Ultrasonic contrast video preprocessing method, device, equipment and medium

Abstract

The invention discloses an ultrasonic contrast video preprocessing method, device and equipment and a medium, and belongs to the field of ultrasonic contrast video preprocessing. The method comprises the following steps: firstly, acquiring an ultrasound contrast video data set containing different ultrasound types, inputting the ultrasound contrast video data set into a first deep neural network model to obtain an ultrasound contrast timestamp image data set, and then inputting the ultrasound contrast video data set into a second deep neural network model for time period division to obtain an arterial phase video data set, a portal phase video data set and a delay phase video data set; inputting the video data set in each period into a third deep neural network model for information desensitization and image cutting; and finally, removing redundant image frames in each data set based on a similarity calculation method to obtain a processed video data set in each period. Therefore, by implementing the method and the device, the problem of low ultrasonic contrast video preprocessing efficiency in the prior art can be solved.

Description

Technical Field

[0001] The present invention relates to the field of ultrasound contrast imaging video preprocessing, and in particular to an ultrasound contrast imaging video preprocessing method, device, equipment and medium. Background Art

[0002] In recent years, contrast-enhanced ultrasound (CEUS) has gradually become an indispensable and important means in clinical diagnosis due to its advantages such as no radiation, easy operation, low cost, and real-time observation of liver structure and dynamic blood flow changes. It has played an important role in the diagnosis, efficacy evaluation and postoperative follow-up of liver diseases.

[0003] However, in the field of ultrasound contrast imaging video preprocessing, ultrasound contrast imaging videos contain a large number of similar frames. The video formats, resolutions, and image features generated by different ultrasound devices vary greatly. The timestamps and contrast imaging area distribution positions of different models are not fixed, and noise or occlusion may also exist. Existing technologies often rely on manual operations to complete time region interception, privacy information desensitization, and redundant image frame removal. Due to the low efficiency and time consumption of manual operations, and the significant influence of subjective factors, the compatibility and processing accuracy of videos from different models are insufficient. Therefore, it is difficult to meet the growing clinical needs and cannot complete ultrasound contrast imaging video preprocessing efficiently and accurately. Summary of the Invention

[0004] The present invention provides an ultrasound contrast imaging video preprocessing method, device, equipment and medium, which can solve the problem of low efficiency of ultrasound contrast imaging video preprocessing in the prior art.

[0005] In a first aspect, an embodiment of the present invention provides a method for preprocessing ultrasound contrast imaging videos, comprising:

[0006] Acquire an ultrasound contrast imaging video data set; wherein the ultrasound contrast imaging video data set includes ultrasound contrast imaging video data of a plurality of different ultrasound machine models, and each ultrasound contrast imaging video data includes a plurality of image frames arranged in chronological order;

[0007] Inputting the ultrasound contrast-enhanced video dataset into a first deep neural network model to obtain an ultrasound contrast-enhanced timestamp image dataset, and then inputting the ultrasound contrast-enhanced timestamp image dataset into a second deep neural network model for time period division to obtain a first arterial phase video dataset, a first portal venous phase video dataset, and a first delayed phase video dataset;

[0008] Inputting the first arterial phase video dataset, the first portal venous phase video dataset, and the first delayed phase video dataset into a third deep neural network model for information desensitization and image segmentation to obtain a second arterial phase video dataset, a second portal venous phase video dataset, and a second delayed phase video dataset, respectively;

[0009] Redundant image frames in the second arterial phase video dataset, the second portal phase video dataset, and the second delayed phase video dataset are removed based on a similarity calculation method to obtain a third arterial phase video dataset, a third portal phase video dataset, and a third delayed phase video dataset, respectively.

[0010] Compared with the prior art, the above embodiment has the following beneficial effects: by obtaining an ultrasound contrast imaging video data set containing multiple ultrasound models, the present application can be compatible with and process multiple ultrasound models. Then, the ultrasound contrast imaging video data set is input into the first deep neural network model to obtain an ultrasound contrast imaging timestamp image data set, and then the ultrasound contrast imaging timestamp image data set is input into the second deep neural network model for time period division. The present application solves the problem of non-fixed timestamp positions of different models and low efficiency of manual division, so that users can obtain the divided video data of each period; the divided video data sets of each period are subjected to information desensitization and image cutting, which can effectively remove irrelevant information and accurately retain the contrast imaging area, while protecting the patient's privacy, improving the pertinence and purity of the image, and providing a more focused effective area for subsequent clinical analysis; based on the similarity calculation method, redundant image frames in the video data sets of each period are removed, which can significantly reduce the amount of data while retaining important information, reduce storage costs and processing burden, and improve data processing efficiency. Therefore, through the present application, the problem of low efficiency of ultrasound contrast imaging video preprocessing in the prior art can be solved.

[0011] As a preferred example of the first aspect, the method of removing redundant image frames from the second arterial phase video dataset, the second portal venous phase video dataset, and the second delayed phase video dataset based on the similarity calculation method to obtain a third arterial phase video dataset, a third portal venous phase video dataset, and a third delayed phase video dataset, respectively, includes:

[0012] traversing all image frames in the second arterial phase video dataset, the second portal venous phase video dataset, and the second delayed phase video dataset in chronological order, and calculating structural similarity and peak signal-to-noise ratio between adjacent image frames in sequence;

[0013] If the structural similarity and peak signal-to-noise ratio of the current adjacent image frames meet the preset first judgment condition, then mark the next frame in the current adjacent image frames as a redundant image frame, otherwise no processing is performed;

[0014] After the traversal is completed, all redundant image frames in the second arterial phase video dataset, the second portal phase video dataset, and the second delayed phase video dataset are removed, and the unlabeled image frames are reordered in chronological order to obtain a third arterial phase video dataset, a third portal phase video dataset, and a third delayed phase video dataset, respectively.

[0015] In this preferred example, the image frames of each video dataset are first traversed in chronological order. Redundant image frames are identified by calculating the structural similarity and peak signal-to-noise ratio between adjacent frames, enabling accurate identification of frames with minimal image changes. Next, redundant image frames are marked using a pre-set first judgment criterion, ensuring objectivity and accuracy in the screening. Removing redundant image frames significantly reduces the data size of the video dataset. Finally, unmarked image frames are reordered chronologically to ensure the temporal continuity and logicality of the video datasets for each period.

[0016] As a preferred example of the first aspect, the calculating the structural similarity and peak signal-to-noise ratio between adjacent image frames includes:

[0017] The adjacent image frames include a first image frame and a second image frame. The structural similarity and peak signal-to-noise ratio calculation formulas of the first image frame and the second image frame are as follows:

[0018]

[0019] Among them, SSIM(x,y) is the structural similarity, PSNR(x,y) is the peak signal-to-noise ratio, μ x is the mean of the first image frame, μ y is the mean of the second image frame, σ x is the standard deviation of the first image frame, σ y is the standard deviation of the second image frame, the value of L is 255, and C1 and C2 are obtained by converting L.

[0020] In this preferred example, by using the above-mentioned structural similarity (SSIM) and peak signal-to-noise ratio (PSNR) calculation formula to measure the similarity of adjacent image frames (the first image frame and the second image frame), the structural features and quality differences between the image frames can be objectively quantified.

[0021] As a preferred example of the first aspect, inputting the ultrasound contrast-enhanced video dataset into a first deep neural network model to obtain an ultrasound contrast-enhanced timestamp image dataset, and then inputting the ultrasound contrast-enhanced timestamp image dataset into a second deep neural network model for time period division to obtain a first arterial phase video dataset, a first portal venous phase video dataset, and a first delayed phase video dataset, including:

[0022] Inputting the ultrasound contrast-enhanced video data set into a first deep neural network model, outputting the timestamp of each image frame in each ultrasound contrast-enhanced video data, and then making a one-to-one correspondence between each image frame in each ultrasound contrast-enhanced video data and the timestamp to obtain an ultrasound contrast-enhanced timestamp image data set;

[0023] The ultrasound contrast imaging timestamp image dataset is input into a second deep neural network model for time period division, and a first arterial phase video dataset, a first portal venous phase video dataset, and a first delayed phase video dataset are output.

[0024] In this preferred example, by inputting the ultrasound contrast-enhanced video dataset into a first deep neural network model, the timestamps for each image frame are automatically output and a one-to-one correspondence is achieved. This effectively and accurately constructs a timestamp ultrasound contrast-enhanced image dataset, resolving the compatibility issue of timestamp processing across different ultrasound models. This timestamp image dataset is then input into a second deep neural network model for time segmentation, automatically generating the first arterial phase, portal venous phase, and delayed phase video datasets. This allows users to more accurately capture video data for each phase, improving the overall efficiency and quality of the preprocessing process.

[0025] In a second aspect, the present invention provides an ultrasound contrast imaging video preprocessing device, comprising: a data acquisition module, a video interception module, a video cutting module, and a video frame reduction module;

[0026] The data acquisition module is configured to acquire an ultrasound contrast imaging video data set; wherein the ultrasound contrast imaging video data set includes ultrasound contrast imaging video data of a plurality of different ultrasound machine models, and each ultrasound contrast imaging video data set includes a plurality of image frames arranged in chronological order;

[0027] The video capture module is configured to input the ultrasound contrast-enhanced video dataset into a first deep neural network model to obtain an ultrasound contrast-enhanced timestamp image dataset, and further input the ultrasound contrast-enhanced timestamp image dataset into a second deep neural network model for time period division to obtain a first arterial phase video dataset, a first portal venous phase video dataset, and a first delayed phase video dataset;

[0028] The video segmentation module is configured to input the first arterial phase video dataset, the first portal venous phase video dataset, and the first delayed phase video dataset into a third deep neural network model for information desensitization and image segmentation, thereby obtaining a second arterial phase video dataset, a second portal venous phase video dataset, and a second delayed phase video dataset, respectively;

[0029] The video frame reduction module is used to remove redundant image frames in the second arterial phase video dataset, the second portal phase video dataset, and the second delayed phase video dataset based on a similarity calculation method to obtain a third arterial phase video dataset, a third portal phase video dataset, and a third delayed phase video dataset, respectively.

[0030] As a preferred example of the second aspect, the video frame reduction module includes: a calculation unit, a judgment unit and a removal unit;

[0031] The calculation unit is configured to sequentially traverse all image frames in the second arterial phase video dataset, the second portal venous phase video dataset, and the second delayed phase video dataset in chronological order, and sequentially calculate structural similarity and peak signal-to-noise ratio between adjacent image frames;

[0032] The judgment unit is configured to mark the next frame among the current adjacent image frames as a redundant image frame if the structural similarity and the peak signal-to-noise ratio of the current adjacent image frames meet a preset first judgment condition, and not process the next frame otherwise;

[0033] The removal unit is configured to remove all redundant image frames from the second arterial phase video dataset, the second portal venous phase video dataset, and the second delayed phase video dataset after the traversal is completed, and reorder the unmarked image frames in chronological order to obtain a third arterial phase video dataset, a third portal venous phase video dataset, and a third delayed phase video dataset, respectively.

[0034] As a preferred example of the second aspect, the calculating the structural similarity and peak signal-to-noise ratio between adjacent image frames includes:

[0035] The adjacent image frames include a first image frame and a second image frame. The structural similarity and peak signal-to-noise ratio calculation formulas of the first image frame and the second image frame are as follows:

[0036]

[0037] Among them, SSIM(x,y) is the structural similarity, PSNR(x,y) is the peak signal-to-noise ratio, μ x is the mean of the first image frame, μ y is the mean of the second image frame, σ x is the standard deviation of the first image frame, σ y is the standard deviation of the second image frame, the value of L is 255, and C1 and C2 are obtained by converting L.

[0038] As a preferred example of the second aspect, the video capture module includes a first processing unit and a second processing unit;

[0039] The first processing unit is configured to input the ultrasound contrast-enhanced video data set into a first deep neural network model, output a timestamp of each image frame in each ultrasound contrast-enhanced video data, and then associate each image frame in each ultrasound contrast-enhanced video data with the timestamp one-to-one to obtain an ultrasound contrast-enhanced timestamp image data set;

[0040] The second processing unit is configured to input the ultrasound contrast imaging timestamp image dataset into a second deep neural network model for time period division, and output a first arterial phase video dataset, a first portal venous phase video dataset, and a first delayed phase video dataset.

[0041] In summary, compared with the prior art, the above embodiment has the following beneficial effects: by obtaining an ultrasound contrast imaging video data set containing multiple ultrasound models, the present application can be compatible with and process multiple ultrasound models. Then, the ultrasound contrast imaging video data set is input into the first deep neural network model to obtain an ultrasound contrast imaging timestamp image data set, and then the ultrasound contrast imaging timestamp image data set is input into the second deep neural network model for time period division. The present application solves the problem of non-fixed timestamp positions of different models and low efficiency of manual division, so that users can obtain the divided video data of each period; the divided video data sets of each period are subjected to information desensitization and image cutting, which can effectively remove irrelevant information and accurately retain the contrast imaging area, while protecting the patient's privacy, improving the pertinence and purity of the image, and providing a more focused effective area for subsequent clinical analysis; based on the similarity calculation method, redundant image frames in the video data sets of each period are removed, which can significantly reduce the amount of data while retaining important information, reduce storage costs and processing burden, and improve data processing efficiency. Therefore, through the present application, the problem of low efficiency of ultrasound contrast imaging video preprocessing in the prior art can be solved.

[0042] Another embodiment of the present invention further provides a terminal device, comprising: a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, the steps of the ultrasound contrast imaging video preprocessing method of the present invention are implemented.

[0043] Another embodiment of the present invention further provides a computer-readable storage medium item, comprising: a stored computer program, which controls the device where the computer-readable storage medium is located to execute the steps of the ultrasound contrast imaging video preprocessing method of the present invention when the computer program is running. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the technical solution of the present application, the following is a brief introduction to the drawings required for use in the implementation. Obviously, the drawings described below are only some implementation methods of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0045] Figure 1 A schematic flow chart of an embodiment of a method for preprocessing ultrasound contrast imaging videos provided by the present invention;

[0046] Figure 2 A schematic diagram of video capture of an embodiment of a method for preprocessing ultrasound contrast imaging videos provided by the present invention;

[0047] Figure 3A schematic diagram of video cutting according to an embodiment of a method for preprocessing ultrasound contrast imaging videos provided by the present invention;

[0048] Figure 4 A schematic diagram of video frame reduction according to an embodiment of a method for preprocessing ultrasound contrast imaging videos provided by the present invention;

[0049] Figure 5 This is a module structure diagram of an embodiment of an ultrasound contrast imaging video preprocessing device provided by the present invention. DETAILED DESCRIPTION

[0050] To make the objectives, technical solutions, and advantages of this application more clear, the technical solutions in this application will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making any creative efforts are within the scope of protection of this application.

[0051] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this application belongs; the terms used herein are only for the purpose of describing specific embodiments and are not intended to limit this application; the terms "including" and "having" and any variations thereof in the specification and claims of this application and the above-mentioned figure descriptions are intended to cover non-exclusive inclusions.

[0052] In the description of the embodiments of this application, the technical terms "first" and "second" are used only to distinguish different objects and should not be understood to indicate or imply relative importance or implicitly specify the quantity, specific order, or primary and secondary relationship of the indicated technical features. In the description of the embodiments of this application, the meaning of "plurality" is more than two, unless otherwise clearly and specifically defined.

[0053] References herein to "embodiments" mean that a particular feature, structure, or characteristic described in connection with the embodiments may be included in at least one embodiment of the present application. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it constitute an independent or alternative embodiment that is mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0054] In the description of the embodiments of this application, the term "and / or" is simply a description of the association relationship between associated objects, indicating that three relationships can exist. For example, A and / or B can represent the following three situations: A exists alone, A and B exist simultaneously, and B exists alone. In addition, the character " / " in this document generally indicates that the associated objects are in an "or" relationship.

[0055] In the description of the embodiments of the present application, the term "multiple" refers to more than two (including two). Similarly, "multiple groups" refers to more than two groups (including two groups), and "multiple pieces" refers to more than two pieces (including two pieces).

[0056] In the description of the embodiments of the present application, unless otherwise expressly specified or limited, technical terms such as "installed," "connected," "connected," and "fixed" should be understood in a broad sense. For example, they can refer to fixed connections, detachable connections, or integration; mechanical connections or electrical connections; direct connections or indirect connections through an intermediate medium; internal connections between two components or interactions between two components. Those skilled in the art can understand the specific meanings of the above terms in the embodiments of the present application based on specific circumstances.

[0057] In recent years, contrast-enhanced ultrasound (CEUS) has gradually become an indispensable and important tool in clinical diagnosis due to its advantages such as being radiation-free, easy to use, low-cost, and capable of real-time observation of liver structure and dynamic blood flow changes. It has played a particularly important role in the diagnosis, efficacy evaluation, and postoperative follow-up of liver diseases. However, the efficient analysis and processing of CEUS video data still faces many challenges: dynamic ultrasound video contains a large number of similar frames. This redundant information not only takes up storage space but also increases the computational complexity of subsequent data processing and analysis. The video formats, resolutions, and image features generated by different ultrasound devices vary significantly, and the lack of a unified processing standard increases the complexity of information processing. The timestamps and contrast region distribution locations of different models are not fixed and may be subject to noise or occlusion, increasing the difficulty of accurate identification and segmentation. Existing technologies often rely on manual operations to complete time region interception, privacy information desensitization, and redundant image frame removal, which is difficult to meet the growing clinical needs.

[0058] Example 1

[0059] See also Figure 1 To solve the problem of low efficiency in ultrasound contrast imaging video preprocessing in the prior art, an embodiment of the present invention provides an ultrasound contrast imaging video preprocessing method, comprising:

[0060] S1. Acquire an ultrasound contrast imaging video dataset; wherein the ultrasound contrast imaging video dataset includes ultrasound contrast imaging video data of a plurality of different ultrasound machine models, and each ultrasound contrast imaging video data includes a plurality of image frames arranged in chronological order;

[0061] Specifically, the ultrasound contrast imaging video data in the ultrasound contrast imaging video dataset may be derived from a variety of ultrasound models, cover a variety of clinical scenarios and patient data, and include videos with different resolutions, frame rates, and contrast agent injection conditions.

[0062] S2. Inputting the ultrasound contrast-enhanced video dataset into a first deep neural network model to obtain an ultrasound contrast-enhanced timestamp image dataset, and then inputting the ultrasound contrast-enhanced timestamp image dataset into a second deep neural network model for time period division to obtain a first arterial phase video dataset, a first portal venous phase video dataset, and a first delayed phase video dataset, respectively;

[0063] Furthermore, in some embodiments of the present application, inputting the ultrasound contrast-enhanced video dataset into a first deep neural network model to obtain an ultrasound contrast-enhanced timestamp image dataset, and then inputting the ultrasound contrast-enhanced timestamp image dataset into a second deep neural network model for time period division to obtain a first arterial phase video dataset, a first portal venous phase video dataset, and a first delayed phase video dataset, includes:

[0064] Inputting the ultrasound contrast-enhanced video data set into a first deep neural network model, outputting the timestamp of each image frame in each ultrasound contrast-enhanced video data, and then making a one-to-one correspondence between each image frame in each ultrasound contrast-enhanced video data and the timestamp to obtain an ultrasound contrast-enhanced timestamp image data set;

[0065] The ultrasound contrast imaging timestamp image dataset is input into a second deep neural network model for time period division, and a first arterial phase video dataset, a first portal venous phase video dataset, and a first delayed phase video dataset are output.

[0066] Specifically, the first deep neural network model can be constructed in the following way:

[0067] YOLOv8 (the first deep neural network model) uses CSPDarknet53 as the backbone network, combined with the path aggregation network (PANet) to achieve multi-scale feature fusion, output detection box coordinates and confidence. The training set contains labeled data for various ultrasound models. The labeled area is the timestamp area in the image, and the timestamp is located in various positions such as the upper left, lower right, and middle of the screen. Data enhancement methods include random rotation, scaling, and brightness adjustment to improve cross-model generalization capabilities.

[0068] Specifically, the second deep neural network model can be constructed in the following way:

[0069] The CnOCR (i.e., the second deep neural network model) architecture is a ResNet34-BiLSTM-CTC combination. The pre-trained weights come from public Chinese-English mixed text datasets (ICDAR2015, LSVT, etc.), and are fine-tuned based on the characteristics of ultrasound timestamp text (low contrast, blurred text). The training data is manually annotated ultrasound timestamp images.

[0070] Specifically, the ultrasound contrast imaging time-stamp image dataset is input into a second deep neural network model for time period division to obtain a first arterial phase video dataset, a first portal venous phase video dataset, and a first delayed phase video dataset, respectively. This can be implemented by the following scheme:

[0071] like Figure 2 As shown, according to the preset three-period division criteria (arterial phase 0-40 seconds, portal venous phase 40-120 seconds and delayed phase 121-360 seconds), the input video is accurately intercepted through the second deep neural network model, and the video clips corresponding to each period are output and saved in the preset folder.

[0072] By inputting the ultrasound contrast imaging video dataset into the first deep neural network model, the timestamps of each image frame are automatically output and a one-to-one correspondence is achieved. This effectively and accurately constructs a timestamp ultrasound contrast imaging dataset, resolving the compatibility issue of timestamp processing across different ultrasound machine models. This timestamp image dataset is then input into the second deep neural network model for time segmentation, automatically generating the first arterial phase, portal venous phase, and delayed phase video datasets. This allows users to more accurately capture video data for each phase, improving the overall efficiency and quality of the preprocessing process.

[0073] S3. Inputting the first arterial phase video dataset, the first portal venous phase video dataset, and the first delayed phase video dataset into a third deep neural network model for information desensitization and image segmentation to obtain a second arterial phase video dataset, a second portal venous phase video dataset, and a second delayed phase video dataset, respectively;

[0074] Specifically, the third deep neural network model can be constructed in the following way:

[0075] The training set contains labeled data for various ultrasound models. The labeled areas are the contrast areas in the images. The positions of the ultrasound contrast areas in images of different models are different. Data enhancement methods include random rotation, scaling, and brightness adjustment to improve the generalization ability across models.

[0076] Specifically, in order to fully explain step S3, the following scheme is used for illustration:

[0077] like Figure 3 As shown, the video is desensitized by the third deep neural network model, the areas containing useless information in the video are cropped out, and the contrast area is cut out.

[0078] S4. Removing redundant image frames from the second arterial phase video dataset, the second portal venous phase video dataset, and the second delayed phase video dataset based on a similarity calculation method to obtain a third arterial phase video dataset, a third portal venous phase video dataset, and a third delayed phase video dataset, respectively.

[0079] Furthermore, in some embodiments of the present application, removing redundant image frames from the second arterial phase video dataset, the second portal venous phase video dataset, and the second delayed phase video dataset based on the similarity calculation method to obtain a third arterial phase video dataset, a third portal venous phase video dataset, and a third delayed phase video dataset, respectively, includes:

[0080] traversing all image frames in the second arterial phase video dataset, the second portal venous phase video dataset, and the second delayed phase video dataset in chronological order, and calculating structural similarity and peak signal-to-noise ratio between adjacent image frames in sequence;

[0081] If the structural similarity and peak signal-to-noise ratio of the current adjacent image frames meet the preset first judgment condition, then mark the next frame in the current adjacent image frames as a redundant image frame, otherwise no processing is performed;

[0082] After the traversal is completed, all redundant image frames in the second arterial phase video dataset, the second portal phase video dataset, and the second delayed phase video dataset are removed, and the unlabeled image frames are reordered in chronological order to obtain a third arterial phase video dataset, a third portal phase video dataset, and a third delayed phase video dataset, respectively.

[0083] Specifically, the preset first judgment condition may be:

[0084] Through experimental analysis of multiple groups of ultrasound videos of various ultrasound models, the SSIM and PSNR distributions of adjacent image frames were statistically analyzed, and redundant image frames were determined when the SSIM threshold was set to ≥0.98 or the PSNR was ≥40dB.

[0085] Specifically, in order to fully explain step S4, the following scheme is used for illustration:

[0086] like Figure 4 As shown in the figure, through the similarity calculation method of image quality evaluation indicators, the structural similarity and peak signal-to-noise ratio are used to calculate the similarity of adjacent frames, and redundant image frames are screened out. Then, these redundant image frames are deleted while ensuring that the specified number of frames are retained at the end, and the video after frame reduction is output and saved.

[0087] This approach first traverses the image frames of each video dataset in chronological order. Redundant frames are identified by calculating the structural similarity and peak signal-to-noise ratio between adjacent frames, accurately identifying frames with minimal image changes. Next, redundant frames are marked using a pre-set first judgment criterion, ensuring objectivity and accuracy in screening. Removing redundant frames significantly reduces the size of the video dataset. Finally, unmarked frames are reordered chronologically to ensure temporal continuity and logical consistency within each video dataset.

[0088] Furthermore, in some embodiments of the present application, the calculating of the structural similarity and peak signal-to-noise ratio between adjacent image frames includes:

[0089] The adjacent image frames include a first image frame and a second image frame. The structural similarity and peak signal-to-noise ratio calculation formulas of the first image frame and the second image frame are as follows:

[0090]

[0091] Among them, SSIM(x,y) is the structural similarity, PSNR(x,y) is the peak signal-to-noise ratio, μ x is the mean of the first image frame, μ y is the mean of the second image frame, σ x is the standard deviation of the first image frame, σ y is the standard deviation of the second image frame, the value of L is 255, and C1 and C2 are obtained by converting L.

[0092] In this way, by using the above-mentioned structural similarity (SSIM) and peak signal-to-noise ratio (PSNR) calculation formula to measure the similarity of adjacent image frames (the first image frame and the second image frame), the structural characteristics and quality differences between the image frames can be objectively quantified.

[0093] In summary, compared with the prior art, the above embodiment has the following beneficial effects: by obtaining an ultrasound contrast imaging video data set containing multiple ultrasound models, the present application can be compatible with and process multiple ultrasound models. Then, the ultrasound contrast imaging video data set is input into the first deep neural network model to obtain an ultrasound contrast imaging timestamp image data set, and then the ultrasound contrast imaging timestamp image data set is input into the second deep neural network model for time period division. The present application solves the problem of non-fixed timestamp positions of different models and low efficiency of manual division, so that users can obtain the divided video data of each period; the divided video data sets of each period are subjected to information desensitization and image cutting, which can effectively remove irrelevant information and accurately retain the contrast imaging area, while protecting the patient's privacy, improving the pertinence and purity of the image, and providing a more focused effective area for subsequent clinical analysis; based on the similarity calculation method, redundant image frames in the video data sets of each period are removed, which can significantly reduce the amount of data while retaining important information, reduce storage costs and processing burden, and improve data processing efficiency. Therefore, through the present application, the problem of low efficiency of ultrasound contrast imaging video preprocessing in the prior art can be solved.

[0094] Example 2

[0095] like Figure 5 As shown, based on the above method embodiment, a corresponding device embodiment is provided;

[0096] An embodiment of the present invention provides an ultrasound contrast imaging video preprocessing device, comprising: a data acquisition module 51, a video interception module 52, a video cutting module 53 and a video frame reduction module 54;

[0097] The data acquisition module 51 is configured to acquire an ultrasound contrast imaging video data set; wherein the ultrasound contrast imaging video data set includes ultrasound contrast imaging video data of a plurality of different ultrasound machine models, and each ultrasound contrast imaging video data set includes a plurality of image frames arranged in chronological order;

[0098] a video capture module 52 configured to input the ultrasound contrast-enhanced video dataset into a first deep neural network model to obtain an ultrasound contrast-enhanced time-stamp image dataset, and further input the ultrasound contrast-enhanced time-stamp image dataset into a second deep neural network model for time segmentation to obtain a first arterial phase video dataset, a first portal venous phase video dataset, and a first delayed phase video dataset;

[0099] a video segmentation module 53 configured to input the first arterial phase video dataset, the first portal venous phase video dataset, and the first delayed phase video dataset into a third deep neural network model for information desensitization and image segmentation, thereby obtaining a second arterial phase video dataset, a second portal venous phase video dataset, and a second delayed phase video dataset, respectively;

[0100] The video frame reduction module 54 is configured to remove redundant image frames from the second arterial phase video dataset, the second portal venous phase video dataset, and the second delayed phase video dataset based on a similarity calculation method, to obtain a third arterial phase video dataset, a third portal venous phase video dataset, and a third delayed phase video dataset, respectively.

[0101] Furthermore, in some embodiments of the present application, the video frame reduction module 54 includes: a calculation unit, a judgment unit, and a removal unit;

[0102] The calculation unit is configured to sequentially traverse all image frames in the second arterial phase video dataset, the second portal venous phase video dataset, and the second delayed phase video dataset in chronological order, and sequentially calculate structural similarity and peak signal-to-noise ratio between adjacent image frames;

[0103] The judgment unit is configured to mark the next frame among the current adjacent image frames as a redundant image frame if the structural similarity and the peak signal-to-noise ratio of the current adjacent image frames meet a preset first judgment condition, and not process the next frame otherwise;

[0104] The removal unit is configured to remove all redundant image frames from the second arterial phase video dataset, the second portal venous phase video dataset, and the second delayed phase video dataset after the traversal is completed, and reorder the unmarked image frames in chronological order to obtain a third arterial phase video dataset, a third portal venous phase video dataset, and a third delayed phase video dataset, respectively.

[0105] Furthermore, in some embodiments of the present application, the calculating of the structural similarity and peak signal-to-noise ratio between adjacent image frames includes:

[0106] The adjacent image frames include a first image frame and a second image frame. The structural similarity and peak signal-to-noise ratio calculation formulas of the first image frame and the second image frame are as follows:

[0107]

[0108] Among them, SSIM(x,y) is the structural similarity, PSNR(x,y) is the peak signal-to-noise ratio, μ x is the mean of the first image frame, μ y is the mean of the second image frame, σ x is the standard deviation of the first image frame, σ y is the standard deviation of the second image frame, the value of L is 255, and C1 and C2 are obtained by converting L.

[0109] Furthermore, in some embodiments of the present application, the video capture module 52 includes a first processing unit and a second processing unit;

[0110] The first processing unit is configured to input the ultrasound contrast-enhanced video data set into a first deep neural network model, output a timestamp of each image frame in each ultrasound contrast-enhanced video data, and then associate each image frame in each ultrasound contrast-enhanced video data with the timestamp one-to-one to obtain an ultrasound contrast-enhanced timestamp image data set;

[0111] The second processing unit is configured to input the ultrasound contrast imaging timestamp image dataset into a second deep neural network model for time period division, and output a first arterial phase video dataset, a first portal venous phase video dataset, and a first delayed phase video dataset.

[0112] For more detailed steps and working principles of this embodiment, please refer to, but not limited to, the relevant records of Embodiment 1.

[0113] In summary, compared with the prior art, the above embodiment has the following beneficial effects: by obtaining an ultrasound contrast imaging video data set containing multiple ultrasound models, the present application can be compatible with and process multiple ultrasound models. Then, the ultrasound contrast imaging video data set is input into the first deep neural network model to obtain an ultrasound contrast imaging timestamp image data set, and then the ultrasound contrast imaging timestamp image data set is input into the second deep neural network model for time period division. The present application solves the problem of non-fixed timestamp positions of different models and low efficiency of manual division, so that users can obtain the divided video data of each period; the divided video data sets of each period are subjected to information desensitization and image cutting, which can effectively remove irrelevant information and accurately retain the contrast imaging area, while protecting the patient's privacy, improving the pertinence and purity of the image, and providing a more focused effective area for subsequent clinical analysis; based on the similarity calculation method, redundant image frames in the video data sets of each period are removed, which can significantly reduce the amount of data while retaining important information, reduce storage costs and processing burden, and improve data processing efficiency. Therefore, through the present application, the problem of low efficiency of ultrasound contrast imaging video preprocessing in the prior art can be solved.

[0114] It can be understood that the above-mentioned device embodiment corresponds to the method embodiment of the present invention, and can implement any one of the above-mentioned method embodiments of the present invention to provide an ultrasound imaging video preprocessing method.

[0115] It should be noted that the device embodiments described above are merely illustrative, and some or all of the modules may be selected according to actual needs to achieve the purpose of the present embodiment. Furthermore, in the drawings of the device embodiments provided by the present invention, the connection relationship between modules indicates that they have a communication connection, which may be implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement the present invention without inventive effort.

[0116] Example 3

[0117] Based on the above-mentioned embodiment of the ultrasound contrast imaging video preprocessing method, another embodiment of the present invention provides a terminal device, which includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, the ultrasound contrast imaging video preprocessing method of any embodiment of the present invention is implemented.

[0118] For example, in this embodiment, the computer program may be divided into one or more modules, which are stored in the memory and executed by the processor to implement the present invention. The one or more module elements may be a series of computer program instruction segments capable of performing specific functions, and the instruction segments are used to describe the execution process of the computer program in the terminal device.

[0119] The terminal device may be a computing device such as a desktop computer, a notebook computer, a PDA, a cloud server, etc. The terminal device may include, but is not limited to, a processor and a memory.

[0120] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the terminal device, connecting various parts of the entire terminal device using various interfaces and lines.

[0121] Example 4

[0122] Based on the above-mentioned method embodiments, another embodiment of the present invention provides a computer-readable storage medium, including a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute the ultrasound contrast imaging video preprocessing method described in any one of the above-mentioned method embodiments of the present invention.

[0123] Wherein, the module / unit integrated in the device / terminal equipment, if implemented in the form of a software functional unit and sold or used as an independent product, can be stored in a computer-readable storage medium. Based on this understanding, the present invention implements all or part of the process in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and when the computer program is executed by the processor, it can implement the steps of the above-mentioned various method embodiments. Wherein, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device that can carry the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium, etc.

[0124] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A method for preprocessing ultrasound contrast imaging video, characterized in that: include: Acquire an ultrasound contrast imaging video data set; wherein the ultrasound contrast imaging video data set includes ultrasound contrast imaging video data of a plurality of different ultrasound machine models, and each ultrasound contrast imaging video data includes a plurality of image frames arranged in chronological order; Inputting the ultrasound contrast-enhanced video dataset into a first deep neural network model to obtain an ultrasound contrast-enhanced timestamp image dataset, and then inputting the ultrasound contrast-enhanced timestamp image dataset into a second deep neural network model for time period division to obtain a first arterial phase video dataset, a first portal venous phase video dataset, and a first delayed phase video dataset, respectively; Inputting the first arterial phase video dataset, the first portal venous phase video dataset, and the first delayed phase video dataset into a third deep neural network model for information desensitization and image segmentation to obtain a second arterial phase video dataset, a second portal venous phase video dataset, and a second delayed phase video dataset, respectively; Redundant image frames in the second arterial phase video dataset, the second portal phase video dataset, and the second delayed phase video dataset are removed based on a similarity calculation method to obtain a third arterial phase video dataset, a third portal phase video dataset, and a third delayed phase video dataset, respectively.

2. The ultrasound contrast imaging video preprocessing method according to claim 1, wherein: The method of removing redundant image frames from the second arterial phase video dataset, the second portal venous phase video dataset, and the second delayed phase video dataset based on the similarity calculation method to obtain a third arterial phase video dataset, a third portal venous phase video dataset, and a third delayed phase video dataset, respectively, includes: traversing all image frames in the second arterial phase video dataset, the second portal venous phase video dataset, and the second delayed phase video dataset in chronological order, and calculating structural similarity and peak signal-to-noise ratio between adjacent image frames in sequence; If the structural similarity and peak signal-to-noise ratio of the current adjacent image frames meet the preset first judgment condition, then mark the next frame in the current adjacent image frames as a redundant image frame, otherwise no processing is performed; After the traversal is completed, all redundant image frames in the second arterial phase video dataset, the second portal phase video dataset, and the second delayed phase video dataset are removed, and the unlabeled image frames are reordered in chronological order to obtain a third arterial phase video dataset, a third portal phase video dataset, and a third delayed phase video dataset, respectively.

3. The ultrasound contrast imaging video preprocessing method according to claim 2, wherein: The calculating of the structural similarity and the peak signal-to-noise ratio between adjacent image frames includes: The adjacent image frames include a first image frame and a second image frame. The structural similarity and peak signal-to-noise ratio calculation formulas of the first image frame and the second image frame are as follows: Among them, SSIM(x,y) is the structural similarity, PSNR(x,y) is the peak signal-to-noise ratio, μ x is the mean of the first image frame, μ y is the mean of the second image frame, σ x is the standard deviation of the first image frame, σ y is the standard deviation of the second image frame, the value of L is 255, and C1 and C2 are obtained by converting L.

4. The ultrasound contrast imaging video preprocessing method according to claim 1, wherein: The step of inputting the ultrasound contrast-enhanced video dataset into a first deep neural network model to obtain an ultrasound contrast-enhanced timestamp image dataset, and then inputting the ultrasound contrast-enhanced timestamp image dataset into a second deep neural network model for time period division to obtain a first arterial phase video dataset, a first portal venous phase video dataset, and a first delayed phase video dataset, comprising: Inputting the ultrasound contrast-enhanced video data set into a first deep neural network model, outputting the timestamp of each image frame in each ultrasound contrast-enhanced video data, and then making a one-to-one correspondence between each image frame in each ultrasound contrast-enhanced video data and the timestamp to obtain an ultrasound contrast-enhanced timestamp image data set; The ultrasound contrast imaging timestamp image dataset is input into a second deep neural network model for time period division, and a first arterial phase video dataset, a first portal venous phase video dataset, and a first delayed phase video dataset are output.

5. An ultrasound contrast imaging video preprocessing device, characterized in that: include: Data acquisition module, video interception module, video cutting module and video frame reduction module; The data acquisition module is configured to acquire an ultrasound contrast imaging video data set; wherein the ultrasound contrast imaging video data set includes ultrasound contrast imaging video data of a plurality of different ultrasound machine models, and each ultrasound contrast imaging video data set includes a plurality of image frames arranged in chronological order; The video capture module is configured to input the ultrasound contrast-enhanced video dataset into a first deep neural network model to obtain an ultrasound contrast-enhanced timestamp image dataset, and further input the ultrasound contrast-enhanced timestamp image dataset into a second deep neural network model for time period division to obtain a first arterial phase video dataset, a first portal venous phase video dataset, and a first delayed phase video dataset; The video segmentation module is configured to input the first arterial phase video dataset, the first portal venous phase video dataset, and the first delayed phase video dataset into a third deep neural network model for information desensitization and image segmentation, thereby obtaining a second arterial phase video dataset, a second portal venous phase video dataset, and a second delayed phase video dataset, respectively; The video frame reduction module is used to remove redundant image frames in the second arterial phase video dataset, the second portal phase video dataset, and the second delayed phase video dataset based on a similarity calculation method to obtain a third arterial phase video dataset, a third portal phase video dataset, and a third delayed phase video dataset, respectively.

6. The ultrasound contrast imaging video preprocessing device according to claim 5, characterized in that: The video frame reduction module includes: a calculation unit, a judgment unit and a removal unit; The calculation unit is configured to sequentially traverse all image frames in the second arterial phase video dataset, the second portal venous phase video dataset, and the second delayed phase video dataset in chronological order, and sequentially calculate structural similarity and peak signal-to-noise ratio between adjacent image frames; The judgment unit is configured to mark the next frame among the current adjacent image frames as a redundant image frame if the structural similarity and the peak signal-to-noise ratio of the current adjacent image frames meet a preset first judgment condition, and not process the next frame otherwise; The removal unit is configured to remove all redundant image frames from the second arterial phase video dataset, the second portal venous phase video dataset, and the second delayed phase video dataset after the traversal is completed, and reorder the unmarked image frames in chronological order to obtain a third arterial phase video dataset, a third portal venous phase video dataset, and a third delayed phase video dataset, respectively.

7. The ultrasound contrast imaging video preprocessing device according to claim 6, characterized in that: The calculating of the structural similarity and the peak signal-to-noise ratio between adjacent image frames includes: The adjacent image frames include a first image frame and a second image frame. The structural similarity and peak signal-to-noise ratio calculation formulas of the first image frame and the second image frame are as follows: Among them, SSIM(x,y) is the structural similarity, PSNR(x,y) is the peak signal-to-noise ratio, μ x is the mean of the first image frame, μ y is the mean of the second image frame, σ x is the standard deviation of the first image frame, σ y is the standard deviation of the second image frame, the value of L is 255, and C1 and C2 are obtained by converting L.

8. The ultrasound contrast imaging video preprocessing device according to claim 6, characterized in that: The video capture module includes a first processing unit and a second processing unit; The first processing unit is configured to input the ultrasound contrast-enhanced video data set into a first deep neural network model, output a timestamp of each image frame in each ultrasound contrast-enhanced video data, and then associate each image frame in each ultrasound contrast-enhanced video data with the timestamp one-to-one to obtain an ultrasound contrast-enhanced timestamp image data set; The second processing unit is configured to input the ultrasound contrast imaging timestamp image dataset into a second deep neural network model for time period division, and output a first arterial phase video dataset, a first portal venous phase video dataset, and a first delayed phase video dataset.

9. A terminal device, characterized in that: The method comprises a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, the ultrasound contrast imaging video preprocessing method according to any one of claims 1 to 4 is implemented.

10. A computer-readable storage medium, characterized in that include: A stored computer program, wherein when the computer program is run, the device where the computer-readable storage medium is located is controlled to execute the ultrasound contrast imaging video preprocessing method according to any one of claims 1 to 4.

Images