidelity of the core diagnostic region and adjusting parameters accordingly. Specifically, the proxy device extracts image blocks corresponding to the core diagnostic region from both the scaled target-adapted video stream and the original standardized base video stream. Then, it calculates the structural similarity index (SSIM) between these two image blocks. SSIM is an index that comprehensively measures image brightness, contrast, and structural similarity; the closer its value is to 1, the more similar the two images are. For the core diagnostic region, this invention sets a higher fidelity threshold, SSIM ≥ 0.98, to ensure lossless transmission of diagnostic information.

[0030] If the calculated SSIM value is lower than this preset threshold, it means that the scaling process may have caused unacceptable loss or distortion of detail in the core diagnostic area. In this case, the system triggers a parameter feedback adjustment loop. This adjustment process may involve adjusting the convolution kernel parameters or weight function parameters in the bicubic interpolation algorithm, such as modifying the shape parameters in the Mitchell-Netravali interpolation kernel to better preserve image sharpness. After adjusting the parameters, the system recalculates the interpolation of the core diagnostic area and recalculates the SSIM value, iterating until the calculated SSIM value meets the preset fidelity threshold. Only when the SSIM meets the threshold will the system transmit the finally adjusted, fidelity-verified target adaptation video stream to the target display terminal for display via the communication network. This feedback mechanism ensures that the final display effect meets stringent medical diagnostic standards in any complex adaptation scenario.

[0031] Based on the above, the present invention also provides several preferred embodiments to further improve the adaptability and robustness of the solution.

[0032] In a preferred embodiment, step S4 is followed by an upsampling optimization step for high-performance terminals. When the calculated scaling factor is greater than 1, meaning the original resolution of the base video stream is lower than the screen resolution of the target display terminal (e.g., a 1080P normalized video stream to be displayed on a 4K computer screen), simply scaling up may result in image blurring. Therefore, before scaling the base video stream, a lightweight EDSR super-resolution network pre-trained for ultrasound image features is invoked to reconstruct details from the base video stream. In this embodiment, the EDSR super-resolution network preferably has 8 residual blocks and 3×3 convolutional kernels. This network optimizes for high-frequency details such as tissue boundaries and blood flow signals in ultrasound images through residual learning, reconstructing low-resolution images into high-resolution images while enhancing and restoring these key details, thereby achieving a high-definition display effect that surpasses the original resolution on high-performance terminals.

[0033] In another preferred embodiment, the differential scaling in step S5 can also be combined with network status. If the terminal characteristic parameters obtained in step S3 include real-time network bandwidth information, and it is determined that the current network bandwidth is lower than a preset threshold (e.g., lower than 5Mbps), then further bitrate control measures can be taken during scaling. Specifically, the quantization parameter (QP) of the blank background area during encoding can be reduced to make it more compressed, or the blank background area can be directly subjected to frame skipping, i.e., its temporal resolution can be reduced. This reduces the bitrate of the overall video stream without affecting the image quality of the core diagnostic area, making it suitable for weak network environments and avoiding stuttering.

[0034] Finally, after calculating and verifying the SSIM of the core diagnostic area, before transmitting the target adapted video stream to the terminal display, the proxy device will perform final encoding and encapsulation of the verified target adapted video stream based on the hardware decoding capabilities of the target display terminal. For example, for older mobile phones that only support H.264 hardware decoding, the system will encode the video stream into H.264 format; while for high-performance computers that support the latest H.266 standard, the H.266 format with a higher compression ratio can be selected, and an appropriate encoding level and bitrate can be selected within its decoding capability range, thereby generating a transmission stream that perfectly matches the terminal's hardware decoding capabilities, ensuring that the terminal can decode and display smoothly and stably.

[0035] In addition, the present invention also includes a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of a method for adapting a multi-terminal high-fidelity display of remote ultrasound images.

[0036] It also includes a data processing device, comprising: A memory on which computer programs are stored; A processor is configured to execute a computer program in the memory to implement the steps of a method for adapting a remote ultrasound image to a multi-terminal high-fidelity display.

[0037] In summary, the present invention proposes a multi-terminal high-fidelity display adaptation method for remote ultrasound images. By standardizing and parsing heterogeneous video streams from ultrasound instruments of different brands and models, a base stream of uniform format is generated, thereby eliminating source-end differences. Then, based on the screen parameters of the target terminal, the unstretched rendering size is determined by proportional calculation, and combined with black border filling, the distortion of anatomical structures caused by improper display ratio is solved.

[0038] We introduce a region scaling process based on ultrasound image feature optimization. We identify and distinguish the core diagnostic area, information overlay area and blank background area through a semantic segmentation model. We also use bicubic interpolation, bilinear interpolation and nearest neighbor interpolation algorithms to ensure that the fidelity of high-value diagnostic details such as lesions and blood flow is preserved to the maximum extent while reducing the overall computational load and transmission bit rate.

[0039] The scaling quality of the core area is quantitatively fed back and dynamically adjusted by the structural similarity index. The bitrate can be adaptively encoded according to the terminal's decoding capability and real-time network status, so as to achieve clear, stable, distortion-free image display effect that meets the professional requirements of medical diagnosis on various terminals.

[0040] It should be noted that all directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present invention are only used to explain the relative positional relationship and movement of each component in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indication will also change accordingly.

[0041] Furthermore, in this invention, descriptions involving terms such as "first," "second," and "a" are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0042] In this invention, unless otherwise explicitly specified and limited, the terms "connection," "fixed," etc., should be interpreted broadly. For example, "fixed" can mean a fixed connection, a detachable connection, or an integral part; it can mean a mechanical connection or an electrical connection; it can mean a direct connection or an indirect connection through an intermediate medium; it can mean the internal communication of two components or the interaction between two components, unless otherwise explicitly limited. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0043] Furthermore, the technical solutions of the various embodiments of the present invention can be combined with each other, but only if they are feasible for those skilled in the art. If the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such combination of technical solutions does not exist and is not within the scope of protection claimed by the present invention.

Claims

1. A method for multi-terminal high-fidelity display adaptation of remote ultrasound images, characterized in that, Including the following steps: S1: Acquire heterogeneous raw ultrasound video streams from multiple ultrasound instruments of different brands and / or models at the clinical end, which differ in one or more parameters such as resolution, frame rate, encoding format, color space and grayscale level. S2: Standardize and resample heterogeneous raw ultrasound image video streams through multi-protocol adaptive parsing to generate a basic video stream in a unified format containing complete diagnostic details; S3: Obtain terminal characteristic parameters of the target display terminal, including at least screen resolution, display ratio, and hardware decoding capability; S4: Calculate the scaling factor based on the original resolution of the base video stream and the screen resolution, and determine the stretch-free target rendering size adapted to the target display terminal based on the scaling factor and the display ratio. S5: Scaling the base video stream according to the unstretched target rendering size, the scaling process includes: identifying different image regions in the base video stream through a semantic segmentation model optimized by ultrasound image features, resampling pixels using different interpolation algorithms respectively, and generating a target-adapted video stream; S6: Calculate the structural similarity index between the core diagnostic area of ​​the target adapted video stream and the corresponding area of ​​the basic video stream, and perform parameter feedback adjustment when the structural similarity index is lower than the preset threshold. Then, transmit the adjusted target adapted video stream to the target display terminal for display through the communication network.

2. The method for multi-terminal high-fidelity display adaptation of remote ultrasound images as described in claim 1, characterized in that, In step S2, the heterogeneous original ultrasound image video stream is standardized and parsed, specifically as follows: The heterogeneous raw ultrasound video stream is decoded and uniformly converted into a preset resolution range, frame rate range, encoding format, and color space.

3. The method for multi-terminal high-fidelity display adaptation of remote ultrasound images as described in claim 1, characterized in that, In step S4, determining the stretch-free target rendering size adapted to the target display terminal specifically includes: Calculate the original aspect ratio of the base video stream and the screen aspect ratio of the target display terminal; If the original aspect ratio is greater than or equal to the screen aspect ratio, the screen width of the target display terminal is set as the target rendering width, and the target rendering height is calculated based on the original aspect ratio. At the same time, black borders are filled at the top and bottom of the screen to center the image, and the size of the black borders is non-negatively verified. If the original aspect ratio is less than the screen aspect ratio, the screen height of the target display terminal is set as the target rendering height, and the target rendering width is calculated based on the original aspect ratio. At the same time, black borders are filled on the left and right sides of the screen to center the image, and the size of the black borders is non-negatively verified.

4. The method for multi-terminal high-fidelity display adaptation of remote ultrasound images as described in claim 1, characterized in that, The S4 step also includes: When the calculated scaling factor is greater than 1, that is, when the original resolution of the base video stream is lower than the screen resolution of the target display terminal, before scaling the base video stream, a lightweight EDSR super-resolution network pre-trained for ultrasound image features is used to reconstruct the details of the base video stream. The EDSR super-resolution network has 8 residual blocks and 3×3 convolution kernels.

5. The method for multi-terminal high-fidelity display adaptation of remote ultrasound images as described in claim 1, characterized in that, In step S5, different image regions in the basic video stream are identified using a semantic segmentation model optimized by ultrasound image features. Specifically, the image frames in the basic video stream are divided into a core diagnostic region, an information overlay region, and a blank background region using the semantic segmentation model, wherein: The core diagnostic area includes the lesion tissue, tissue boundaries, and blood flow signals; the information overlay area includes parameter text, measurement values, or scale markings; and the blank background area is the area other than the core diagnostic area and the information overlay area.

6. The method for multi-terminal high-fidelity display adaptation of remote ultrasound images as described in claim 5, characterized in that, In step S5, different interpolation algorithms are used for pixel resampling, specifically including: The core diagnostic area is resampled using a bicubic interpolation algorithm with an ultrasound-optimized weighting function. The specific formula for the ultrasound-optimized weighting function is as follows: In the formula, This is used to control the smoothness or sharpness of the interpolation result. The distance from a pixel to the target interpolation point. The distance to the target interpolation point is The weight of each pixel; The information overlay area is resampled using a bilinear interpolation algorithm; The blank background area is resampled using the nearest neighbor interpolation algorithm.

7. The method for multi-terminal high-fidelity display adaptation of remote ultrasound images as described in claim 6, characterized in that, After resampling the core diagnostic area using a bicubic interpolation algorithm, the following steps are also included: Calculate the structural similarity index between the resampled core diagnostic region and the original core diagnostic region before resampling, and determine whether the structural similarity index is greater than or equal to the preset fidelity threshold. If not, adjust the convolution kernel parameters or weight function parameters of the bicubic interpolation algorithm until the structural similarity index reaches the fidelity threshold.

8. The method for multi-terminal high-fidelity display adaptation of remote ultrasound images as described in claim 5, characterized in that, In step S3, the terminal characteristic parameters also include real-time network bandwidth and / or screen physical size. When the real-time network bandwidth is lower than a preset threshold, the scaling process in step S5 further includes: Reduce the quantization parameters of the blank background area, or perform frame skipping processing on the blank background area.

9. The method for multi-terminal high-fidelity display adaptation of remote ultrasound images as described in claim 1, characterized in that, In step S6, before transmitting the adjusted target adapted video stream to the target display terminal for display via the communication network after calculating the structural similarity index between the core diagnostic area of ​​the target adapted video stream and the corresponding area of ​​the basic video stream, the following steps are also included: Based on the hardware decoding capabilities of the target display terminal, the target adaptive video stream is encoded using bitrate adaptive encoding to generate a transport stream that matches the hardware decoding capabilities.

Images