Enhanced scanning triggering method and device, computer equipment and storage medium

Abstract

The invention relates to an enhanced scanning triggering method and device, computer equipment and a storage medium, and the method comprises the steps: obtaining a bolus injection tracking image sequence corresponding to a target scanning region after a contrast agent is injected; the group injection tracking image sequence comprises a current image; inputting the group injection tracking image sequence into a well-trained deep learning network, predicting the current image based on contrast agent inflow information in the current image, and outputting a prediction result corresponding to the current image; the prediction result is used for indicating whether the current image is a trigger frame; and under the condition that the current image is the trigger frame, triggering scanning of the target scanning area. According to the method and the device, the problem of inaccurate time for triggering enhanced scanning is solved, and accurate triggering of enhanced scanning is realized.

Description

Technical Field

[0001] The present application relates to the field of medical imaging technology, and in particular to a triggering method, device, computer equipment and storage medium for enhanced scanning. Background Art

[0002] Enhanced scanning is a common medical examination method. It is a medical scan performed after contrast agent is injected into the blood vessels, such as dynamic enhanced scanning of the abdomen and enhanced scanning of the neck blood vessels. It is also applied to medical imaging technologies such as computed tomography (CT) and magnetic resonance imaging (MRI).

[0003] In the existing enhanced scanning method, a baseline image is obtained after contrast agent is injected, and the corresponding region of interest is automatically obtained based on the baseline image, and the time when the contrast agent reaches the region of interest is predicted, and the scan is triggered based on the predicted time. However, the above method cannot ensure that the appropriate region of interest is obtained when there is a deviation in the baseline image, resulting in inaccurate time to trigger the enhanced scan.

[0004] With respect to the problem of inaccurate time for triggering enhanced scanning in related technologies, no effective solution has been proposed yet. Summary of the invention

[0005] In this embodiment, a method, apparatus, computer device and storage medium for triggering enhanced scanning are provided to solve the problem of inaccurate time for triggering enhanced scanning in the related art.

[0006] In a first aspect, a method for triggering enhanced scanning is provided in this embodiment, and the method includes:

[0007] Acquire a bolus tracking image sequence corresponding to the target scanning area after contrast agent injection; the bolus tracking image sequence includes a current image;

[0008] Inputting the bolus tracking image sequence into a well-trained deep learning network, predicting the current image based on contrast agent inflow information in the current image, and outputting a prediction result corresponding to the current image; the prediction result is used to indicate whether the current image is a trigger frame;

[0009] When the current image is a trigger frame, scanning of the target scanning area is triggered.

[0010] In some embodiments, when the bolus tracking image sequence further includes a plurality of preceding images that are sequentially adjacent to the current image, after acquiring the bolus tracking image sequence corresponding to the target scanning area after injection of contrast agent, the method further includes:

[0011] Inputting the bolus tracking image sequence into the well-trained deep learning network, predicting the current image based on the change of contrast agent inflow information between the current image and each of the previous images, and outputting the prediction result corresponding to the current image; the prediction result is used to indicate whether the current image is a trigger frame;

[0012] When the current image is a trigger frame, scanning of the target scanning area is triggered.

[0013] In some embodiments, predicting the current image based on the contrast agent inflow information in the current image and outputting the prediction result corresponding to the current image includes:

[0014] According to the contrast agent inflow information in the current image, determining whether the contrast agent state in the target scanning area reaches a preset state;

[0015] When the state of the contrast agent in the target scanning area reaches a preset state, the current image is determined to be a trigger frame.

[0016] In some embodiments, when the output of the deep learning network includes the credibility of the prediction result, after outputting the prediction result corresponding to the current image, the method further includes:

[0017] When the current image is a trigger frame and the credibility reaches a preset threshold, scanning of the target scanning area is triggered.

[0018] In some of the embodiments, outputting the credibility of the prediction result through the deep learning network includes:

[0019] Determining, by means of the deep learning network, whether the current image contains contrast agent inflow information;

[0020] Determining the credibility of the prediction result based on the judgment result; and / or,

[0021] Performing quality detection on the current image through the deep learning network;

[0022] According to the detection result, the credibility of the prediction result is determined.

[0023] In some of the embodiments, when the output of the deep learning network includes the credibility of the prediction result, after outputting the corresponding prediction result, the method further includes:

[0024] When the current image is a trigger frame and the credibility does not reach a preset threshold, determining a trigger time based on the injection time of the contrast agent or the start scanning time corresponding to the bolus tracking image sequence;

[0025] According to the trigger time, a scan of the target scanning area is triggered.

[0026] In some of the embodiments, before predicting the current image based on the change of contrast agent inflow information between the current image and each of the previous images through the trained deep learning network, the method further includes:

[0027] Acquire a sample image set; the sample image set includes positive sample images and negative sample images, the positive sample images are bolus tracking images in which the contrast agent inflow state reaches the trigger standard, and the negative sample images are bolus tracking images in which the contrast agent inflow state does not reach the trigger standard;

[0028] The deep learning network is trained based on the sample image set to obtain the fully trained deep learning network.

[0029] In a second aspect, a triggering device for enhanced scanning is provided in this embodiment, the device comprising: an acquisition module, a prediction module and a triggering module;

[0030] The acquisition module is used to acquire a bolus tracking image sequence corresponding to the target scanning area after the contrast agent is injected; the bolus tracking image sequence includes the current image;

[0031] The prediction module is used to input the bolus tracking image sequence into a well-trained deep learning network, predict the current image based on the contrast agent inflow information in the current image, and output a prediction result corresponding to the current image; the prediction result is used to indicate whether the current image is a trigger frame;

[0032] The trigger module is used to trigger scanning of the target scanning area when the current image is a trigger frame.

[0033] In a third aspect, a computer device is provided in this embodiment, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method for triggering enhanced scanning described in the first aspect when executing the computer program.

[0034] In a fourth aspect, in this embodiment, a storage medium is provided, on which a computer program is stored, and when the program is executed by a processor, the method for triggering enhanced scanning described in the first aspect above is implemented.

[0035] Compared with the related art, the enhanced scanning triggering method, device, computer equipment and storage medium provided in this embodiment obtain a bolus tracking image sequence corresponding to the target scanning area after the injection of contrast agent; the bolus tracking image sequence includes the current image; the bolus tracking image sequence is input into a well-trained deep learning network, the current image is predicted based on the contrast agent inflow information in the current image, and the prediction result corresponding to the current image is output; the prediction result is used to indicate whether the current image is a trigger frame; when the current image is a trigger frame, the scanning of the target scanning area is triggered, thereby solving the problem of inaccurate time of triggering enhanced scanning and realizing accurate triggering of enhanced scanning.

[0036] Details of one or more embodiments of the present application are set forth in the following drawings and description to make other features, objects, and advantages of the present application more readily apparent. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0038] Figure 1 It is a hardware structure block diagram of a terminal device of a method for triggering enhanced scanning provided in an embodiment of the present application;

[0039] Figure 2 is a flow chart of a method for triggering enhanced scanning provided in an embodiment of the present application;

[0040] Figure 3 is a schematic diagram of different display states of a target organization provided by an embodiment of the present application;

[0041] Figure 4 is a schematic diagram of a prediction result of a bolus tracking image sequence provided by an embodiment of the present application;

[0042] Figure 5 It is a flowchart of a method for triggering enhanced scanning provided in an embodiment of the present application;

[0043] Figure 6 This is a flow chart of a method for triggering enhanced scanning provided in a preferred embodiment of the present application;

[0044] Figure 7 It is a structural block diagram of a trigger device for enhanced scanning provided in one embodiment of the present application.

[0045] In the figure: 102, processor; 104, memory; 106, transmission device; 108, input and output device; 10, acquisition module; 20, prediction module; 30, trigger module. DETAILED DESCRIPTION

[0046] In order to more clearly understand the purpose, technical solutions and advantages of the present application, the present application is described and illustrated below in conjunction with the accompanying drawings and embodiments.

[0047] Unless otherwise defined, the technical terms or scientific terms involved in this application shall have the general meaning understood by people with ordinary skills in the technical field to which this application belongs. The words "one", "a", "a", "the", "these" and the like in this application do not represent quantitative restrictions, and they can be singular or plural. The terms "include", "comprise", "have" and any variants thereof involved in this application are intended to cover non-exclusive inclusions; for example, a process, method and system, product or device comprising a series of steps or modules (units) is not limited to the listed steps or modules (units), but may include unlisted steps or modules (units), or may include other steps or modules (units) inherent to these processes, methods, products or devices. The words "connect", "connected", "coupled" and the like involved in this application are not limited to physical or mechanical connections, but may include electrical connections, whether directly or indirectly. The "multiple" involved in this application refers to two or more. "And / or" describes the association relationship of associated objects, indicating that there can be three relationships. For example, "A and / or B" can mean: A exists alone, A and B exist at the same time, and B exists alone. Usually, the character " / " indicates that the objects associated with each other are in an "or" relationship. The terms "first", "second", "third", etc. involved in this application are only used to distinguish similar objects and do not represent a specific ordering of the objects.

[0048] The method embodiment provided in this embodiment can be executed in a terminal, a computer or a similar computing device. For example, running on a terminal, Figure 1 FIG. 4 is a hardware structure diagram of a terminal of the method for triggering enhanced scanning of this embodiment. Figure 1 As shown, the terminal may include one or more ( Figure 1 The terminal may further include a transmission device 106 and an input / output device 108 for communication functions. A person skilled in the art will appreciate that Figure 1 The structure shown is only for illustration and does not limit the structure of the above terminal. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations shown.

[0049] The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as the computer program corresponding to the triggering method of enhanced scanning in this embodiment. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, that is, to implement the above method. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories may be connected to the terminal via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0050] The transmission device 106 is used to receive or send data via a network. The above network includes a wireless network provided by the communication provider of the terminal. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, referred to as NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 can be a radio frequency (Radio Frequency, referred to as RF) module, which is used to communicate with the Internet wirelessly.

[0051] In this embodiment, a method for triggering enhanced scanning is provided, which can be applied to enhanced scanning in different modalities such as CT and MRI, for example, using CT to perform dynamic enhanced scanning of the abdomen, enhanced scanning of the cervical blood vessels, etc. Figure 2 is a flow chart of the method for triggering enhanced scanning of this embodiment, such as Figure 2 As shown, the process includes the following steps:

[0052] Step S210, acquiring a bolus tracking image sequence corresponding to the target scanning area after contrast agent injection; the bolus tracking image sequence includes the current image;

[0053] Step S220, inputting the bolus tracking image sequence into a well-trained deep learning network, predicting the current image based on the contrast agent inflow information in the current image, and outputting a prediction result corresponding to the current image; the prediction result is used to indicate whether the current image is a trigger frame;

[0054] Step S230, when the current image is a trigger frame, trigger scanning of the target scanning area.

[0055] Specifically, after the contrast agent is injected into the body of the object to be scanned, a bolus tracking image of the target scanning area is acquired to obtain a corresponding bolus tracking image sequence, and the acquired bolus tracking image is monitored in real time. The bolus tracking image sequence includes at least a current image, and the current image may be a bolus tracking image acquired for the target scanning area at the current moment.

[0056] The current image is input into a well-trained deep learning network. In the deep learning network, the contrast agent inflow information in the current image is extracted, and the current image is predicted based on the contrast agent inflow information. The deep learning network can adopt different network combinations, such as a combination of a convolutional neural network and a recurrent network, a combination of a convolutional neural network and a transformer module with an attention mechanism, etc. The recurrent network can adopt a long short-term memory (Long Short-Term Memory, LSTM) network, etc. This embodiment does not specifically limit the form of the convolutional neural network.

[0057] Among them, the feature information of the current image is extracted in advance through the first network in the deep learning network to obtain the contrast agent inflow information in the current image, and the contrast agent inflow information can be the light and dark information of the target tissue in the current image. This is because in the bolus tracking image, the position where the contrast agent has not arrived is displayed as a darker state, the position where the contrast agent is about to arrive is displayed as a brighter state, and the position where the contrast agent arrives is highlighted. The contrast agent enters the blood circulation through intravenous injection. As time changes, the contrast agent will follow the blood flow of the aorta to be distributed throughout the body. When the contrast agent is about to flow to the target tissue in the target scanning area following the blood flow of the aorta, the display state of the target tissue in the bolus tracking image changes from dark to bright. Taking the enhanced scanning of the carotid blood vessels as an example, when the contrast agent is just injected into the body, the carotid artery in the collected bolus tracking image is displayed as a darker state, and the display state of the carotid artery changes from dark to bright over time, such as Figure 3 As shown in (a) in FIG. 1 ; when the contrast agent flows to the carotid artery, the carotid artery in the collected bolus tracking image becomes highlighted, as shown in FIG. Figure 3 As shown in (b) in .

[0058] Furthermore, the contrast agent inflow information in the current image is input into the second network in the deep learning network, and the current image is predicted based on the contrast agent inflow information, that is, whether the display state of the target tissue in the current image reaches the preset brightness state is judged according to the contrast agent inflow information, and whether the current image is a trigger frame is predicted according to the judgment result, and the corresponding prediction result is output, which is used to indicate whether the current image is a trigger frame, such as outputting a specific prediction value, using a prediction value of 0 to indicate that the current image is not a trigger frame, and a prediction value of 1 to indicate that the current image is a trigger frame. If it is detected that the display state of the target tissue in the current image reaches the preset brightness state, the prediction result is output to indicate that the current image is a trigger frame, triggering the scanning of the target scanning area; if it is detected that the display state of the target tissue in the current image does not reach the preset brightness state, the prediction result is output to indicate that the current image is not a trigger frame, and the bolus tracking image of the target scanning area at the next moment is used as the current image, and the bolus tracking image of the target scanning area is continuously monitored. It can be understood that the preset brightness state in this embodiment is usually a highlighted state or a near-highlight state to indicate that the contrast agent state in the target scanning area has reached a triggerable scanning state. The specific brightness can be reasonably set during the training process of the deep learning network according to the actual application situation.

[0059] For example, Figure 4 As shown, the first frame of bolus tracking image acquired by scanning is input into the deep learning network as the current image. If the prediction result of the first frame of bolus tracking image indicates that the first frame of bolus tracking image is not a trigger frame, the output prediction value y0 is 0, then the second frame of bolus tracking image is scanned and acquired, and the second frame of bolus tracking image is input into the deep learning network as the current image for prediction, and the input is sequentially input according to the acquisition timing until the trigger frame is predicted and output, for example, the prediction values ​​y1~y3 corresponding to the second to fourth frames of bolus tracking images are all 0, until the fifth frame of bolus tracking image is predicted to be a trigger frame, the output prediction value y4 is 1, then the scanning of the target scanning area is triggered, the scanning of the current sequence is stopped, and the scanning of the next diagnostic sequence is triggered.

[0060] It should be noted that the acquired bolus tracking image sequence may also be bolus tracking images of the target scanning area at multiple moments, and the image sequence includes a current image and multiple preceding images that are sequentially adjacent to the current image. The current image may be a bolus tracking image of the target scanning area at the current moment, and the preceding image is the bolus tracking image of the first n frames of the current image. For example, the first two bolus tracking images of the current image are taken as preceding images each time.

[0061] The current image and each previous image are input into a well-trained deep learning network. The first network of the deep learning network detects and obtains the changes in the contrast agent inflow information between each previous image and the current image. The changes in the contrast agent inflow information are input into the second network of the deep learning network. Based on the changes in the contrast agent inflow information, it is predicted whether the current image is a trigger frame, and the corresponding prediction result is output. The prediction result is used to indicate whether the current image is a trigger frame. If the current image is a trigger frame, the scanning of the target scanning area is triggered; if the current image is not a trigger frame, the bolus tracking image of the target scanning area at the next moment is used as the current image, and the bolus tracking image of the target scanning area is continuously monitored.

[0062] Afterwards, by triggering the scan of the target scanning area, the scanning device (such as CT or MRI) is controlled to collect images according to the preset parameters of the target scanning area, and the enhanced scanning image of the target scanning area is collected. The enhanced scanning image clearly shows the distribution of the contrast agent in the blood vessels and tissues, so as to obtain the best enhanced scanning image through precise triggering, while avoiding the risk of increased contrast dose due to repeated scanning caused by inaccurate triggering scanning time points. Among them, the original scanning image of the target scanning area can be compared with the enhanced scanning image, or the original scanning image of the target scanning area can be subtracted from the enhanced scanning image to highlight the area where the contrast agent is distributed, improve the visibility of the vascular structure and lesions, and further help identify the vascular structure and lesion area.

[0063] In the existing enhanced scanning method, a baseline image is obtained after contrast agent is injected, and the corresponding region of interest is automatically obtained based on the baseline image, and the time when the contrast agent reaches the region of interest is predicted, and the scan is triggered based on the predicted time. However, the above method cannot ensure that the appropriate region of interest is obtained when there is a deviation in the baseline image, resulting in inaccurate time to trigger the enhanced scan.

[0064] Compared with the prior art, the present application obtains a bolus tracking image sequence corresponding to the target scanning area after contrast agent injection; the bolus tracking image sequence includes the current image; the bolus tracking image sequence is input into a well-trained deep learning network, the current image is predicted based on the contrast agent inflow information in the current image, and the prediction result corresponding to the current image is output; the prediction result is used to indicate whether the current image is a trigger frame; when the current image is a trigger frame, the scanning of the target scanning area is triggered. Based on this, the contrast agent state of the input frame is accurately identified through the deep learning network, and based on the contrast agent inflow information in the target scanning area, it is predicted whether the current image is a trigger frame, and the scanning is triggered when the current image reaches the corresponding state of the trigger frame, which solves the problem of inaccurate time of triggering enhanced scanning and realizes accurate triggering of enhanced scanning.

[0065] In some of the embodiments, when the bolus tracking image sequence further includes a plurality of preceding images that are sequentially adjacent to the current image, after acquiring the bolus tracking image sequence corresponding to the target scanning area after the contrast agent is injected, the triggering method for enhanced scanning further includes the following steps:

[0066] The bolus tracking image sequence is input into a well-trained deep learning network, and the current image is predicted based on the change of contrast agent inflow information between the current image and each previous image, and the prediction result corresponding to the current image is output; the prediction result is used to indicate whether the current image is a trigger frame;

[0067] When the current image is a trigger frame, scanning of the target scanning area is triggered.

[0068] Specifically, bolus tracking images of the target scanning area at different times are collected to obtain a bolus tracking image sequence, where the image sequence includes a current image and a plurality of preceding images that are time-sequentially adjacent to the current image. The current image may be a bolus tracking image of the target scanning area at the current moment, and the preceding image is a bolus tracking image of the first n frames of the current image. For example, the first two bolus tracking images of the current image are taken as preceding images each time.

[0069] The current image and each previous image are input into a well-trained deep learning network, and the change of contrast agent inflow information between each previous image and the current image is detected through the first network of the deep learning network. It should be noted that since the position where the contrast agent has not arrived is displayed as a darker state in the bolus tracking image, the position where the contrast agent is about to arrive is displayed as a brighter state, and the position where the contrast agent arrives will be highlighted, the change of the contrast agent inflow information can be the change of brightness of the target tissue in the current image.

[0070] Furthermore, the change of contrast agent inflow information is input into the second network in the deep learning network, and based on the change of contrast agent inflow information, it is predicted whether the current image is a trigger frame, that is, whether the display state of the target tissue in the current image reaches a preset brightness state according to the change of contrast agent inflow information, and whether the current image is a trigger frame is predicted according to the judgment result, and the corresponding prediction result is output, and the prediction result is used to indicate whether the current image is a trigger frame. If it is detected that the display state of the target tissue in the current image reaches a preset brightness state, the prediction result is output to indicate that the current image is a trigger frame, triggering the scanning of the target scanning area; if it is detected that the display state of the target tissue in the current image does not reach the preset brightness state, the prediction result is output to indicate that the current image is not a trigger frame, and the bolus tracking image of the target scanning area at the next moment is used as the current image, and the bolus tracking image of the target scanning area is continuously monitored.

[0071] In this embodiment, the deep learning network can adopt different network combinations, such as a combination of a convolutional neural network and a recurrent network, etc. The recurrent network can adopt a long short-term memory network, etc., and the form of the convolutional neural network is not specifically limited.

[0072] Through this embodiment, the bolus tracking image sequence is input into a well-trained deep learning network, the current image is predicted based on the change of contrast agent inflow information between the current image and each previous image, and the prediction result corresponding to the current image is output. When the prediction result indicates that the current image is a trigger frame, the scanning of the target scanning area is triggered, so that based on the contrast agent inflow change information in the target scanning area, it is possible to predict whether the current image is a trigger frame, and trigger the scan when the current image reaches the corresponding state of the trigger frame, so as to accurately determine whether the current image is a trigger frame.

[0073] In some embodiments, predicting the current image based on contrast agent inflow information in the current image and outputting a prediction result corresponding to the current image includes the following steps:

[0074] According to the contrast agent inflow information in the current image, judging whether the contrast agent state in the target scanning area reaches a preset state;

[0075] When the state of the contrast agent in the target scanning area reaches a preset state, the current image is determined to be a trigger frame.

[0076] Specifically, the contrast agent inflow information in the current image is obtained through the first network in the deep learning network, and whether the display state of the target tissue in the current image reaches the preset brightness state is judged according to the contrast agent inflow information, and whether the current image is a trigger frame is predicted according to the judgment result, and the corresponding prediction result is output, which is used to indicate whether the current image is a trigger frame.

[0077] If the display state of the target tissue in the current image reaches the preset brightness state, that is, the contrast agent state in the target scanning area has reached a triggerable scanning state, then the prediction result is output to indicate that the current image is a trigger frame; if the display state of the target tissue in the current image does not reach the preset brightness state, that is, the contrast agent state in the target scanning area has not reached a triggerable scanning state, then the prediction result is output to indicate that the current image is not a trigger frame. The contrast agent state in the target scanning area refers to whether the contrast agent has reached the target tissue in the target scanning area, whether the concentration of the contrast agent in the target tissue blood has reached an effective level, etc., to ensure that the target tissue can be clearly displayed.

[0078] It should be noted that the state of the contrast agent in the target scanning area can be determined according to the display state of the target tissue. In the bolus tracking image acquired by scanning, the position where the contrast agent has not arrived is displayed as a darker state, the position where the contrast agent is about to arrive is displayed as a brighter state, and the position where the contrast agent arrives will be highlighted. The contrast agent enters the blood circulation through intravenous injection. As time changes, the contrast agent will follow the blood flow of the aorta to be distributed throughout the body. When the contrast agent is about to flow to the target tissue in the target scanning area following the blood flow of the aorta, the display state of the target tissue in the bolus tracking image changes from dark to bright, and will eventually reach a highlighted state.

[0079] Through this embodiment, based on the contrast agent inflow information in the current image, it is determined whether the contrast agent area on the target scanning area reaches the preset state. When the contrast agent state on the target scanning area reaches the preset state, the current image is determined to be a trigger frame, thereby improving the accuracy of the triggered scan.

[0080] In some of the embodiments, when the output of the deep learning network includes the credibility of the prediction result, after outputting the prediction result corresponding to the current image, the triggering method of the enhanced scan further includes the following steps:

[0081] When the current image is a trigger frame and the credibility reaches a preset threshold, scanning of the target scanning area is triggered.

[0082] Specifically, after the combination of the current image and each previous image is input into the deep learning network, or the current image is input into the deep learning network, the output of the deep learning network includes the prediction result of the current image, that is, whether the current image is a trigger frame, and the credibility of the prediction result. If the credibility value of the prediction result is high, it indicates that the prediction result is credible, and if the credibility value of the prediction result is low, it indicates that the prediction result is unreliable. Among them, in the case where the output of the deep learning network includes the credibility of the prediction result, a corresponding node or subnetwork can be set in the output layer of the deep learning network to analyze the credibility of the prediction result, and then train the deep learning network to achieve multi-task output.

[0083] In this embodiment, the credibility of the prediction result is related to whether effective contrast agent inflow information can be obtained from the current image, for example, detecting whether the current image itself contains contrast agent inflow information, whether the quality of the current image affects the accurate extraction of contrast agent inflow information, etc.

[0084] For example, Figure 5As shown, a combination of the first network and the second network is used as a deep learning network, the current image and each previous image are input into the deep learning network, and the current image is predicted based on the change of contrast agent inflow information between the current image and each previous image to obtain a prediction result and a credibility corresponding to the prediction result. The prediction results of each group of images are y0, y1, y2, y3, y4, ..., and correspondingly, the credibility corresponding to each group of images is p0, p1, p2, p3, p4, ..., wherein the prediction result can be 0 or 1, and the value 0 and the value 1 respectively indicate that the current image is not a trigger frame and the current image is a trigger frame.

[0085] Furthermore, when the deep learning network outputs the prediction result and the credibility of the prediction result, if the current image is predicted to be a trigger frame and the credibility reaches the preset threshold, the scan of the target scanning area is triggered; if the current image is predicted to be a trigger frame and the credibility does not reach the preset threshold, it indicates that the prediction result is abnormal. At this time, a pop-up window prompt, voice prompt, etc. are used to inform the operator of the current abnormal situation. In the case of abnormal prediction results, the enhanced scan is automatically triggered based on the prediction result, or it is configured to determine the trigger time according to the system information to implement the subsequent enhanced scan, including determining the trigger time based on the type of syringe currently used, determining the trigger time based on the start scan time corresponding to the bolus tracking image sequence, etc. When the operator receives the abnormal prompt information, the trigger mode of the enhanced scan can be adjusted in real time according to the current scanning requirements.

[0086] It should be noted that, when it is predicted that the current image is not a trigger frame, if the credibility reaches the preset threshold, the bolus tracking image of the target scanning area will continue to be monitored. When it is predicted that the current image is not a trigger frame and the credibility does not reach the preset threshold, the operator may be prompted to confirm the status of the contrast agent in the current image to confirm whether to trigger the scan.

[0087] Through this embodiment, the output of the deep learning network includes the prediction results and the credibility of the prediction results, so that when the current image is a trigger frame and the credibility reaches a preset threshold, the scan of the target scanning area can be triggered, thereby avoiding abnormal conditions in image detection that lead to inaccurate scanning trigger time and improving the accuracy and stability of the triggered scan.

[0088] In some embodiments, outputting the credibility of the prediction result through the deep learning network includes the following steps:

[0089] Through the deep learning network, determine whether the current image contains contrast agent inflow information;

[0090] Determine the credibility of the prediction results based on the judgment results; and / or,

[0091] Perform quality detection on the current image through deep learning network;

[0092] Based on the test results, determine the credibility of the prediction results.

[0093] In this embodiment, the credibility of the prediction result is evaluated based on whether effective contrast agent inflow information can be obtained from the current image. A higher credibility value output indicates that the prediction result is credible, and a lower credibility value indicates that the prediction result is unreliable. The evaluation method includes detecting whether the current image contains contrast agent inflow information, whether the current image quality affects the accuracy of the prediction result, etc. A variety of evaluation methods can be used to calculate the credibility of the prediction result. The credibility is usually a positive value. When it is detected that the current image does not contain contrast agent inflow information, the credibility is set to a negative value to distinguish other abnormal situations.

[0094] Specifically, through the deep learning network, it is determined whether the current image contains contrast agent inflow information, and the credibility of the prediction result is determined according to the judgment result. If it is detected that the current image contains contrast agent inflow information, it indicates that the credibility of the prediction result is high, otherwise, it indicates that the credibility of the prediction result is low. In the case of predicting that the current image is a trigger frame, the credibility is compared with the preset threshold. If the credibility does not reach the preset threshold, it indicates that the prediction result is abnormal.

[0095] In addition, since the movement of the object to be scanned or the generation of metal artifacts and other factors will affect the quality of the bolus tracking image, the quality of the current image can be detected through the deep learning network, and the quality of the current image can be determined according to the preset image quality evaluation standard. The image quality evaluation standard corresponds to the influencing factors of the above-mentioned image quality, and the credibility of the prediction result is determined according to the current image quality. The current image quality is positively correlated with the credibility of the prediction result.

[0096] Through this embodiment, a deep learning network is used to determine whether the current image contains contrast agent inflow information, and the credibility of the prediction result is determined based on the judgment result; and / or, the current image is quality tested through a deep learning network, and the credibility of the prediction result is determined based on the test result, which helps to determine whether an abnormal situation occurs and improve the accuracy of detecting abnormal situations.

[0097] In some of the embodiments, when the output of the deep learning network includes the credibility of the prediction result, after outputting the prediction result corresponding to the current image, the triggering method of the enhanced scan further includes the following steps:

[0098] When the current image is a trigger frame and the credibility does not reach a preset threshold, the trigger time is determined based on the injection time of the contrast agent or the start scanning time corresponding to the bolus tracking image sequence;

[0099] According to the trigger time, the scanning of the target scanning area is triggered.

[0100] Specifically, when the prediction result is that the current image is a trigger frame and the credibility does not reach the preset threshold, it indicates that the prediction result is abnormal. While prompting the operator that the current prediction result is abnormal, the scan is triggered according to the preset configuration. The preset configuration includes automatically triggering an enhanced scan based on the prediction result, or determining the trigger time according to system information to implement subsequent enhanced scans, etc. The operator can also adjust the triggering method of the enhanced scan in real time according to the current scanning requirements.

[0101] Among them, in the case where the system information includes the type of syringe currently used and the injection time of the contrast agent, since different syringes have different injection speeds and control accuracy, the time for the contrast agent to reach a triggerable state in the body is different when using a manual syringe, a high-pressure syringe, etc. According to the type of syringe and the injection time of the contrast agent, the time when the contrast agent reaches a triggerable state in the target scanning area is calculated as the trigger time; in the case where the system information includes the start scan time corresponding to the bolus tracking image sequence, based on historical experience information, a time interval between the injection of the contrast agent into the body and the contrast reaching a triggerable state is preset, and then the trigger time is calculated according to the start scan time and the preset time interval. The start scan time usually refers to the scan time for obtaining the first bolus tracking image in the bolus tracking image sequence.

[0102] It should be noted that since the prediction time of enhanced scanning is short, when the current image is predicted to be a trigger frame but the credibility does not reach the preset threshold, additional time is usually not available for the control system to reacquire the bolus tracking image for prediction. Therefore, before triggering the enhanced scan, it is necessary to identify in advance whether the prediction result is abnormal, so that the system or operator can correct the current enhanced scan trigger.

[0103] Through this embodiment, when the current image is a trigger frame and the credibility does not reach the preset threshold, the trigger time is determined based on the injection time of the contrast agent or the start scan time corresponding to the bolus tracking image sequence, thereby avoiding the inability to achieve accurate scan triggering when the credibility does not reach the preset threshold, thereby improving the accuracy and stability of the scan triggering.

[0104] In some of the embodiments, before predicting the current image based on the change of contrast agent inflow information between the current image and each previous image by using a fully trained deep learning network, the following steps are also included:

[0105] Acquire a sample image set; the sample image set includes a positive sample image and a negative sample image, the positive sample image is a bolus tracking image in which the contrast agent inflow state reaches the trigger standard, and the negative sample image is a bolus tracking image in which the contrast agent inflow state does not reach the trigger standard;

[0106] The deep learning network is trained based on the sample image set to obtain a fully trained deep learning network.

[0107] Specifically, a sample image set for training a deep learning network is obtained, the sample image set including positive sample images and negative sample images, wherein the positive sample images are bolus tracking images in which the contrast agent inflow state reaches the trigger standard, and the negative sample images are bolus tracking images in which the contrast agent inflow state does not reach the trigger standard. The deep learning network can use different network combinations, such as a combination of a convolutional neural network and a recurrent network, to predict whether the current image is a trigger frame.

[0108] It should be noted that, when the output of the deep learning network includes the credibility of the prediction results, corresponding nodes or sub-networks can be set in the output layer of the deep learning network to analyze the credibility of the prediction results, and the deep learning network can be trained to achieve multi-task output.

[0109] Furthermore, the deep learning network is trained based on the sample image set, so that the deep learning network can learn and capture the contrast agent inflow changes between different images, thereby learning whether the corresponding state of the trigger frame is reached.

[0110] Among them, based on different network inputs, that is, inputting the combination of the current image and each previous image into the deep learning network, or inputting the current image into the deep learning network, the deep learning network can be trained to make a judgment based on the degree of change of the contrast agent state between the current image and the previous image. If the degree of change of the contrast agent state in the target scanning area of ​​the current image compared with the previous images reaches a preset standard, the current image is judged to be a trigger frame, and / or, the deep learning network is trained to make a judgment based on the contrast agent state in the current image. If it is detected that the contrast agent state in the target scanning area of ​​the current image reaches a preset state, the current image is judged to be a trigger frame.

[0111] Through this embodiment, a sample image set including positive sample images and negative sample images is obtained, the positive sample images are bolus tracking images in which the contrast agent inflow state reaches the trigger standard, and the negative sample images are bolus tracking images in which the contrast agent inflow state does not reach the trigger standard. Based on the sample image set, a deep learning network is trained to obtain a fully trained deep learning network, so that the trigger frame prediction can be realized through the deep learning network to ensure the accuracy of the prediction.

[0112] The present embodiment is described and illustrated below through preferred embodiments.

[0113] Figure 6 Flow chart of the method for triggering enhanced scanning of the preferred embodiment. Figure 6 As shown, the triggering method of the enhanced scan includes the following steps:

[0114] Step S610, acquiring a sample image set; the sample image set includes positive sample images and negative sample images, the positive sample images are bolus tracking images in which the contrast agent inflow state reaches the trigger standard, and the negative sample images are bolus tracking images in which the contrast agent inflow state does not reach the trigger standard;

[0115] Step S620, training the deep learning network based on the sample image set to obtain a fully trained deep learning network;

[0116] Step S630, acquiring a bolus tracking image sequence corresponding to the target scanning area after contrast agent injection; the bolus tracking image sequence includes a current image and a plurality of preceding images that are sequentially adjacent to the current image;

[0117] Step S640, inputting the bolus tracking image sequence into a well-trained deep learning network, predicting the current image based on the change of contrast agent inflow information between the current image and each previous image, and outputting the prediction result corresponding to the current image and the credibility of the prediction result; the prediction result is used to indicate whether the current image is a trigger frame;

[0118] Step S650, when the current image is a trigger frame and the credibility reaches a preset threshold, triggering scanning of the target scanning area;

[0119] Step S660, when the current image is a trigger frame and the credibility does not reach a preset threshold, the trigger time is determined based on the injection time of the contrast agent or the start scanning time corresponding to the bolus tracking image sequence;

[0120] Step S670, triggering scanning of the target scanning area according to the trigger time.

[0121] Through this embodiment, a sample image set is obtained; the sample image set includes positive sample images and negative sample images, the positive sample images are bolus tracking images in which the contrast agent inflow state reaches the trigger standard, and the negative sample images are bolus tracking images in which the contrast agent inflow state does not reach the trigger standard, and the deep learning network is trained based on the sample image set to obtain a fully trained deep learning network.

[0122] Furthermore, a bolus tracking image sequence corresponding to the target scanning area after contrast agent injection is obtained, the bolus tracking image sequence includes the current image and multiple previous images that are adjacent to the current image in time sequence, the bolus tracking image sequence is input into a well-trained deep learning network, the current image is predicted based on the change of contrast agent inflow information between the current image and each previous image, and the prediction result corresponding to the current image and the credibility of the prediction result are output, and the prediction result is used to indicate whether the current image is a trigger frame. When the current image is a trigger frame and the credibility reaches a preset threshold, the scanning of the target scanning area is triggered, and when the current image is a trigger frame and the credibility does not reach the preset threshold, the trigger time is determined based on the injection time of the contrast agent or the start scanning time corresponding to the bolus tracking image sequence, and the scanning of the target scanning area is triggered according to the trigger time, which solves the problem of inaccurate time of triggering enhanced scanning and realizes accurate triggering of enhanced scanning.

[0123] It should be noted that the steps shown in the above process or the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0124] In this embodiment, a trigger device for enhanced scanning is also provided, which is used to implement the above-mentioned embodiments and preferred implementation modes, and will not be repeated hereafter. The terms "module", "unit", "subunit", etc. used below can implement a combination of software and / or hardware for a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0125] Figure 7 is a structural block diagram of the trigger device for enhanced scanning of this embodiment, such as Figure 7 As shown, the device includes: an acquisition module 10, a prediction module 20 and a trigger module 30;

[0126] An acquisition module 10 is used to acquire a bolus tracking image sequence corresponding to the target scanning area after the contrast agent is injected; the bolus tracking image sequence includes the current image;

[0127] The prediction module 20 is used to input the bolus tracking image sequence into the well-trained deep learning network, predict the current image based on the contrast agent inflow information in the current image, and output the prediction result corresponding to the current image; the prediction result is used to indicate whether the current image is a trigger frame;

[0128] The trigger module 30 is used to trigger scanning of the target scanning area when the current image is a trigger frame.

[0129] Through the device provided in this embodiment, a bolus tracking image sequence corresponding to the target scanning area after injection of contrast agent is obtained; the bolus tracking image sequence includes the current image; the bolus tracking image sequence is input into a well-trained deep learning network, the current image is predicted based on the contrast agent inflow information in the current image, and the prediction result corresponding to the current image is output; the prediction result is used to indicate whether the current image is a trigger frame; when the current image is a trigger frame, scanning of the target scanning area is triggered, thereby solving the problem of inaccurate time for triggering enhanced scanning and realizing accurate triggering of enhanced scanning.

[0130] In some of the embodiments, the prediction module 30 is further used to input the bolus tracking image sequence into a well-trained deep learning network, predict the current image based on the change in contrast agent inflow information between the current image and each previous image, and output a prediction result corresponding to the current image; the prediction result is used to indicate whether the current image is a trigger frame; when the current image is a trigger frame, scanning of the target scanning area is triggered.

[0131] In some of the embodiments, the prediction module 20 is also used to determine whether the contrast agent state in the target scanning area reaches a preset state based on the contrast agent inflow information in the current image; when the contrast agent state in the target scanning area reaches the preset state, the current image is determined to be a trigger frame.

[0132] In some of the embodiments, the trigger module 30 is further configured to trigger scanning of the target scanning area when the current image is a trigger frame and the credibility reaches a preset threshold.

[0133] In some of these embodiments, Figure 7 On the basis of this, the device also includes a detection module, which is used to determine whether the current image contains contrast agent inflow information through a deep learning network; determine the credibility of the prediction result based on the judgment result; and / or, perform quality detection on the current image through a deep learning network; determine the credibility of the prediction result based on the detection result.

[0134] In some of the embodiments, the trigger module 30 is also used to determine the trigger time based on the injection time of the contrast agent or the start scanning time corresponding to the bolus tracking image sequence when the current image is a trigger frame and the credibility does not reach a preset threshold; and trigger the scanning of the target scanning area according to the trigger time.

[0135] In some of these embodiments, Figure 7On the basis of, the device also includes a training module for acquiring a sample image set; the sample image set includes positive sample images and negative sample images, the positive sample images are bolus tracking images whose contrast agent inflow status reaches the trigger standard, and the negative sample images are bolus tracking images whose contrast agent inflow status does not reach the trigger standard; the deep learning network is trained based on the sample image set to obtain a fully trained deep learning network.

[0136] It should be noted that the above modules can be functional modules or program modules, and can be implemented by software or hardware. For modules implemented by hardware, the above modules can be located in the same processor; or the above modules can be located in different processors in any combination.

[0137] In this embodiment, a computer device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0138] Optionally, the computer device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor, and the input / output device is connected to the processor.

[0139] Optionally, in this embodiment, the processor may be configured to perform the following steps through a computer program:

[0140] S1, acquiring a bolus tracking image sequence corresponding to a target scanning area after contrast agent injection; the bolus tracking image sequence includes a current image and a plurality of preceding images that are sequentially adjacent to the current image;

[0141] S2, inputting the bolus tracking image sequence into a well-trained deep learning network, predicting the current image based on the contrast agent inflow information in the current image, and outputting the prediction result corresponding to the current image; the prediction result is used to indicate whether the current image is a trigger frame;

[0142] S3, when the current image is a trigger frame, trigger scanning of the target scanning area.

[0143] It should be noted that the specific examples in this embodiment can refer to the examples described in the above embodiments and optional implementation modes, and will not be repeated in this embodiment.

[0144] In addition, in combination with the enhanced scanning triggering method provided in the above embodiments, a storage medium may be provided in this embodiment to implement the method. The storage medium stores a computer program, and when the computer program is executed by a processor, any of the enhanced scanning triggering methods in the above embodiments is implemented.

[0145] It should be understood that the specific embodiments described herein are only used to explain the application, rather than to limit it. Based on the embodiments provided in this application, all other embodiments obtained by ordinary technicians in this field without creative work are within the protection scope of this application.

[0146] Obviously, the drawings are only some examples or embodiments of the present application. For ordinary technicians in the field, the present application can also be applied to other similar situations based on these drawings without creative work. In addition, it is understandable that although the work done in this development process may be complicated and lengthy, for ordinary technicians in the field, certain changes in design, manufacturing or production based on the technical content disclosed in this application are only conventional technical means and should not be regarded as insufficient content disclosed in this application.

[0147] The term "embodiment" in this application refers to a specific feature, structure or characteristic described in conjunction with the embodiment that can be included in at least one embodiment of the present application. The appearance of this phrase in various locations in the specification does not necessarily mean the same embodiment, nor does it mean that it is mutually exclusive with other embodiments and is independent or optional. It is clearly or implicitly understood by those of ordinary skill in the art that the embodiments described in this application can be combined with other embodiments without conflict.

[0148] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of patent protection. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the scope of protection of the present application. Therefore, the scope of protection of the present application shall be subject to the attached claims.

Claims

1. A method for triggering enhanced scanning, characterized in that: The method comprises: Acquire a bolus tracking image sequence corresponding to the target scanning area after contrast agent injection; the bolus tracking image sequence includes a current image; Inputting the bolus tracking image sequence into a well-trained deep learning network, predicting the current image based on contrast agent inflow information in the current image, and outputting a prediction result corresponding to the current image; the prediction result is used to indicate whether the current image is a trigger frame; When the current image is a trigger frame, scanning of the target scanning area is triggered.

2. The method for triggering enhanced scanning according to claim 1, characterized in that: In the case where the bolus tracking image sequence further includes a plurality of preceding images that are sequentially adjacent to the current image, after acquiring the bolus tracking image sequence corresponding to the target scanning area after injection of contrast agent, the method further includes: Inputting the bolus tracking image sequence into the well-trained deep learning network, predicting the current image based on the change of contrast agent inflow information between the current image and each of the previous images, and outputting the prediction result corresponding to the current image; the prediction result is used to indicate whether the current image is a trigger frame; When the current image is a trigger frame, scanning of the target scanning area is triggered.

3. The method for triggering enhanced scanning according to claim 1, characterized in that: The predicting the current image based on the contrast agent inflow information in the current image and outputting the prediction result corresponding to the current image includes: According to the contrast agent inflow information in the current image, determining whether the contrast agent state in the target scanning area reaches a preset state; When the state of the contrast agent in the target scanning area reaches a preset state, the current image is determined to be a trigger frame.

4. The method for triggering enhanced scanning according to any one of claims 1 to 3, characterized in that: In the case where the output of the deep learning network includes the credibility of the prediction result, after outputting the prediction result corresponding to the current image, the method further includes: When the current image is a trigger frame and the credibility reaches a preset threshold, scanning of the target scanning area is triggered.

5. The method for triggering enhanced scanning according to claim 4, characterized in that: Outputting the credibility of the prediction result through the deep learning network includes: Determining, by means of the deep learning network, whether the current image contains contrast agent inflow information; Determining the credibility of the prediction result based on the judgment result; and / or, Performing quality detection on the current image through the deep learning network; According to the detection result, the credibility of the prediction result is determined.

6. The method for triggering enhanced scanning according to claim 4, characterized in that: In the case where the output of the deep learning network includes the credibility of the prediction result, after outputting the prediction result corresponding to the current image, the method further includes: When the current image is a trigger frame and the credibility does not reach a preset threshold, determining a trigger time based on the injection time of the contrast agent or the start scanning time corresponding to the bolus tracking image sequence; According to the trigger time, a scan of the target scanning area is triggered.

7. The method for triggering enhanced scanning according to claim 1, characterized in that: Before inputting the bolus tracking image sequence into a well-trained deep learning network and predicting the current image based on the contrast agent inflow information in the current image, the method further includes: Acquire a sample image set; the sample image set includes positive sample images and negative sample images, the positive sample images are bolus tracking images in which the contrast agent inflow state reaches the trigger standard, and the negative sample images are bolus tracking images in which the contrast agent inflow state does not reach the trigger standard; The deep learning network is trained based on the sample image set to obtain the fully trained deep learning network.

8. A trigger device for enhanced scanning, characterized in that: The device comprises: an acquisition module, a prediction module and a trigger module; The acquisition module is used to acquire a bolus tracking image sequence corresponding to the target scanning area after the contrast agent is injected; the bolus tracking image sequence includes the current image; The prediction module is used to input the bolus tracking image sequence into a well-trained deep learning network, predict the current image based on the contrast agent inflow information in the current image, and output a prediction result corresponding to the current image; the prediction result is used to indicate whether the current image is a trigger frame; The trigger module is used to trigger scanning of the target scanning area when the current image is a trigger frame.

9. A computer device comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to run the computer program to execute the steps of the method for triggering enhanced scanning according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for triggering enhanced scanning according to any one of claims 1 to 7 are implemented.

Images