Method, device, equipment and medium for predicting success rate of cto intervention treatment

Abstract

The application relates to a CTO interventional therapy success rate prediction method, device, equipment and medium, the method comprising the following steps: acquiring a coronary CTA image of a diseased heart; determining a blood vessel segmentation mask of the coronary CTA image by using a pre-trained blood vessel segmentation model; receiving a user-input PCI pre-intervention position, and cutting the coronary CTA image based on the PCI pre-intervention position to obtain a cut image; inputting the blood vessel segmentation mask and the cut image into a pre-trained feature extraction model to obtain image features corresponding to the coronary CTA image; extracting imageomics features corresponding to the blood vessel segmentation mask and the cut image; and inputting the image features and the imageomics features into a pre-trained success rate prediction model to obtain a predicted success rate of a PCI operation on the diseased heart. The application can automatically, quickly and accurately predict the success rate of the PCI operation only by inputting the PCI pre-intervention position.

Description

Technical Field

[0001] This application relates to the field of medical image processing technology, and in particular to a method, device, equipment and medium for predicting the success rate of CTO interventional treatment. Background Technology

[0002] Chronic total occlusion (CTO) refers to coronary artery obstruction with a positive TIMI flow grade of 0 and an occlusion duration of more than three months. CTO remains a significant threat to human cardiovascular health, accounting for approximately 20% of coronary heart disease cases. CTO is currently a key focus and challenge in the field of interventional coronary artery disease. Recent data shows that the average success rate of interventional treatment for CTO is only around 75%. More accurate assessments of the success rate, risks, and surgical difficulty of interventional treatment before the procedure can help physicians make more precise judgments and improve the success rate. However, existing scoring models for interventional treatment success rates have certain limitations and fail to provide sufficiently accurate predictive results.

[0003] Currently, methods for predicting the success rate of interventional treatment for chronic total occlusion (CTO) coronary artery disease generally combine scoring models with manual scoring. These scoring models typically break down the model into several scoring variables, such as 1 point for blunt occlusion, 1 point for calcification, 1 point for tortuosity, and 1 point for occlusion length ≥20mm, etc., and then predict the success rate of interventional treatment based on the comprehensive score. This method requires physicians to score each indicator by observing the patient's CTA images, thus heavily relying on physician experience and being extremely time-consuming. Furthermore, even the widely accepted CT-RECTOR scoring model only achieves an average accuracy of around 70% in predicting interventional treatment success rates. To improve the efficiency and accuracy of predicting the success rate of interventional treatment for CTO and to free physicians from tedious image interpretation, there is an urgent need to develop an automated method for predicting the success rate of interventional treatment for CTO entirely based on CTA medical images. Summary of the Invention

[0004] In order to solve the above-mentioned technical problems, or at least partially solve the above-mentioned technical problems, this application provides a method, device, equipment and medium for predicting the success rate of CTO interventional treatment.

[0005] According to a first aspect of this application, a method for predicting the success rate of CTO interventional therapy is provided, comprising:

[0006] Acquire coronary CTA images;

[0007] Using a pre-trained vessel segmentation model, a vessel segmentation mask for the coronary CTA image is determined;

[0008] Receive the PCI pre-intervention location input by the user, and crop the coronary CTA image based on the PCI pre-intervention location to obtain a cropped image;

[0009] The blood vessel segmentation mask and the cropped image are input into a pre-trained feature extraction model to obtain the image features corresponding to the coronary CTA image;

[0010] Extract the radiomics features corresponding to the blood vessel segmentation mask and the cropped image;

[0011] The image features and the radiomics features are input into a pre-trained success rate prediction model to obtain the predicted success rate of PCI surgery.

[0012] According to a second aspect of this application, a device for predicting the success rate of CTO interventional therapy is provided, comprising:

[0013] The first acquisition module is used to acquire coronary CTA images;

[0014] A determination module is used to determine the vascular segmentation mask of the coronary CTA image using a pre-trained vascular segmentation model;

[0015] The cropping module is used to receive the PCI pre-intervention location input by the user and crop the coronary CTA image based on the PCI pre-intervention location to obtain a cropped image.

[0016] The second acquisition module is used to input the blood vessel segmentation mask and the cropped image into a pre-trained feature extraction model to obtain the image features corresponding to the coronary CTA image.

[0017] The extraction module is used to extract the radiomics features corresponding to the blood vessel segmentation mask and the cropped image;

[0018] The third acquisition module is used to input the image features and the radiomics features into a pre-trained success rate prediction model to obtain the predicted success rate of PCI surgery.

[0019] According to a third aspect of this application, an electronic device is provided, comprising: a processor, the processor being configured to execute a computer program stored in a memory, the computer program, when executed by the processor, implementing the method for predicting the success rate of CTO interventional therapy as described in the first aspect.

[0020] According to a fourth aspect of this application, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the method for predicting the success rate of CTO interventional therapy as described in the first aspect.

[0021] According to a fifth aspect of this application, a computer program product is provided that, when the computer program product is run on a computer, causes the computer to perform the method for predicting the success rate of CTO interventional therapy as described in the first aspect.

[0022] The technical solution provided in this application has the following advantages compared with the prior art:

[0023] By acquiring coronary CTA images and utilizing a pre-trained vessel segmentation model, a vessel segmentation mask is determined for the coronary CTA images. The system receives the user-inputted pre-intervention location for PCI and crops the coronary CTA image based on this location. The cropped image is then input into a pre-trained feature extraction model to obtain the corresponding image features of the coronary CTA image and to extract the corresponding radiomics features. These image features and radiomics features are then input into a pre-trained success rate prediction model to predict the success rate of PCI surgery. Using this technical solution, the success rate of PCI surgery can be automatically, quickly, and accurately predicted simply by the user inputting the pre-intervention location, eliminating the need for users to score various evaluation indicators and avoiding tedious image reading work. Attached Figure Description

[0024] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0025] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0026] Figure 1 A flowchart illustrating a method for predicting the success rate of CTO interventional therapy according to an embodiment of this application;

[0027] Figure 2 This is a schematic diagram of a device for predicting the success rate of CTO interventional therapy according to an embodiment of this application. Detailed Implementation

[0028] Embodiments of this application will now be described in more detail with reference to the accompanying drawings. While some embodiments of this application are shown in the drawings, it should be understood that this application can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this application. It should be understood that the drawings and embodiments of this application are for illustrative purposes only and are not intended to limit the scope of protection of this application.

[0029] It should be understood that the steps described in the method embodiments of this application may be performed in different orders and / or in parallel. Furthermore, the method embodiments may include additional steps and / or omit the steps shown. The scope of this application is not limited in this respect.

[0030] The term "comprising" and its variations as used herein are open-ended, meaning "including but not limited to". The term "based on" means "at least partially based on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". Definitions of other terms will be given in the following description. It should be noted that the concepts of "first", "second", etc., mentioned in this application are used only to distinguish different devices, modules, or units, and are not intended to limit the order of functions performed by these devices, modules, or units or their interdependencies.

[0031] It should be noted that the terms "a" and "a plurality of" used in this application are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0032] Before explaining the embodiments of the method for predicting the success rate of CTO interventional therapy provided in this application, the terms involved in this application and their roles in this application are explained as follows:

[0033] CTA, or Computed Tomography (CT) angiography, is a crucial part of clinical CT applications. Because of the poor natural contrast between blood vessels and their background soft tissues, conventional CT scans often fail to visualize blood vessels. During CTA examinations, a contrast agent is introduced to alter the image contrast between blood vessels and background tissues, thereby highlighting the vessels. CTA is widely used in the visualization of blood vessels in the head and neck, coronary arteries, pulmonary arteries and veins, thoracic aorta, abdominal aorta, and lower extremity arteries and veins.

[0034] For coronary CTA, the heart rate should ideally be kept below 70 bpm. However, with increasingly advanced equipment, the requirements for the heart rate have become less stringent, and even a heart rate greater than 70 bpm can still produce a good coronary angiogram.

[0035] PCI, or Percutaneous Coronary Intervention, is a treatment method that uses cardiac catheterization to open narrowed or even blocked coronary arteries, thereby improving myocardial blood flow.

[0036] Existing scoring models combined with manual scoring for predicting PCI treatment success rates suffer from high reliance on physician experience, time-consuming processes, and low accuracy. This disclosure provides an automated success rate prediction scheme for percutaneous coronary intervention (PCI) of chronic total occlusion (CTO) lesions based on deep learning and machine learning technologies. This scheme can be used to directly assess and predict the risk of failure in CTO percutaneous coronary intervention, offering advantages such as low time consumption, strong predictive power, and high accuracy. The scheme in this application addresses the following technical problems:

[0037] (1) Existing methods rely heavily on manual evaluation, making the assessment and prediction of CTO percutaneous coronary intervention (PCI) failure extremely time-consuming. Physicians typically need about 5 minutes to observe CTA images of the blood vessels and then score various evaluation indicators. To address this issue, this application proposes a more efficient assessment and prediction method that only requires the physician to select the pre-intervention site for PCI treatment. After that, no manual intervention is required, and the success rate of the PCI treatment can be assessed in a short time (about 10 seconds).

[0038] (2) Existing methods rely on physician experience, which leads to individual differences in experience. Therefore, the assessment and prediction of CTO percutaneous coronary intervention (PCI) treatment failure is unstable and inaccurate. To address this issue, this application proposes a method based on deep learning and machine learning technologies to analyze coronary CTA image data to assess the success rate of PCI treatment. This method can automatically, efficiently, and stably assess and predict the success rate of PCI treatment for different CTO patients.

[0039] The proposed solution includes the following steps: acquiring coronary CTA images; determining a vascular segmentation mask for the coronary CTA images using a pre-trained vascular segmentation model; receiving a user-inputted PCI pre-intervention location and cropping the coronary CTA images based on the PCI pre-intervention location to obtain a cropped image; inputting the vascular segmentation mask and the cropped image into a pre-trained feature extraction model to obtain image features corresponding to the coronary CTA images; extracting radiomics features corresponding to the vascular segmentation mask and the cropped image; and inputting the image features and the radiomics features into a pre-trained success rate prediction model to obtain a predicted success rate for PCI surgery. The proposed solution has the following advantages over existing solutions:

[0040] (1) Compared with manual scoring models, this application can automatically, quickly and accurately assess and predict the risk of failure of CTO percutaneous coronary intervention, reducing the assessment and prediction time from 5 minutes to about 10 seconds, which improves diagnostic efficiency while significantly reducing the workload of doctors and the waiting time of patients.

[0041] (2) Compared with manual rating models, this application can eliminate unstable factors such as individual differences of raters and differences in data distribution by learning and analyzing a large amount of multi-center data without relying on the experience of physicians, and can more accurately and stably evaluate and predict case data from different clinical centers.

[0042] (3) Compared with other deep learning or machine learning technologies, this application successfully combines deep learning technology, radiomics and traditional machine learning models and applies them to evaluate and predict the success rate of CTO percutaneous coronary intervention, providing a new, fully automated and rapid and accurate evaluation scheme for preoperative risk prediction of CTO percutaneous coronary intervention.

[0043] Figure 1 This is a flowchart illustrating a method for predicting the success rate of CTO interventional treatment according to an embodiment of this application. This method can be executed by a device for predicting the success rate of CTO interventional treatment according to an embodiment of this application. The device can be implemented using software and / or hardware and can be integrated into an electronic device, such as a computer used by a doctor or other devices.

[0044] like Figure 1 As shown, the method for predicting the success rate of CTO interventional treatment may include the following steps:

[0045] Step 101: Obtain coronary CTA images.

[0046] Coronary arteries are the vessels that carry the blood vessels.

[0047] When a patient with CTO lesions is considered for PCI treatment, coronary CTA images of the patient can be obtained first. CTA image sequences generally conform to the medical imaging format of the Digital Imaging and Communications in Medicine (DICOM) protocol. In actual use, it is necessary to ensure that the selected image sequences meet the basic requirements of CTA, such as no contrast agent filling and no obvious motion artifacts.

[0048] Step 102: Using a pre-trained vessel segmentation model, determine the vessel segmentation mask for the coronary CTA image.

[0049] The vessel segmentation model is pre-trained. It typically employs a 3D U-Net network structure. Training methods include: first, acquiring training samples, which can be obtained by experienced radiologists labeling vessels in training sample images. The labeled training sample images (acquired coronary CTA images) are used as sample images, and the corresponding labels are used as the ground truth labels during training, thus forming a training sample. Multiple training samples are then obtained. During training, the training sample images are input into the model to obtain the vessel segmentation results. The loss between the segmentation results and the ground truth labels is calculated, and the network parameters of the vessel segmentation model are adjusted based on this loss. When the loss is less than or equal to a preset threshold or convergence is achieved, the training of the vessel segmentation model is considered successful, resulting in a well-trained model. Optionally, the Dice loss function, cross-entropy loss function, or other types of loss functions can be used to calculate the loss. When adjusting the network parameters of the vessel segmentation model, a stochastic gradient descent (SGD) optimizer or other types of optimizers can be used; this disclosure does not impose any limitations on these methods. A well-trained vessel segmentation model can segment input coronary CTA images to obtain vessel segmentation results. The segmentation results are mask images, i.e., vessel segmentation masks. In other words, a well-trained vessel segmentation model can identify vessels in coronary CTA images and output vessel segmentation masks.

[0050] In this embodiment of the disclosure, after acquiring a coronary CTA image, the acquired coronary CTA image can be input into a pre-trained blood vessel segmentation model, and the blood vessel segmentation model outputs the blood vessel segmentation mask corresponding to the coronary CTA image.

[0051] It should be noted that since PCI treatment is an interventional treatment of the coronary arteries, the vascular segmentation mask obtained in this embodiment may only include the segmentation results of the coronary arteries.

[0052] Step 103: Receive the PCI pre-intervention location input by the user, and crop the coronary CTA image based on the PCI pre-intervention location to obtain a cropped image.

[0053] The pre-intervention position for PCI refers to the opening position where the doctor would consider performing PCI surgery if the patient were to undergo PCI treatment.

[0054] In this embodiment of the application, the user (i.e., the doctor) can input the PCI pre-intervention location. After receiving the PCI pre-intervention location, the electronic device can crop the acquired coronary CTA image based on the PCI pre-intervention location to obtain a cropped image.

[0055] In one optional embodiment of this disclosure, when cropping a coronary CTA image based on the pre-PCI intervention location to obtain a cropped image, a circular region can be defined on the coronary CTA image with the pre-PCI intervention location as the center and a preset radius. This circular region is then cropped to obtain the cropped image. Thus, the cropped image is a circular image cropped from the coronary CTA image with the pre-PCI intervention location as the center and the preset radius as the radius.

[0056] The preset radius can be set in advance according to actual needs.

[0057] In one optional embodiment of this disclosure, when cropping a coronary CTA image based on the pre-PCI intervention location to obtain a cropped image, a square region can be defined on the coronary CTA image with the pre-PCI intervention location as the center and a preset half-side length as the side length. This square region is then cropped to obtain the cropped image. Thus, the cropped image is a square image cropped from the coronary CTA image with the pre-PCI intervention location as the center and a side length of twice the preset half-side length.

[0058] The preset half-side length can be set in advance according to actual needs.

[0059] Step 104: Input the blood vessel segmentation mask and the cropped image into a pre-trained feature extraction model to obtain the image features corresponding to the coronary CTA image.

[0060] The feature extraction model is pre-trained.

[0061] In one optional embodiment of this application, the feature extraction model is obtained through the following steps:

[0062] Multiple training samples are obtained. Each training sample includes a sample image and a label corresponding to the sample image. The sample image includes a vessel segmentation mask sample belonging to the same coronary CTA image sample and a cropped image sample based on the PCI intervention location. The label is success or failure.

[0063] The initial deep neural network is trained using the multiple training samples to obtain a trained deep neural network.

[0064] The feature extraction model is obtained by discarding the output layer and activation function of the deep neural network.

[0065] The initial deep neural network could be, for example, a ResNet network.

[0066] In this embodiment, multiple training samples can be obtained before training the ResNet network. For example, with the patient's permission, real coronary CTA image samples can be obtained from cooperating hospitals. A pre-trained vessel segmentation model is then used to determine the vessel segmentation mask sample of the coronary CTA image sample, and a cropped image sample is extracted from the coronary CTA image sample based on the interventional location during PCI treatment. The vessel segmentation mask sample and the cropped image sample of the same coronary CTA image sample constitute a sample image, and the corresponding surgical outcome (success (opening) or failure (not open)) is labeled as the tag of that sample image.

[0067] Next, the initial deep neural network can be iteratively trained using the acquired training samples. The loss is calculated based on the output of the initial deep neural network and the labels of the sample images. Training ends when the loss is less than a preset value or convergence is achieved, resulting in a trained deep neural network. After model training is complete, the last layer of the deep neural network (i.e., the output layer and its activation function) is discarded, while the structure and parameters of the previous layers are retained, resulting in a feature extraction model. If new data is input again, the output of this feature extraction model will be the image feature vector extracted from the input data.

[0068] Therefore, in this embodiment of the application, after inputting the vascular segmentation mask and cropped image of the coronary CTA image into the pre-trained feature extraction model, the feature extraction model outputs the image features corresponding to the coronary CTA image, and the image features are feature vectors.

[0069] Step 105: Extract the radiomics features corresponding to the blood vessel segmentation mask and the cropped image.

[0070] In this embodiment of the application, corresponding radiomics features can be extracted from the blood vessel segmentation mask and cropped image based on radiomics technology.

[0071] Step 106: Input the image features and the radiomics features into a pre-trained success rate prediction model to obtain the predicted success rate of PCI surgery.

[0072] The success rate prediction model is pre-trained. During training, the success rate of PCI surgery at the PCI intervention site is used as the true label based on real cases (training samples). The image feature samples and radiomics feature samples of the training samples can be input into the random forest model for feature selection and model selection. The model is trained according to the optimal feature combination selected, and K-fold cross-validation (e.g., K=10) is performed. Finally, the prediction model with the highest accuracy is selected to obtain the success rate prediction model.

[0073] In this embodiment of the application, after obtaining the image features and radiomics features of the coronary CTA image, the image features and radiomics features can be input into a pre-trained success rate prediction model. The success rate prediction model outputs the predicted success rate of PCI surgery for the patient to whom the coronary CTA image belongs. The predicted success rate indicates the probability of success of PCI surgery for the patient. The higher the predicted success rate, the greater the probability of success of PCI treatment for the patient and the higher the chance of cure.

[0074] It is understood that in this embodiment, the success rate prediction model outputs the probability of success when performing PCI surgery at the pre-interventional PCI location. When the success rate is low, it indicates a higher surgical risk. In this case, the user can re-enter the pre-interventional PCI location to reassess the success rate until a satisfactory assessment result is obtained. This allows for the selection of a better PCI interventional location, thereby improving the success rate of PCI treatment.

[0075] The method for predicting the success rate of CTO interventional treatment according to this application embodiment acquires coronary CTA images, uses a pre-trained vessel segmentation model to determine the vessel segmentation mask of the coronary CTA images, receives the pre-interventional PCI location input by the user, and crops the coronary CTA images based on the pre-interventional PCI location to obtain a cropped image. Then, the vessel segmentation mask and the cropped image are input into a pre-trained feature extraction model to obtain the image features corresponding to the coronary CTA images, and to extract the radiomics features corresponding to the vessel segmentation mask and the cropped image. Finally, the image features and radiomics features are input into a pre-trained success rate prediction model to obtain the predicted success rate of PCI surgery. Using the above technical solution, the success rate of PCI surgery can be automatically, quickly, and accurately predicted simply by the user inputting the pre-interventional PCI location, eliminating the need for the user to score various evaluation indicators and avoiding the tedious image reading work.

[0076] In one optional embodiment of this application, the features input to the success rate prediction model can be fused features. Therefore, inputting the image features and the radiomics features into the pre-trained success rate prediction model includes:

[0077] The image features and the radiomics features are fused to obtain a fused feature vector;

[0078] The fused feature vector is input into a pre-trained success rate prediction model.

[0079] In one optional embodiment of this application, when performing feature fusion, the image features and the radiomics features can be concatenated to obtain the fused feature vector.

[0080] For example, radiomics features can be concatenated after image features to obtain a fused feature vector. Alternatively, image features can be concatenated after radiomics features to obtain a fused feature vector. This application does not restrict the concatenation method.

[0081] In one optional embodiment of this application, when performing feature fusion, the image features and the radiomics features can be added together to obtain the fused feature vector.

[0082] Then, the fused feature vector obtained by feature fusion can be input into the success rate prediction model, and the success rate prediction model outputs the predicted success rate of PCI surgery.

[0083] In one optional embodiment of this application, the user can select a point on the coronary CTA image, i.e., the location to be opened for percutaneous CTO intervention, and the electronic device will automatically crop the area near the selected point. Therefore, in this embodiment, receiving the user-inputted PCI pre-intervention location includes:

[0084] Receive user click operations on the coronary CTA image;

[0085] The location where the click operation is performed is determined as the PCI pre-intervention location.

[0086] Typically, a location approximately 1 mm upstream of the site of a chronic occlusive lesion is selected as the PCI intervention site. Therefore, physicians can perform a click operation approximately 1 mm upstream of the lesion location on the coronary CTA image. In this embodiment, the electronic device receives the user's click operation on the coronary CTA image, determines the location of the click operation, and then identifies this location as the pre-PCI intervention site.

[0087] Corresponding to the above method embodiments, this application also provides a device for predicting the success rate of CTO interventional treatment.

[0088] Figure 2 This is a schematic diagram of the structure of a CTO interventional treatment success rate prediction device provided in an embodiment of this application, as shown below. Figure 2 As shown, the CTO interventional treatment success rate prediction device 20 may include: a first acquisition module 210, a determination module 220, a trimming module 230, a second acquisition module 240, an extraction module 250, and a third acquisition module 260.

[0089] The first acquisition module 210 is used to acquire coronary CTA images;

[0090] The determination module 220 is used to determine the vascular segmentation mask of the coronary CTA image using a pre-trained vascular segmentation model;

[0091] The cropping module 230 is used to receive the PCI pre-intervention location input by the user and crop the coronary CTA image based on the PCI pre-intervention location to obtain a cropped image;

[0092] The second acquisition module 240 is used to input the blood vessel segmentation mask and the cropped image into a pre-trained feature extraction model to obtain the image features corresponding to the coronary CTA image.

[0093] Extraction module 250 is used to extract the radiomics features corresponding to the blood vessel segmentation mask and the cropped image;

[0094] The third acquisition module 260 is used to input the image features and the radiomics features into a pre-trained success rate prediction model to obtain the predicted success rate of PCI surgery.

[0095] Optionally, the CTO interventional treatment success rate prediction device 20 further includes: a training module; the training module is used for:

[0096] Multiple training samples are obtained. Each training sample includes a sample image and a label corresponding to the sample image. The sample image includes a vessel segmentation mask sample belonging to the same coronary CTA image sample and a cropped image sample based on the PCI intervention location. The label is success or failure.

[0097] The initial deep neural network is trained using the multiple training samples to obtain a trained deep neural network.

[0098] The feature extraction model is obtained by discarding the output layer and activation function of the deep neural network.

[0099] Optionally, the third acquisition module 260 includes:

[0100] The feature fusion unit is used to fuse the image features and the radiomics features to obtain a fused feature vector.

[0101] The output unit is used to input the fused feature vector into a pre-trained success rate prediction model.

[0102] Optionally, the feature fusion unit is further configured to:

[0103] The image features and the radiomics features are concatenated to obtain the fused feature vector; or

[0104] The image features and the radiomics features are added together to obtain the fused feature vector.

[0105] Optionally, the trimming module 230 is further configured to:

[0106] Using the pre-PCI intervention location as the center, a circular area is determined on the coronary CTA image according to a preset radius;

[0107] The circular region is cropped to obtain the cropped image.

[0108] Optionally, the trimming module 230 is further configured to:

[0109] Centered on the pre-PCI intervention location, a square region is determined on the coronary CTA image according to a preset half-side length;

[0110] The square region is cropped to obtain the cropped image.

[0111] Optionally, the trimming module 230 is further configured to:

[0112] Receive user click operations on the coronary CTA image;

[0113] The location where the click operation is performed is determined as the PCI pre-intervention location.

[0114] The CTO interventional treatment success rate prediction device provided in this application embodiment can execute any CTO interventional treatment success rate prediction method provided in this application embodiment, and has the corresponding functional modules and beneficial effects for executing the method. Content not described in detail in the device embodiments of this application can be referred to the description in any method embodiment of this application.

[0115] It should be noted that although several modules or units for the device used to perform actions have been mentioned in the detailed description above, this division is not mandatory. In fact, according to the embodiments of this application, the features and functions of two or more modules or units described above can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided and embodied by multiple modules or units.

[0116] In an exemplary embodiment of this application, an electronic device is also provided, comprising: a processor, the processor being configured to execute a computer program stored in a memory, the computer program being executed by the processor to implement the steps of the method for predicting the success rate of CTO interventional treatment as described in the above embodiments.

[0117] In an exemplary embodiment of this application, a computer-readable storage medium is also provided, on which a computer program is stored, which, when executed by a processor, implements the steps of the method for predicting the success rate of CTO interventional treatment as described in the above embodiments.

[0118] It should be noted that the computer-readable storage medium shown in this application can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory, read-only memory, erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this application, the computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. The program code contained on the computer-readable storage medium can be transmitted using any suitable medium, including but not limited to: wireless, wire, optical fiber, radio frequency, etc., or any suitable combination thereof.

[0119] In an exemplary embodiment of this application, a computer program product is also provided, which, when run on a computer, causes the computer to perform the steps of the method for predicting the success rate of CTO interventional treatment as described in the above embodiments.

[0120] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0121] The above description is merely a specific embodiment of this application, enabling those skilled in the art to understand or implement this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments described herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for predicting the success rate of CTO interventional treatment, characterized in that, The method includes: Acquire coronary CTA images; Using a pre-trained vessel segmentation model, a vessel segmentation mask for the coronary CTA image is determined; Receive the PCI pre-intervention location input by the user, and crop the coronary CTA image based on the PCI pre-intervention location to obtain a cropped image; The blood vessel segmentation mask and the cropped image are input into a pre-trained feature extraction model to obtain the image features corresponding to the coronary CTA image; Extract the radiomics features corresponding to the blood vessel segmentation mask and the cropped image; The image features and radiomics features are input into a pre-trained success rate prediction model to obtain the predicted success rate of PCI surgery. The success rate prediction model is pre-trained. During training, the success of PCI surgery at the PCI intervention site in real cases is used as the real label for training. The image feature samples and radiomics feature samples of the training samples are input into a random forest model for feature selection and model selection. The model is trained according to the selected optimal feature combination and K-fold cross-validation is performed. Finally, the prediction model with the highest accuracy is selected as the success rate prediction model. The step of cropping the coronary CTA image based on the pre-PCI intervention location to obtain a cropped image includes: Using the pre-PCI intervention location as the center, a circular region is determined on the coronary CTA image according to a preset radius. The circular region is then cropped to obtain the cropped image. or, Centered on the pre-PCI intervention location, a square region is determined on the coronary CTA image according to a preset half-side length. The square region is then cropped to obtain the cropped image.

2. The method according to claim 1, characterized in that, The feature extraction model is obtained through the following steps: Multiple training samples are obtained. Each training sample includes a sample image and a label corresponding to the sample image. The sample image includes a vessel segmentation mask sample belonging to the same coronary CTA image sample and a cropped image sample based on the PCI intervention location. The label is success or failure. The initial deep neural network is trained using the multiple training samples to obtain a trained deep neural network. The feature extraction model is obtained by discarding the output layer and activation function of the deep neural network.

3. The method according to claim 1, characterized in that, The step of inputting the image features and the radiomics features into a pre-trained success rate prediction model includes: The image features and the radiomics features are fused to obtain a fused feature vector; The fused feature vector is input into a pre-trained success rate prediction model.

4. The method according to claim 3, characterized in that, The step of fusing the image features and the radiomics features to obtain a fused feature vector includes: The image features and the radiomics features are concatenated to obtain the fused feature vector; or The image features and the radiomics features are added together to obtain the fused feature vector.

5. The method according to claim 1, characterized in that, The PCI pre-intervention location input by the user includes: Receive user click operations on the coronary CTA image; The location where the click operation is performed is determined as the PCI pre-intervention location.

6. A device for predicting the success rate of CTO interventional treatment, characterized in that, include: The first acquisition module is used to acquire coronary CTA images; A determination module is used to determine the vascular segmentation mask of the coronary CTA image using a pre-trained vascular segmentation model; The cropping module is used to receive the PCI pre-intervention location input by the user and crop the coronary CTA image based on the PCI pre-intervention location to obtain a cropped image. The second acquisition module is used to input the blood vessel segmentation mask and the cropped image into a pre-trained feature extraction model to obtain the image features corresponding to the coronary CTA image. The extraction module is used to extract the radiomics features corresponding to the blood vessel segmentation mask and the cropped image; The third acquisition module is used to input the image features and the radiomics features into a pre-trained success rate prediction model to obtain the predicted success rate of PCI surgery. The success rate prediction model is pre-trained. During training, the success of PCI surgery at the PCI intervention site in real cases is used as the real label for training. The image feature samples and radiomics feature samples of the training samples are input into a random forest model for feature selection and model selection. The model is trained according to the selected optimal feature combination and K-fold cross-validation is performed. Finally, the prediction model with the highest accuracy is selected as the success rate prediction model. The cropping module is also used for: Using the pre-PCI intervention location as the center, a circular region is determined on the coronary CTA image according to a preset radius. The circular region is then cropped to obtain the cropped image. or, Centered on the pre-PCI intervention location, a square region is determined on the coronary CTA image according to a preset half-side length. The square region is then cropped to obtain the cropped image.

7. An electronic device, comprising: A processor for executing a computer program stored in a memory, characterized in that, when executed by the processor, the computer program implements the steps of the method for predicting the success rate of CTO interventional treatment as described in any one of claims 1-5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the method for predicting the success rate of CTO interventional treatment as described in any one of claims 1-5.

Images