Thyroid scanning system and method based on AI assistance

Abstract

The invention provides a thyroid scanning system and method based on AI assistance, and relates to the technical field of ultrasonic examination, and the method comprises the steps: obtaining a video stream, dividing the video stream into a plurality of video clips according to a time sequence, and associating time intervals; extracting an optimal thyroid ultrasound image from the video clips, uploading the optimal thyroid ultrasound image to a cloud analysis system to obtain a corresponding cloud analysis result, and distributing each video clip to a plurality of AI analysis lightweight models to obtain a corresponding local analysis result; and establishing a cache queue, adding the video clips which have obtained the local analysis result and the cloud analysis result into the cache queue, adjusting the sequence, and then outputting the video clips and the analysis results in the cache queue in sequence. The system has the beneficial effects that the cost is reduced, and the universality of the system is improved; the real-time performance is improved, and the data transmission pressure is reduced; the diagnosis result delay caused by poor network conditions is reduced, and the reliability of the diagnosis result is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of ultrasonic examination, and in particular to a thyroid scanning system and method based on AI assistance. Background Art

[0002] Existing ultrasound equipment is mainly divided into two categories: high-resolution equipment used by professional medical institutions and portable equipment. Professional equipment is known for its high-resolution image acquisition capabilities. It can capture high-quality ultrasound images in real time and quickly transmit image data to smart devices or servers for AI analysis through high-bandwidth and low-latency transmission technology. The AI ​​analysis system uses deep learning models to process large amounts of data in a short period of time and provide highly accurate diagnostic results. However, these devices are expensive and unaffordable for ordinary users. In contrast, portable ultrasound devices are more affordable and easy to carry, but usually only provide image acquisition functions. Users need to have certain professional knowledge to correctly use and interpret images, which limits their popularity in non-professional fields.

[0003] In addition, both professional and portable devices face the problem of data transmission and real-time performance. The transmission of high-resolution ultrasound image data requires high bandwidth and low latency technical support, but the existing network conditions may not meet this requirement, resulting in data transmission delays and affecting the real-time performance of diagnosis. At the same time, the data transmission and analysis process is highly dependent on Internet connection. Once the network conditions are poor, data upload and analysis will be affected, which may lead to delays in diagnostic results.

[0004] In addition, for non-professionals, the operation of existing ultrasound equipment is relatively complicated and it is difficult to accurately judge the image results, which also limits the widespread application of portable ultrasound equipment. Therefore, the development of ultrasound equipment needs to be improved and innovated to address these problems in order to enhance the practicality and popularity of the equipment. Summary of the invention

[0005] In view of the problems existing in the prior art, the present invention provides an AI-assisted thyroid scanning system, comprising:

[0006] An image acquisition module, connected to the probe of the portable ultrasonic scanning device, is used to acquire a video stream containing multiple frames of thyroid ultrasound images during the scanning process of the patient, and divide the video stream into multiple video segments in chronological order and associate time intervals;

[0007] an image distribution module, connected to the image acquisition module, for extracting an optimal thyroid ultrasound image from each of the video clips, uploading the optimal thyroid ultrasound image to a cloud analysis system to obtain a corresponding cloud analysis result, and respectively distributing each of the video clips to a plurality of AI analysis lightweight models configured in a host of a portable ultrasound scanning device to obtain a corresponding local analysis result;

[0008] A result output module is connected to the image distribution module and is used to establish a cache queue, add the video clips that have obtained the local analysis results and the cloud analysis results to the cache queue, and adjust the order in the cache queue according to the associated time interval, and then output each video clip in the cache queue and its associated local analysis results and cloud analysis results in sequence.

[0009] Preferably, the image acquisition module includes:

[0010] The video segmentation unit is used to divide the video stream into a plurality of video segments in sequence according to a preset initial interval length, and then for each video segment, the corresponding system time of the thyroid ultrasound image at the beginning and the end of the video segment is associated with the video segment as the time interval of the video segment.

[0011] Preferably, the image acquisition module further includes an interval adjustment unit, connected to the video segmentation unit, for obtaining the image processing capability of the AI ​​analysis lightweight model, and adjusting the initial interval length according to the image processing capability;

[0012] The video segmentation unit divides the video stream in sequence according to the adjusted initial interval length to obtain a plurality of the video segments.

[0013] Preferably, the interval adjustment unit includes:

[0014] A storage subunit, used for storing a standard video segment whose interval length is equal to the initial interval length, wherein the standard video segment is associated with a corresponding standard processing duration;

[0015] a capability testing subunit, connected to the storage subunit, and configured to input the standard video clip into the AI ​​analysis lightweight model multiple times to obtain multiple test processing times, and use the average value of all the test processing times as the image processing capability of the AI ​​analysis lightweight model;

[0016] The length adjustment subunit is connected to the capability testing unit and is used to calculate the ratio of the image processing capability to the standard processing time as the adjustment ratio, and multiply the initial interval length by the adjustment ratio to obtain the adjusted initial interval length.

[0017] Preferably, the image distribution module comprises:

[0018] A quality evaluation unit is used to, for each of the video clips, use each frame of the thyroid ultrasound image contained therein as a candidate image, and then perform a comprehensive scoring quality evaluation on each frame of the candidate image according to clarity, contrast, noise, integrity and artifacts to obtain a corresponding image quality score;

[0019] An image extraction unit is connected to the quality evaluation unit and is used to select the candidate image with the highest image quality score as the optimal thyroid ultrasound image.

[0020] The present invention also provides an AI-assisted thyroid scanning method, which is applied to the above-mentioned thyroid scanning system, comprising:

[0021] Step S1, the thyroid scanning system obtains a video stream containing multiple frames of thyroid ultrasound images during the scanning process of the patient, divides the video stream into multiple video segments in chronological order and associates them with time intervals;

[0022] Step S2, the thyroid scanning system obtains, for each of the video clips, extracts the optimal thyroid ultrasound image from the video clip, uploads the optimal thyroid ultrasound image to the cloud analysis system to obtain the corresponding cloud analysis result, and assigns each of the video clips to a plurality of AI analysis lightweight models configured in the host of the portable ultrasound scanning device to obtain the corresponding local analysis result;

[0023] Step S3, the thyroid system is scanned to obtain and establish a cache queue, the video clips that have obtained the local analysis results and the cloud analysis results are added to the cache queue, and the order in the cache queue is adjusted according to the associated time interval, and then each video clip in the cache queue and its associated local analysis results and the cloud analysis results are output in sequence.

[0024] Preferably, the step S1 includes:

[0025] Step S11, the thyroid scanning system obtains a video stream containing multiple frames of thyroid ultrasound images during the scanning process of the patient by the probe of the portable ultrasound scanning device;

[0026] In step S12, the thyroid system is scanned by dividing the video stream in sequence according to a preset initial interval length to obtain a plurality of video segments, and then for each video segment, the corresponding system time of the thyroid ultrasound image at the beginning and the end of the video segment is associated with the video segment as the time interval of the video segment.

[0027] Preferably, before executing step S11, step S1 further includes:

[0028] Step S10, the thyroid scanning system obtains the image processing capability of the AI ​​analysis lightweight model, and adjusts the initial interval length according to the image processing capability;

[0029] In the step S12, the video stream is divided in sequence according to the adjusted initial interval length to obtain a plurality of the video segments.

[0030] Preferably, the thyroid scanning system stores a standard video clip with an interval length equal to the initial interval length, and the standard video clip is associated with a corresponding standard processing duration;

[0031] The step S10 comprises:

[0032] Step S101, the thyroid scanning system inputs the standard video clip into the AI ​​analysis lightweight model multiple times to obtain multiple test processing times, and uses the average value of all the test processing times as the image processing capability of the AI ​​analysis lightweight model;

[0033] Step S102: the thyroid scanning system calculates a ratio of the image processing capability to the standard processing time as an adjustment ratio, and multiplies the initial interval length by the adjustment ratio to obtain the adjusted initial interval length.

[0034] Preferably, the step S2 includes:

[0035] Step S21, the thyroid scanning system uses each frame of the thyroid ultrasound image contained in each video clip as a candidate image, and then performs a comprehensive scoring quality assessment on each frame of the candidate image according to clarity, contrast, noise, integrity and artifacts to obtain a corresponding image quality score;

[0036] Step S22, the thyroid scanning system uses the candidate image with the highest image quality score as the optimal thyroid ultrasound image, and uploads the optimal thyroid ultrasound image to the cloud analysis system to obtain a corresponding cloud analysis result;

[0037] In step S23, the thyroid scanning system distributes each of the video clips to a plurality of AI analysis lightweight models configured in a host of the portable ultrasound scanning device to obtain corresponding local analysis results.

[0038] The above technical solution has the following advantages or beneficial effects:

[0039] 1. The use of portable ultrasound scanning equipment as the front end reduces the cost burden of professional high-resolution equipment, making thyroid ultrasound testing affordable for ordinary users. At the same time, the combination of cloud and local AI analysis reduces the dependence on a single high-performance device, further indirectly reducing costs and increasing the popularity of the system.

[0040] 2. Uploading the optimal thyroid ultrasound image to the cloud analysis system through the image distribution module greatly reduces the network bandwidth requirement and the dependence on network conditions compared to transmitting the entire video stream. At the same time, multiple AI analysis lightweight models are used locally for parallel processing, and the speed is guaranteed through wired data transmission, which improves real-time performance and reduces data transmission pressure.

[0041] 3. The result output module establishes a cache queue and adjusts the order of video clips according to the time interval, ensuring the order of the output results and avoiding output confusion caused by network delays. This design reduces the delay of diagnosis results caused by poor network conditions and improves the reliability of diagnosis results and user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 This is a schematic diagram of the results of an AI-assisted scan of the thyroid system in a preferred embodiment of the present invention;

[0043] Figure 2 The figure is a flow chart of a method for scanning the thyroid gland based on AI assistance in a preferred embodiment of the present invention;

[0044] Figure 3 A schematic diagram of a sub-process of step S1 in a preferred embodiment of the present invention;

[0045] Figure 4 This is a schematic diagram of a sub-process of step S10 in a preferred embodiment of the present invention; Figure 5 This is a schematic diagram of a sub-process of step S2 in a preferred embodiment of the present invention. DETAILED DESCRIPTION

[0046] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. The present invention is not limited to this embodiment, and other embodiments may also fall within the scope of the present invention as long as they conform to the gist of the present invention.

[0047] In a preferred embodiment of the present invention, based on the above problems existing in the prior art, a thyroid scanning system based on AI assistance is provided. Figure 1 As shown, including:

[0048] An image acquisition module 1 is connected to a probe of a portable ultrasonic scanning device, and is used to acquire a video stream containing multiple frames of thyroid ultrasound images during a patient's scanning process, and divide the video stream into multiple video segments in chronological order and associate time intervals;

[0049] The image distribution module 2 is connected to the image acquisition module 2 and is used for extracting the optimal thyroid ultrasound image from each video segment, uploading the optimal thyroid ultrasound image to the cloud analysis system to obtain the corresponding cloud analysis result, and respectively distributing each video segment to a plurality of AI analysis lightweight models configured in the host of the portable ultrasound scanning device to obtain the corresponding local analysis result;

[0050] The result output module 3 is connected to the image distribution module 3 and is used to establish a cache queue, add the video clips that have obtained local analysis results and cloud analysis results to the cache queue, and adjust the order in the cache queue according to the associated time interval, and then output each video clip in the cache queue and its associated local analysis results and cloud analysis results in sequence.

[0051] Specifically, the high-resolution equipment used by professional medical institutions is expensive and unaffordable for ordinary users. The system in this embodiment uses a portable ultrasound scanning device as the front end for image acquisition. This type of equipment is usually more affordable and easy to carry. At the same time, the system improves diagnostic efficiency and reduces dependence on a single high-performance device by combining cloud and local AI analysis, thereby indirectly reducing costs.

[0052] Specifically, existing ultrasound equipment faces problems with data transmission and real-time performance, especially the transmission of high-resolution image data requires technical support with high bandwidth and low latency. The cloud analysis system has powerful computing power and rich data resources, and can perform in-depth analysis on the uploaded optimal thyroid ultrasound images to provide highly accurate diagnostic results. In this embodiment, the image allocation module uploads the optimal thyroid ultrasound image in each video clip to the cloud analysis system. Since multiple single-frame images are uploaded to the cloud analysis system, compared with the existing ultrasound equipment transmission video stream method, only a small amount of network bandwidth is required, which reduces the dependence on network conditions and stringent requirements, and is less affected by the network.

[0053] At the same time, multiple AI analysis lightweight models configured in the portable device host are used for local analysis. The data transmission rate is guaranteed by local wired data transmission. This parallel processing method reduces the pressure of single data transmission and improves real-time performance.

[0054] The result output module establishes a cache queue and adjusts the order of video clips according to the time interval to ensure the sequentiality of the output results. At the same time, it avoids output confusion caused by network delays and reduces delays in diagnostic results caused by poor network conditions.

[0055] In a preferred embodiment of the present invention, the image acquisition module 1 comprises:

[0056] The video segmentation unit 11 is used to divide the video stream into multiple video segments in sequence according to a preset initial interval length, and then for each video segment, the corresponding system time of the thyroid ultrasound image at the beginning and end of the video segment is associated with the video segment as the time interval of the video segment.

[0057] In a preferred embodiment of the present invention, the image acquisition module further includes an interval adjustment unit 12, connected to the video segmentation unit 11, for obtaining the image processing capability of the AI ​​analysis lightweight model, and adjusting the initial interval length according to the image processing capability;

[0058] The video segmentation unit 11 divides the video stream into a plurality of video segments in sequence according to the adjusted initial interval length.

[0059] Specifically, in this embodiment, the image processing capability of the AI ​​analysis lightweight model is obtained through the interval adjustment unit, and the initial interval length used in the video segmentation unit can be dynamically adjusted. This means that the system can optimize the division of video segments according to the performance of the actual analysis model, ensuring that the length of each video segment is neither too long (causing processing delays) nor too short (affecting image continuity).

[0060] The optimized video clip length ensures that the lightweight AI analysis model can fully utilize its performance when processing each clip, without increasing the computational burden due to processing too many image frames, or missing key information due to too few image frames. This helps improve the efficiency and accuracy of the analysis.

[0061] Different AI analysis lightweight models may have different image processing capabilities and processing speeds. Through the adjustment of the interval adjustment unit, the system can adapt to different analysis models to ensure the best analysis results in various scenarios.

[0062] After the introduction of the interval adjustment unit 12, the system can more flexibly respond to performance improvements or changes that may occur in the future. For example, when the AI ​​analysis lightweight model is upgraded or a new more efficient model is introduced, the system can make full use of these performance improvements by adjusting the initial interval length.

[0063] By dynamically adjusting the length of video clips, the system can more efficiently utilize computing resources and bandwidth. For example, when bandwidth is limited, the system can reduce the amount of data transmission by reducing the number of image frames in each video clip, thereby ensuring real-time and stable data transmission.

[0064] In summary, by introducing the video segmentation unit 11 and the interval adjustment unit 12 in the image acquisition module 1, the system of this embodiment can dynamically adjust the length of the video clip to adapt to the image processing capabilities of different AI analysis lightweight models. This not only improves the efficiency and accuracy of the analysis, but also enhances the flexibility and scalability of the system, providing better support for future performance improvements and changes. At the same time, this design also helps to optimize resource utilization and ensure that the system can run stably in various scenarios.

[0065] In a preferred embodiment of the present invention, the interval adjustment unit 12 includes:

[0066] The storage subunit 121 is used to store standard video segments whose interval length is equal to the initial interval length, and the standard video segments are associated with corresponding standard processing durations;

[0067] The capability testing subunit 122 is connected to the storage subunit 121 and is used to input the standard video clip into the AI ​​analysis lightweight model multiple times to obtain multiple test processing times, and use the average value of all the test processing times as the image processing capability of the AI ​​analysis lightweight model;

[0068] The length adjustment subunit 123 is connected to the capability testing unit 122 and is used to calculate the ratio of the image processing capability to the standard processing time as the adjustment ratio, and multiply the initial interval length by the adjustment ratio to obtain the adjusted initial interval length.

[0069] Specifically, in this embodiment, the standard video clips are input into the AI ​​analysis lightweight model multiple times through the capability test subunit, and the average of multiple test processing times is calculated, so that the image processing capability of the AI ​​analysis lightweight model can be evaluated more accurately. This evaluation method reduces the contingency of a single test and improves the reliability of the evaluation results.

[0070] The length adjustment subunit dynamically adjusts the initial interval length according to the image processing capabilities of the AI ​​analysis lightweight model. This ensures that the length of each video segment can match the processing capabilities of the AI ​​analysis lightweight model, and neither processing delays caused by too long segments nor the integrity of image information affected by too short segments.

[0071] By dynamically adjusting the length of video clips, the system can use computing resources and bandwidth more efficiently. When the processing power of the AI ​​analysis lightweight model is strong, the system can increase the length of the video clip, thereby analyzing more image frames in one processing and improving processing efficiency. Conversely, when the processing power is weak, the system can reduce the length of the clip to reduce the processing burden.

[0072] This design enables the system to adapt to lightweight AI analysis models with different performance. When the model is upgraded or the performance is improved, the system can automatically adjust the length of the video clip to ensure that the analysis efficiency and accuracy are always maintained at the optimal state.

[0073] The standard video clips and the corresponding standard processing duration stored in the storage subunit 121 can be used as part of the system configuration to facilitate implementation and maintenance. When the system performance needs to be adjusted, only the standard video clips need to be updated or the standard processing duration needs to be adjusted.

[0074] In summary, by introducing the storage subunit 121, the capability test subunit 122, and the length adjustment subunit 123 in the interval adjustment unit, the system of this embodiment can accurately evaluate the image processing capability of the AI ​​analysis lightweight model and dynamically adjust the length of the video clip according to the evaluation results. This not only improves the efficiency and accuracy of the analysis, but also optimizes resource utilization and enhances the flexibility and scalability of the system. At the same time, this design also facilitates the implementation and maintenance of the system, providing better support for future performance improvements and changes.

[0075] In a preferred embodiment of the present invention, the image distribution module 2 comprises:

[0076] The quality evaluation unit 21 is used for taking each frame of thyroid ultrasound image contained in each video clip as a candidate image, and then comprehensively scoring and evaluating the quality of each frame of the candidate image according to clarity, contrast, noise, integrity and artifacts to obtain a corresponding image quality score;

[0077] The image extraction unit 22 is connected to the quality evaluation unit 21 and is used to select the candidate image with the highest image quality score as the optimal thyroid ultrasound image.

[0078] Specifically, in this embodiment, the quality evaluation unit performs a comprehensive scoring quality evaluation on each frame of the thyroid ultrasound image in each video clip, which ensures that the selected image is optimal in terms of clarity, contrast, noise control, integrity, and artifacts. Such images are easier for the AI ​​analysis lightweight model to perform accurate analysis and diagnosis.

[0079] The image with the highest quality score is directly selected as the optimal thyroid ultrasound image by the image extraction unit 22, thereby avoiding the processing of a large number of redundant or low-quality images, thereby improving the analysis efficiency. This helps to reduce the waste of computing resources and speed up the diagnosis.

[0080] High-quality images are the basis for accurate diagnosis. The introduction of the quality assessment unit and the image extraction unit ensures that the AI ​​analysis lightweight model receives the best quality images, thereby improving the accuracy of diagnosis. This is crucial for the early detection and effective treatment of thyroid diseases.

[0081] The comprehensive scoring criteria in the quality evaluation unit can be adjusted according to actual needs to adapt to different application scenarios. For example, in some cases, clarity may be more important than other factors; in other cases, more attention may be paid to the integrity of the image. This flexibility enables the system to be optimized according to different needs. Specifically, the user can select the factors to be emphasized before using the device, and the system selects the corresponding ultrasound image quality evaluation method based on the factors, and a variety of ultrasound image quality evaluation methods already available in the prior art can be used.

[0082] In summary, by introducing the quality evaluation unit and the image extraction unit into the image distribution module, the system of this embodiment can automatically select the optimal thyroid ultrasound image for analysis and diagnosis. This not only improves the image quality and analysis efficiency, but also enhances the accuracy of diagnosis, providing strong support for the early detection and effective treatment of thyroid diseases. At the same time, this design also improves the automation of the system, reduces the cost and risk of manual intervention, and provides more possibilities for future system upgrades and optimizations.

[0083] The present invention also provides an AI-assisted thyroid scanning method, which is applied to the above-mentioned thyroid scanning system, such as Figure 2 As shown, including:

[0084] Step S1, scanning the thyroid system to obtain a video stream containing multiple frames of thyroid ultrasound images during the scanning process of the patient, dividing the video stream into multiple video clips in chronological order and associating time intervals;

[0085] Step S2, scanning the thyroid system to obtain, for each video clip, an optimal thyroid ultrasound image is extracted from the video clip, the optimal thyroid ultrasound image is uploaded to the cloud analysis system to obtain a corresponding cloud analysis result, and each video clip is respectively assigned to a plurality of AI analysis lightweight models configured in a host of the portable ultrasound scanning device to obtain a corresponding local analysis result;

[0086] Step S3, scan the thyroid system to obtain and establish a cache queue, add the video clips that have obtained local analysis results and cloud analysis results to the cache queue, and adjust the order in the cache queue according to the associated time interval, and then output each video clip in the cache queue and its associated local analysis results and cloud analysis results in sequence.

[0087] In a preferred embodiment of the present invention, Figure 3 As shown, step S1 includes:

[0088] Step S11, scanning the thyroid system to obtain a video stream containing multiple frames of thyroid ultrasound images during the process of the probe of the portable ultrasound scanning device scanning the patient;

[0089] Step S12, scanning the thyroid system, dividing the video stream in sequence according to a preset initial interval length to obtain multiple video segments, and then for each video segment, the corresponding system time of the thyroid ultrasound images at the beginning and end of the video segment is associated with the video segment as the time interval of the video segment.

[0090] In a preferred embodiment of the present invention, Figure 3 As shown, before executing step S11, step S1 also includes:

[0091] Step S10, scanning the thyroid system to obtain the image processing capability of the AI ​​analysis lightweight model, and adjusting the initial interval length according to the image processing capability;

[0092] In step S12, the video stream is divided into a plurality of video segments in sequence according to the adjusted initial interval length.

[0093] In a preferred embodiment of the present invention, a standard video clip having an interval duration equal to the initial interval length is stored in the thyroid scanning system, and the standard video clip is associated with a corresponding standard processing duration;

[0094] like Figure 4 As shown, step S10 includes:

[0095] Step S101, scanning the thyroid system, inputting the standard video clips into the AI ​​analysis lightweight model multiple times to obtain multiple test processing times, and taking the average of all the test processing times as the image processing capability of the AI ​​analysis lightweight model;

[0096] Step S102, scanning the thyroid system, calculating the ratio of the image processing capability to the standard processing time as an adjustment ratio, and multiplying the initial interval length by the adjustment ratio to obtain an adjusted initial interval length.

[0097] In a preferred embodiment of the present invention, Figure 5 As shown, step S2 includes:

[0098] Step S21, scanning the thyroid system. For each video clip, each frame of thyroid ultrasound image contained therein is used as a candidate image, and then a comprehensive scoring quality evaluation is performed on each frame of the candidate image according to clarity, contrast, noise, integrity and artifacts to obtain a corresponding image quality score;

[0099] Step S22, the thyroid scanning system uses the candidate image with the highest image quality score as the optimal thyroid ultrasound image, and uploads the optimal thyroid ultrasound image to the cloud analysis system to obtain the corresponding cloud analysis result;

[0100] Step S23, the thyroid scanning system distributes each video clip to a plurality of AI analysis lightweight models configured in the host of the portable ultrasound scanning device to obtain corresponding local analysis results.

[0101] The above are only preferred embodiments of the present invention, and are not intended to limit the implementation methods and protection scope of the present invention. Those skilled in the art should be aware that all solutions obtained by equivalent substitutions and obvious changes made using the contents of this specification and illustrations should be included in the protection scope of the present invention.

Claims

1. An AI-assisted thyroid scanning system, characterized in that: include: An image acquisition module, connected to the probe of the portable ultrasonic scanning device, is used to acquire a video stream containing multiple frames of thyroid ultrasound images during the scanning process of the patient, and divide the video stream into multiple video segments in chronological order and associate time intervals; an image distribution module, connected to the image acquisition module, for extracting an optimal thyroid ultrasound image from each of the video clips, uploading the optimal thyroid ultrasound image to a cloud analysis system to obtain a corresponding cloud analysis result, and respectively distributing each of the video clips to a plurality of AI analysis lightweight models configured in a host of a portable ultrasound scanning device to obtain a corresponding local analysis result; A result output module is connected to the image distribution module and is used to establish a cache queue, add the video clips that have obtained the local analysis results and the cloud analysis results to the cache queue, and adjust the order in the cache queue according to the associated time interval, and then output each video clip in the cache queue and its associated local analysis results and cloud analysis results in sequence.

2. The thyroid scanning system according to claim 1, characterized in that: The image acquisition module comprises: The video segmentation unit is used to divide the video stream into a plurality of video segments in sequence according to a preset initial interval length, and then for each video segment, the corresponding system time of the thyroid ultrasound image at the beginning and the end of the video segment is associated with the video segment as the time interval of the video segment.

3. The thyroid scanning system according to claim 2, characterized in that: The image acquisition module further includes an interval adjustment unit connected to the video segmentation unit, for obtaining the image processing capability of the AI ​​analysis lightweight model, and adjusting the initial interval length according to the image processing capability; The video segmentation unit divides the video stream in sequence according to the adjusted initial interval length to obtain a plurality of the video segments.

4. The thyroid scanning system according to claim 3, characterized in that: The interval adjustment unit includes: A storage subunit, used for storing a standard video segment whose interval length is equal to the initial interval length, wherein the standard video segment is associated with a corresponding standard processing duration; a capability testing subunit, connected to the storage subunit, and configured to input the standard video clip into the AI ​​analysis lightweight model multiple times to obtain multiple test processing times, and use the average value of all the test processing times as the image processing capability of the AI ​​analysis lightweight model; The length adjustment subunit is connected to the capability testing unit and is used to calculate the ratio of the image processing capability to the standard processing time as the adjustment ratio, and multiply the initial interval length by the adjustment ratio to obtain the adjusted initial interval length.

5. The thyroid scanning system according to claim 1, characterized in that: The image distribution module comprises: A quality evaluation unit is used to, for each of the video clips, use each frame of the thyroid ultrasound image contained therein as a candidate image, and then perform a comprehensive scoring quality evaluation on each frame of the candidate image according to clarity, contrast, noise, integrity and artifacts to obtain a corresponding image quality score; An image extraction unit is connected to the quality evaluation unit and is used to select the candidate image with the highest image quality score as the optimal thyroid ultrasound image.

6. A method for scanning the thyroid gland based on AI assistance, characterized in that: A thyroid scanning system as claimed in any one of claims 1 to 5, comprising: Step S1, the thyroid scanning system obtains a video stream containing multiple frames of thyroid ultrasound images during the scanning process of the patient, divides the video stream into multiple video segments in chronological order and associates them with time intervals; Step S2, the thyroid scanning system obtains, for each of the video clips, extracts the optimal thyroid ultrasound image from the video clip, uploads the optimal thyroid ultrasound image to the cloud analysis system to obtain the corresponding cloud analysis result, and assigns each of the video clips to a plurality of AI analysis lightweight models configured in the host of the portable ultrasound scanning device to obtain the corresponding local analysis result; Step S3, the thyroid system is scanned to obtain and establish a cache queue, the video clips that have obtained the local analysis results and the cloud analysis results are added to the cache queue, and the order in the cache queue is adjusted according to the associated time interval, and then each video clip in the cache queue and its associated local analysis results and the cloud analysis results are output in sequence.

7. The method for scanning the thyroid gland according to claim 6, characterized in that: The step S1 includes: Step S11, the thyroid scanning system obtains a video stream containing multiple frames of thyroid ultrasound images during the scanning process of the patient by the probe of the portable ultrasound scanning device; In step S12, the thyroid system is scanned by dividing the video stream in sequence according to a preset initial interval length to obtain a plurality of video segments, and then for each video segment, the corresponding system time of the thyroid ultrasound image at the beginning and the end of the video segment is associated with the video segment as the time interval of the video segment.

8. The method for scanning the thyroid gland according to claim 7, characterized in that: The step S1 also includes, before executing the step S11: Step S10, the thyroid scanning system obtains the image processing capability of the AI ​​analysis lightweight model, and adjusts the initial interval length according to the image processing capability; In the step S12, the video stream is divided in sequence according to the adjusted initial interval length to obtain a plurality of the video segments.

9. The method for scanning the thyroid gland according to claim 8, characterized in that: The thyroid scanning system stores a standard video clip with an interval length equal to the initial interval length, and the standard video clip is associated with a corresponding standard processing duration; The step S10 comprises: Step S101, the thyroid scanning system inputs the standard video clip into the AI ​​analysis lightweight model multiple times to obtain multiple test processing times, and uses the average value of all the test processing times as the image processing capability of the AI ​​analysis lightweight model; Step S102: the thyroid scanning system calculates a ratio of the image processing capability to the standard processing time as an adjustment ratio, and multiplies the initial interval length by the adjustment ratio to obtain the adjusted initial interval length.

10. The method for scanning the thyroid gland according to claim 6, characterized in that: The step S2 includes: Step S21, the thyroid scanning system uses each frame of the thyroid ultrasound image contained in each video clip as a candidate image, and then performs a comprehensive scoring quality assessment on each frame of the candidate image according to clarity, contrast, noise, integrity and artifacts to obtain a corresponding image quality score; Step S22, the thyroid scanning system uses the candidate image with the highest image quality score as the optimal thyroid ultrasound image, and uploads the optimal thyroid ultrasound image to the cloud analysis system to obtain a corresponding cloud analysis result; In step S23, the thyroid scanning system distributes each of the video clips to a plurality of AI analysis lightweight models configured in a host of the portable ultrasound scanning device to obtain corresponding local analysis results.

Images