Oval foramen positioning detection method

Abstract

The invention provides a foramen ovale positioning detection method, which belongs to the technical field of medical detection and comprises the following steps of: scanning a heart area of a patient by using ultrasonic imaging equipment to obtain a two-dimensional image containing the foramen ovale; analyzing the feature points in the two-dimensional image based on a machine learning algorithm to determine the position of the foramen ovale; applying an image enhancement technology to improve the contrast and definition of the two-dimensional image; and the change of the position of the foramen ovale is continuously monitored through a real-time tracking algorithm, so that the positioning stability and accuracy in the whole detection process are ensured. Through the scheme of the invention, the problem of how to improve the positioning accuracy and stability in the positioning detection of the foramen ovale so as to ensure the accuracy of the detection result can be solved.

Description

Technical Field

[0001] This application relates to the field of medical detection technologies, and specifically to a method for detecting the location of the foramen ovale. Background Art

[0002] For the method of detecting the location of the foramen ovale, which is a technique for determining the position of the foramen ovale in the heart and is usually applied in the clinical diagnosis and treatment process, especially for diseases related to patent foramen ovale (PFO). This method uses imaging technologies such as echocardiography to identify and accurately locate the position of the foramen ovale. However, in actual operation, how to improve the accuracy and stability in the detection of the foramen ovale location is a challenging problem. This is mainly because the position of the foramen ovale may vary due to individual differences, and in addition, factors such as the operator's technical level, equipment accuracy, and the patient's physiological state during the detection process may all affect it, which may lead to inaccurate or unstable positioning, thus affecting the accuracy of the final detection result. Therefore, continuously optimizing the detection technology and improving the operator's professional skills are the keys to solving this problem. Summary of the Invention

[0003] In view of this, the present invention provides a method for detecting the location of the foramen ovale, which solves the problem of inaccurate location, identification, and judgment of the foramen ovale in the prior art.

[0004] The method for detecting the location of the foramen ovale includes:

[0005] Using an ultrasonic imaging device to scan the heart region of the patient to obtain a two-dimensional image containing the foramen ovale;

[0006] Analyzing the feature points in the two-dimensional image based on a machine learning algorithm to determine the position of the foramen ovale;

[0007] Applying image enhancement technology to improve the contrast and clarity of the two-dimensional image, thereby enhancing the recognition accuracy of the edge contour of the foramen ovale;

[0008] Continuously monitoring the change of the foramen ovale position through a real-time tracking algorithm to ensure the stability and accuracy of the positioning during the entire detection process.

[0009] Preferably, using an ultrasonic imaging device to scan the heart region of the patient to obtain a two-dimensional image containing the foramen ovale includes:

[0010] Adjusting the ultrasonic probe to a preset scanning position in the heart region;

[0011] Starting the ultrasonic imaging device and setting it to the heart mode to obtain high-quality images;

[0012] Performing an autofocus function to optimize the image clarity;

[0013] Collecting two-dimensional image data containing the foramen ovale.

[0014] Preferably, the step of analyzing the feature points in the two-dimensional image based on a machine learning algorithm to determine the position of the foramen ovale includes:

[0015] Preprocess the two-dimensional image, including grayscale conversion and smoothing filtering;

[0016] Extract the foramen ovale feature points based on the trained machine learning model;

[0017] Determine whether the feature points meet the preset foramen ovale shape condition, that is, calculate whether the compactness of the region formed by the feature points is within a specific range. If it meets, it is considered as the foramen ovale;

[0018] Determine the set of feature points that meet the conditions as the position of the foramen ovale.

[0019] Preferably, the step of extracting the foramen ovale feature points based on the trained machine learning model includes:

[0020] Input the preprocessed image into the trained machine learning model;

[0021] Based on the model prediction results, calculate the probability that each pixel point belongs to the foramen ovale;

[0022] If the probability of a certain pixel point is greater than the set threshold, mark it as a potential feature point;

[0023] Connect the high-probability pixel points to form the contour of the foramen ovale.

[0024] Preferably, calculating the probability that each pixel point belongs to the foramen ovale based on the prediction results of the model includes:

[0025] For each pixel point, use the softmax function to calculate the probability P that it belongs to the foramen ovale;

[0026] The calculation formula of the softmax function is: for the i-th pixel point, the probability P_i that it belongs to the foramen ovale = exp(z_i) / Σ(exp(z_j)), where z_i represents the score of the output layer of the model corresponding to the foramen ovale category, and Σ(exp(z_j)) represents the sum of the exponentials of all category scores.

[0027] Preferably, the step of connecting the high-probability pixel points to form the contour of the foramen ovale includes:

[0028] Perform connected component analysis on the high-probability pixel points;

[0029] Determine whether the area S of the connected component is greater than the minimum foramen ovale area threshold S_min. If so, retain the connected component;

[0030] Extract the boundaries of connected regions using the boundary tracking algorithm;

[0031] Fit the boundary points to obtain the contour of the foramen ovale.

[0032] Beneficial effects: By combining ultrasonic imaging and machine learning techniques, the present invention can automatically and accurately locate the foramen ovale, reduce the errors of manual operations, and improve the accuracy of detection.

[0033] Using image enhancement techniques, the position and shape of the foramen ovale can be clearly displayed, providing higher-quality imaging support.

[0034] Through the real-time tracking algorithm, the changes of the foramen ovale can be dynamically monitored, especially suitable for the real-time monitoring of the position changes of the foramen ovale during cardiac activities, providing a more accurate basis for clinical treatment.

[0035] This method is non-invasive and has high safety, suitable for a wide range of clinical applications, especially for the examination of infants and young children and patients with poor health. Description of the Drawings

[0036] In order to more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the drawings required for the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention, and therefore should not be regarded as limiting the scope. For those of ordinary skill in the art, without creative efforts, other related drawings can also be obtained based on these drawings.

[0037] Figure 1 is a flowchart of the foramen ovale positioning detection method;

[0038] Figure 2 is a flowchart of the step of using an ultrasonic imaging device to scan the cardiac region of a patient to obtain a two-dimensional image containing the foramen ovale;

[0039] Figure 3 is a flowchart of the step of analyzing the feature points in the two-dimensional image based on a machine learning algorithm to determine the position of the foramen ovale;

[0040] Figure 4 is a flowchart of the step of extracting the foramen ovale feature points based on a trained machine learning model;

[0041] Figure 5 is a flowchart of the step of connecting the high-probability pixel points to form the contour of the foramen ovale. Detailed Embodiments

[0042] To make the objectives, technical solutions, and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below with reference to embodiments and drawings. The illustrative embodiments and descriptions of the present invention are only used to explain the present invention and do not limit the present invention.

[0043] Next, refer to Figure 1 Describe this patent foramen ovale positioning detection method:

[0044] S101: Scan the cardiac region of the patient using an ultrasonic imaging device to obtain a two-dimensional image containing the patent foramen ovale. This process first requires a doctor or technician to place the ultrasonic probe on the chest wall of the patient, usually at the location of the left heart. The ultrasonic waves will penetrate the skin, muscle tissue, and ribs, reach the inside of the heart and reflect back to form an image of the heart structure. To obtain a high-quality image, the operator needs to adjust the angle and position of the probe to ensure that the best view of the patent foramen ovale can be captured. For example, in actual operation, the operator may choose the apical four-chamber view to observe the patent foramen ovale because this angle can provide a relatively clear image of the patent foramen ovale. In addition, to reduce noise and improve image quality, it may also be necessary to adjust the gain setting on the ultrasonic device. Once a clear image is obtained, the next step can be entered.

[0045] S102: Analyze the feature points in the two-dimensional image based on a machine learning algorithm to determine the position of the patent foramen ovale. After obtaining the two-dimensional image containing the patent foramen ovale, the next task is to identify the specific position of the patent foramen ovale. This step utilizes advanced machine learning techniques, especially the convolutional neural network CNN in deep learning. CNN can automatically extract features from the image and learn the ability to distinguish different structures through a training data set. In this example, CNN is trained to identify the specific shape and texture features of the patent foramen ovale. For example, the training data set may contain thousands of labeled images of the patent foramen ovale, and these images are preprocessed and then input into CNN for training. CNN continuously adjusts its weights through optimization algorithms such as backpropagation until it can accurately identify the patent foramen ovale. In actual applications, once the model is trained, it can be applied to new images to automatically identify the position of the patent foramen ovale. To further improve the accuracy of identification, other feature points, such as the edge of the atrial septum, can also be combined to assist in positioning.

[0046] S103: Improve the contrast and clarity of the two-dimensional image by applying image enhancement techniques, thereby enhancing the recognition accuracy of the edge contour of the foramen ovale. After obtaining the preliminary foramen ovale image, in order to improve the accuracy of subsequent analysis, it is necessary to perform enhancement processing on the image. This step can be achieved in various ways, such as using histogram equalization to improve the overall contrast of the image, or using a sharpening filter to enhance edge details. For example, by histogram equalization, the distribution range of gray levels in the image can be increased, making the contrast between the foramen ovale and surrounding tissues more obvious. The sharpening filter, on the other hand, can highlight the details of the foramen ovale edge, making its contour clearer. The application of these techniques helps to improve the recognition accuracy of subsequent machine learning algorithms, especially when dealing with low-contrast or blurred images.

[0047] S104: Continuously monitor the change in the position of the foramen ovale through a real-time tracking algorithm to ensure the stability and accuracy of positioning throughout the detection process. After the initial positioning of the foramen ovale is completed, in order to ensure the stability throughout the detection process, it is necessary to use a real-time tracking algorithm to continuously monitor the change in the position of the foramen ovale. This algorithm can predict the position in the next frame based on the position information of the foramen ovale in the previous frame and make corrections by comparing the predicted value with the actual value. For example, a Kalman filter can be used to estimate the movement trajectory of the foramen ovale and update the prediction model according to the position of the foramen ovale in the current frame. In addition, techniques such as optical flow method can be combined to calculate the displacement of the foramen ovale between consecutive frames, thereby more accurately tracking its position. This method can not only improve the accuracy of positioning but also effectively cope with the small changes in the position of the foramen ovale caused by the patient's breathing or heartbeat.

[0048] Through the above steps, the problem of how to improve the accuracy and stability of positioning in the foramen ovale positioning detection to ensure the accuracy of the detection results can be solved. The entire process starts from obtaining high-quality ultrasound images, performs feature point recognition through machine learning algorithms, then uses image enhancement techniques to improve the recognition accuracy, and finally ensures the stability of positioning through a real-time tracking algorithm. The comprehensive application of these techniques greatly improves the accuracy and reliability of the foramen ovale positioning detection.

[0049] Next, refer to Figure 2 Describe the specific steps of using an ultrasound imaging device to scan the patient's cardiac region to obtain a two-dimensional image containing the foramen ovale in the foramen ovale positioning detection method:

[0050] S201: Adjust the ultrasound probe to the preset cardiac region scanning position: The operator needs to place the ultrasound probe on the patient's chest wall, and the specific position should be determined according to anatomical landmarks and clinical experience to ensure that the probe can accurately point to the region where the cardiac foramen ovale is located. This process may require the assistance of auxiliary markers or a pre-set positioning system to help with precise alignment.

[0051] S202: Start the ultrasound imaging device and set it to the cardiac mode to obtain high-quality images: After the probe is correctly placed, the operator should start the ultrasound imaging device and set it to a mode suitable for cardiac imaging. This usually involves selecting specific cardiac imaging parameters, such as frequency, gain, etc., to ensure clear and well-contrasted images are obtained.

[0052] S203: Execute the autofocus function to optimize image clarity: To further improve image quality, the operator should activate the autofocus function on the ultrasound imaging device. This function automatically adjusts the focus position to ensure the details of the foramen ovale and its surrounding structures are optimally presented. In some cases, manual fine-tuning of the focus may also be required to achieve the best results.

[0053] S204: Acquire two-dimensional image data containing the foramen ovale: Once the image quality reaches a satisfactory level, the operator can start acquiring image data. This usually involves pressing the acquisition button or triggering the image capture process in other ways. The acquired two-dimensional image data can then be used for subsequent foramen ovale localization analysis.

[0054] Next, refer to Figure 3 Describe the specific steps of analyzing the feature points in the two-dimensional image based on a machine learning algorithm to determine the position of the foramen ovale in the foramen ovale localization detection method:

[0055] S301: First, preprocess the two-dimensional image. This step mainly includes converting the color image to a grayscale image and applying a smoothing filtering technique to reduce noise and enhance image quality. The grayscale conversion process can be achieved by converting the color information of each pixel to its brightness value, while a Gaussian filter or other methods can be used for smoothing filtering to reduce random noise in the image, thereby improving the accuracy of subsequent feature extraction.

[0056] S302: Then, extract the foramen ovale feature points based on the trained machine learning model. This process usually involves using a pre-trained model such as a convolutional neural network (CNN) to identify and label the key features in the image. The training phase requires a large number of labeled data sets, which contain images with known foramen ovale positions and their corresponding label information. In this way, the model can learn the feature patterns that distinguish the foramen ovale from other structures and automatically identify these patterns in new images.

[0057] S303: Then, determine whether the feature points meet the preset oval foramen shape condition, that is, calculate whether the compactness of the area formed by the feature points is within a specific range. The compactness is defined as the ratio of the area to the perimeter of the area formed by the feature points. Specifically, the oval foramen usually has a certain area range and a relatively stable ratio of area to perimeter. If the calculated ratio falls within the preset threshold range, it is considered that these feature points represent the position of the oval foramen.

[0058] S304: Finally, determine the set of feature points that meet the conditions as the position of the oval foramen. Once the feature points are confirmed as the oval foramen, the system will record the position information of these points, thus completing the positioning of the oval foramen. This information can be further used to guide subsequent medical operations or diagnostic processes. For example, in interventional surgery, accurately locate the oval foramen for occlusion treatment.

[0059] Next, refer to Figure 4 Describe the specific steps of extracting the oval foramen feature points based on the trained machine learning model in the oval foramen positioning detection method:

[0060] S401: First, input the preprocessed image data into the trained machine learning model. The preprocessing here may include, but is not limited to, operations such as grayscale conversion, noise removal, and size normalization to ensure that the input data meets the requirements of the model. After the preprocessed image is fed into the model, the model will analyze it and output the relevant information of each pixel point.

[0061] S402: Then, based on the prediction results of the model, calculate the probability that each pixel point belongs to the oval foramen. This process usually involves post-processing of the model output, such as applying the softmax function to obtain the probability distribution. In this way, a probability map can be obtained, where each pixel point has a corresponding probability value of belonging to the oval foramen.

[0062] S403: Then, if the probability of a certain pixel point is greater than the preset threshold, mark it as a potential oval foramen feature point. The selection of this threshold is crucial for the accuracy of the final detection result. Too high or too low may lead to false detection or missed detection. In this way, the pixel points most likely belonging to the oval foramen area can be screened out.

[0063] Specifically, the true positive rate (TPR) and false positive rate (FPR) can be calculated at different thresholds, and then the ROC curve can be drawn. Optimize the detection performance by maximizing the Youden index (TPR - FPR) to determine the optimal threshold.

[0064] S404: Finally, connect these high-probability pixel points to form the contour of the foramen ovale. This step can be achieved in various ways, such as using connected component analysis or edge detection algorithms to identify and connect these potential feature points, thereby outlining the general shape of the foramen ovale. This process helps to further improve the accuracy and robustness of the detection.

[0065] Through the above steps, the trained machine learning model can be effectively utilized to locate and detect the foramen ovale, providing important support for subsequent medical diagnosis.

[0066] Next, describe the specific steps of calculating the probability of each pixel point belonging to the foramen ovale based on the prediction results of the model in the foramen ovale location detection method:

[0067] For each pixel point, use the softmax function to calculate its probability P of belonging to the foramen ovale. Specifically, for the i-th pixel point, its probability of belonging to the foramen ovale is

[0068] P_i = exp(z_i) / Σ(exp(z_i)), where z_i represents the score of the foramen ovale category corresponding to the output layer of the model, and Σ(exp(z_i)) represents the sum of the exponentials of all category scores. This process ensures that the probability of each pixel point being classified as the foramen ovale can be accurately quantified, and the sum of the probabilities of all possible categories is equal to 1.

[0069] Next, refer to Figure 5 Describe the specific steps of connecting high-probability pixel points to form the contour of the foramen ovale in the foramen ovale location detection method:

[0070] S501: By performing connected component analysis on high-probability pixel points, the regions composed of these pixel points can be identified. Connected component analysis is an image processing technique used to find regions composed of adjacent pixels with the same or similar attributes such as color, grayscale value, etc. In the present invention, this technique is used to identify pixel clusters that may represent the foramen ovale.

[0071] S502: Determine whether the area S of the connected component is greater than the minimum foramen ovale area threshold S_min. This is to exclude those connected components with too small an area, which are less likely to be the foramen ovale. If the area of a certain connected component is greater than the preset minimum threshold S_min, then it is considered that the connected component may be part of the foramen ovale and is retained for further analysis.

[0072] S503: Use the boundary tracking algorithm to extract the boundary of the retained connected component. The boundary tracking algorithm is an algorithm that can move along the edge of the connected component and record the boundary. In this way, the external contour information of the connected component can be accurately obtained.

[0073] S504: Fit the contour of the foramen ovale by fitting boundary points. Once the boundary points of the connected region are obtained, mathematical methods such as ellipse fitting can be used to approximately represent the shape of the foramen ovale. This step is crucial for finally determining the position and size of the foramen ovale.

[0074] For example, in practical applications, assume that we have obtained an image containing multiple high-probability pixel points. First, we perform connected component analysis on these pixel points to identify all possible connected components. Then, we set a reasonable minimum foramen ovale area threshold S_min, such as 50 square pixel units, to filter out connected components with an area greater than 50 square pixel units. Next, we use a boundary tracing algorithm, such as the chain code tracing method, to extract the boundaries of these connected components. Finally, we use the method of ellipse fitting to determine the exact contour of the foramen ovale. In this way, we have completed the entire detection process from high-probability pixel points to the foramen ovale contour.

[0075] During the actual operation process, when this device is used, first, the ultrasonic imaging device scans the patient's heart area. This process aims to obtain a high-quality two-dimensional image containing the foramen ovale. To ensure that the image quality can meet the requirements of subsequent processing, the device usually adopts advanced imaging technologies to obtain clear and detailed image data. Subsequently, the software module based on machine learning algorithms starts to play a role. It can automatically identify and analyze the key feature points in these two-dimensional images, and then accurately determine the specific position of the foramen ovale. To further improve the positioning accuracy, image enhancement technology is applied to enhance the contrast and clarity of the image, especially having a significant improvement effect on the recognition accuracy of the foramen ovale edge contour. On this basis, the real-time tracking algorithm continuously monitors the change of the foramen ovale position to ensure the stability and accuracy of the positioning during the entire detection process. Through the close cooperation between various components of the entire system, not only the efficient and accurate positioning of the foramen ovale is achieved, but also a reliable diagnosis basis is provided for doctors, which helps to improve the safety and success rate of clinical treatment.

[0076] The methods, programs, systems, devices, etc. in the embodiments of the present invention can be executed or implemented in a single or multiple networked computers, and can also be practiced in a distributed computing environment. In the embodiments of this specification, in these distributed computing environments, tasks can be executed by remote processing devices connected through a communication network.

[0077] Those skilled in the art should understand that the embodiments of this specification can be provided as a method, a system, or a computer program product. Therefore, those skilled in the art can think that the implementation of the functional modules / units or controllers and related method steps clarified in the above embodiments can be achieved in a way of software, hardware, and the combination of software and hardware.

[0078] Unless otherwise specified, the acts or steps of the methods and procedures described according to the embodiments of the present invention do not have to be executed in a specific order and still can achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0079] In this document, multiple embodiments of the present invention have been described. For the sake of brevity, the description of each embodiment is not exhaustive, and the same or similar features or parts among the various embodiments may be omitted. In this document, "one embodiment", "some embodiments", "example", "specific example", or "some examples" mean applicable to at least one embodiment or example according to the present invention, rather than all embodiments. The above terms do not necessarily refer to the same embodiment or example. Without contradiction, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples.

[0080] The exemplary systems and methods of the present invention have been specifically shown and described with reference to the above embodiments, which are only examples of the best mode for implementing the systems and methods. Those skilled in the art can understand that various changes can be made to the embodiments of the systems and methods described herein when implementing the systems and / or methods without departing from the spirit and scope of the present invention defined in the appended claims.

Claims

1. A method for detecting the location of the foramen ovale, characterized in that, Comprising: Scanning the cardiac region of a patient using an ultrasonic imaging device to obtain a two-dimensional image containing the foramen ovale; Analyzing the feature points in the two-dimensional image based on a machine learning algorithm to determine the position of the foramen ovale; Applying image enhancement technology to improve the contrast and clarity of the two-dimensional image; Continuously monitoring the change of the position of the foramen ovale through a real-time tracking algorithm.

2. The method for detecting patent foramen ovale position according to claim 1, wherein Scanning the cardiac region of a patient using an ultrasonic imaging device to obtain a two-dimensional image containing the foramen ovale includes: Adjusting the ultrasonic probe to a preset scanning position in the cardiac region; Starting the ultrasonic imaging device and setting it to the cardiac mode to obtain high-quality images; Performing an autofocus function to optimize the image clarity; Collecting two-dimensional image data containing the foramen ovale.

3. The method for detecting the position of the foramen ovale according to claim 1, wherein The steps of analyzing the feature points in the two-dimensional image based on a machine learning algorithm to determine the position of the foramen ovale include: Preprocessing the two-dimensional image, including grayscale conversion and smoothing filtering; Extracting the foramen ovale feature points based on a trained machine learning model; Judging whether the feature points meet the preset foramen ovale shape condition, that is, calculating whether the compactness of the region formed by the feature points is within a specific range; Determining the set of feature points that meet the conditions as the position of the foramen ovale.

4. The method for detecting the positioning of the foramen ovale according to claim 3, wherein The steps of extracting the foramen ovale feature points based on a trained machine learning model include: Inputting the preprocessed image data into a trained machine learning model; Calculating the probability that each pixel point belongs to the foramen ovale based on the prediction result of the model; If the probability of a certain pixel point is greater than a preset threshold, then mark it as a potential foramen ovale feature point; Connecting the high-probability pixel points to form the contour of the foramen ovale.

5. The method for detecting the positioning of the foramen ovale according to claim 4, wherein Calculating the probability that each pixel point belongs to the foramen ovale based on the prediction result of the model includes: For each pixel point, calculating the probability P that it belongs to the foramen ovale using the softmax function; The calculation formula of the softmax function is: for the i-th pixel point, the probability P_i that it belongs to the foramen ovale = exp(z_i) / Σ(exp(z_j)), where z_i represents the score of the output layer of the model corresponding to the foramen ovale category, and Σ(exp(z_j)) represents the sum of the exponents of all category scores.

6. The method for detecting foramen ovale positioning according to claim 4, characterized in that, The steps of connecting the high-probability pixel points to form the contour of the foramen ovale include: Performing connected component analysis on the high-probability pixel points; Judging whether the area S of the connected component is greater than the minimum foramen ovale area threshold S_min, if so, retaining the connected component; Using a boundary tracking algorithm to extract the boundary of the connected component; Obtaining the contour of the foramen ovale by fitting the boundary points.

Images