Ultrasonic 3D modeling method based on ECG
By combining ECG signals and echocardiography models, detecting the peak time of the R wave and calculating the isocapacity time window, screening ultrasound images of the same phase for three-dimensional reconstruction, solving the problems of insufficient time and accuracy in the existing technology, and achieving efficient and accurate three-dimensional model reconstruction.
Patent Information
- Application Number
- CN202510928682.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-07-07
- Publication Date
- 2025-08-08
AI Technical Summary
The existing three-dimensional reconstruction technology takes too long operation and the accuracy of the echocardiac simulation model is insufficient, resulting in motion artifacts and structural distortion, affecting clinical application.
Based on the combination of ECG signal and the ultrasonic heart simulation model, by detecting the peak time image of the R wave and calculating the isocapacity time window, ultrasonic images of the same phase are selected for three-dimensional reconstruction, and image registration and screening are used for image registration and screening.
It improves the accuracy of the echocardiac simulation model and the correctness of clinical diagnosis, reduces operating time and improves the accuracy of three-dimensional reconstruction.
Smart Images
Figure CN120451416A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of ultrasound technology, and in particular to an ECG-based ultrasound 3D modeling method. Background Art
[0002] In the field of cardiac ultrasound, 3D mapping and navigation systems have become a diagnostic aid for cardiologists and are used in clinical treatment. This 3D cardiac reconstruction technology creates a 3D model of human organs or tissues based on 2D image information and its coordinates. This model allows for intuitive observation of the organ's internal structure. Transforming the 3D model allows for multi-directional observation, assisting physicians in accurately locating heart disease lesions. Adding coordinate information to the 3D model identifies the location of the lesion and calculates its size and volume.
[0003] However, existing 3D reconstruction technology has significant bottlenecks: 1. The operation is too time-consuming: to obtain a complete set of images of a single anatomical structure, the catheter needs to be rotated and scanned for hundreds of cardiac cycles; 2. Accuracy of ultrasound heart simulation model: Relying on ECG signals to select only a single frame of specific phase images in each cardiac cycle, resulting in insufficient number of valid images and the reconstructed model is prone to motion artifacts or structural distortion.
[0004] The acquisition efficiency of two-dimensional images is low and the spatiotemporal information within the cardiac cycle is not fully utilized, which restricts the accuracy and clinical practicality of the three-dimensional model. Therefore, an ECG-based ultrasound 3D modeling method is proposed to solve the above problems. Summary of the Invention
[0005] (1) Technical problems solved In response to the shortcomings of the existing technology, the present invention provides an ECG-based ultrasound 3D modeling method. This patent mainly solves the problems of long time and insufficient spatial resolution in acquiring two-dimensional cardiac images. Based on the simulated ECG signal and the ultrasonic cardiac simulation model, the R-wave moment image is selected, and the image of the same phase (isovolumetric contraction period or isovolumetric relaxation period) is calculated according to the dynamic calculation of the isovolumetric period time window and screened. The ultrasonic two-dimensional images with a value greater than or equal to the adaptive threshold of the reference image are screened and stored for later three-dimensional reconstruction, which improves the accuracy of the ultrasonic cardiac simulation model. At the same time, it also greatly improves the application significance of clinical teaching and the accuracy of the doctor's later analysis and diagnosis.
[0006] (2) Technical solution To achieve the above objectives, the present invention is implemented through the following technical solutions: The method for ECG-based ultrasound 3D modeling comprises the following steps: Step (1) Synchronous acquisition: The two-dimensional B-ultrasound image of the ultrasound simulation heart model is periodically scanned by continuous spatial rotation, and the simulated ECG signal is simultaneously acquired, and the continuously acquired heartbeat signal is synchronized with the ultrasound two-dimensional image in time, wherein the ultrasound signal includes the signal continuously acquired within a cardiac cycle, and the cardiac cycle refers to a mechanical wave propagation cycle consisting of each contraction and relaxation of the simulated heart; Step (2) R wave detection: The PT algorithm is used to detect the R wave peak of the ECG signal, and the corresponding cardiac ultrasound image at the R wave peak moment is detected according to time synchronization. The PT algorithm is an adaptive dual-threshold QRS wave detection algorithm that can be used for real-time processing of R waves. The detection is mainly based on the morphological characteristics of the R wave, including amplitude, slope, and time information; Step (3) Dynamic time window: Since the peak of the R wave is the starting point of the isovolumetric contraction period, the isovolumetric contraction period time window starts at the peak of the R wave and lasts for 5%-8% of the current RR interval (RR interval refers to the time interval between two R waves in the electrocardiogram); the starting point of the isovolumetric relaxation period is the end of the T wave, which is determined 40ms after the peak of the R wave is detected by the PT algorithm, and the time window is 7%-10% of the RR interval. Based on this dynamic time window, the ultrasonic two-dimensional images of the isovolumetric contraction period and the isovolumetric relaxation period are saved for later processing; Step (4) Image screening: The user selects an image and calculates the similarity of all ultrasound images in the same phase (isovolumetric contraction or isovolumetric relaxation) with the selected image as the benchmark. The improved NCC (Normalized Cross Correlation) normalized cross correlation algorithm is mainly used to eliminate probe motion artifacts and improve image spatial consistency.
[0007] Preferably, the main process in step (4) image screening is as follows: Automatic selection of reference image: R wave peak time image or T wave time image as the reference Image registration: affine transformation of in-phase images (translation / rotation compensation) Adaptive threshold: When the baseline signal-to-noise ratio (SNR) is greater than 30, the threshold is 0.98; when the SNR is less than 30, the threshold is 0.95. Images greater than or equal to the adaptive threshold are retained for later 3D modeling.
[0008] Preferably, the algorithm in step (2) R wave detection mainly includes two parts: pre-processing to achieve R wave enhancement and R wave comprehensive decision-making: (2a) Preprocessing After bandpass filtering and differential filtering, the ECG signal has been filtered out of power frequency noise and baseline drift noise. The differential filtering operation can enhance the slope of the R wave, highlighting the R wave component while suppressing the P wave and T wave. The subsequent squaring process makes the signal amplitude positive, avoiding downward peaks, such as inverted R wave peaks. The sliding integral is then used to obtain the R wave peak position to be determined. (2b) Threshold screening After multi-channel filtering, the peaks detected may be R waves or mixed noise. The algorithm uses a series of threshold conditions to filter peaks, ignoring all peaks within 200ms before and after the larger peak. The algorithm then compares the peaks to determine whether there are double peaks: if the signal peak appears 360ms after the detected R wave, the peak is considered the R wave. In the dual-threshold detection method adopted, the second threshold is half of the first threshold. If the peak is greater than the first threshold, it is regarded as an R wave, otherwise it is interference; (2c) Retrospective determination of the R wave location According to the Pan-Tompkins R-wave peak detection algorithm, the analog ECG signal is searched and the R-wave is effectively detected through slope, amplitude and width information. After the R-wave peak is found, the corresponding cardiac ultrasound image at the R-wave peak moment is detected according to time synchronization.
[0009] (3) Beneficial effects By combining analog ECG signals with ultrasound image acquisition, the present invention achieves: 1. The simulated ECG signal is combined with the ultrasound image collected by the probe. The time synchronization between the two facilitates the subsequent screening of the ultrasound image; 2. After detecting the image at the peak moment of the R wave using the PT algorithm, dynamically calculate the isovolumetric time window and save the ultrasound image of the same phase (isovolumetric contraction or isovolumetric relaxation) throughout the cardiac cycle; 3. Image similarity comparison: Affine transformation registration and NCC screening are performed on the acquired ultrasound image and the image at the R-wave peak moment, and the required two-dimensional ultrasound image is finally determined for three-dimensional model reconstruction, which greatly improves the accuracy of the subsequent three-dimensional model reconstruction. BRIEF DESCRIPTION OF THE DRAWINGS
[0010] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention and implement it according to the contents of the specification, the following is a detailed description of the preferred embodiments of the present invention with reference to the accompanying drawings.
[0011] Figure 1 Flowchart of the method for ECG-based ultrasound 3D modeling of the present invention. DETAILED DESCRIPTION
[0012] The embodiments of the present application provide an ECG-based ultrasound 3D modeling method to address the problems of long two-dimensional image acquisition time and insufficient spatial resolution in the prior art. Based on the ECG R-wave apex image and the temporal relationship between the cardiac cycle, ultrasound image sequences of specific phases within the cardiac cycle are efficiently extracted. Through NCC calculation, two-dimensional images with high similarity are screened out, thereby improving efficiency and accuracy.
[0013] Example 1 like Figure 1 As shown, the technical solution in the embodiment of the present application is to solve the above-mentioned problems of long acquisition time and insufficient spatial resolution of two-dimensional images. The overall idea is as follows: To address the problems in the prior art, the present invention provides an ECG-based ultrasound 3D modeling method, the specific steps of which are as follows: Step (1) Synchronous acquisition: The two-dimensional B-ultrasound image of the ultrasound simulation heart model is periodically scanned by continuous spatial rotation, and the simulated ECG signal is simultaneously acquired, and the continuously acquired heartbeat signal is synchronized with the ultrasound two-dimensional image in time, wherein the ultrasound signal includes the signal continuously acquired within a cardiac cycle, and the cardiac cycle refers to a mechanical wave propagation cycle consisting of each contraction and relaxation of the simulated heart; Step (2) R wave detection: The PT algorithm is used to detect the R wave peak of the ECG signal, and the corresponding cardiac ultrasound image at the R wave peak moment is detected according to time synchronization. The PT algorithm is an adaptive dual-threshold QRS wave detection algorithm that can be used for real-time processing of R waves. The detection is mainly based on the morphological characteristics of the R wave, including amplitude, slope, and time information; Step (3) Dynamic time window: Since the peak of the R wave is the starting point of the isovolumetric contraction period, the isovolumetric contraction period time window starts at the peak of the R wave and lasts for 5%-8% of the current RR interval; The starting point of the isovolumetric relaxation period is the end of the T wave, which is determined 40 ms after the peak of the R wave detected by the PT algorithm. The time window is 7%-10% of the RR interval. Based on this dynamic time window, the ultrasound two-dimensional images of the isovolumetric contraction and relaxation periods are saved for later processing. Step (4) Image screening: The user selects an image and calculates the similarity of all ultrasound images in the same phase (isovolumetric contraction or isovolumetric relaxation) with the selected image as the benchmark. The improved NCC (Normalized Cross Correlation) normalized cross correlation algorithm is mainly used to eliminate probe motion artifacts and improve image spatial consistency.
[0014] Preferably, the main process in step (4) image screening is as follows: (4a) Automatic selection of the reference image: the image at the peak of the R wave or the image at the T wave is used as the reference; (4b) Image registration: perform affine transformation (translation / rotation compensation) on the same-phase images; (4c) Adaptive threshold: When the baseline signal-to-noise ratio (SNR) is >30, the threshold is 0.98; when the SNR is <30, the threshold is 0.95. Images greater than or equal to the adaptive threshold are retained for later 3D modeling.
[0015] The RR interval in the dynamic time window of step (3) is , and the RR interval refers to the time interval between two R waves in the electrocardiogram.
[0016] The algorithm in step (2) of R wave detection mainly includes two parts: pre-processing to achieve R wave enhancement and R wave comprehensive decision-making: (2a) Preprocessing After bandpass filtering and differential filtering, the ECG signal has been filtered out of power frequency noise and baseline drift noise. The differential filtering operation can enhance the slope of the R wave, highlighting the R wave component while suppressing the P wave and T wave. The subsequent squaring process makes the signal amplitude positive, avoiding downward peaks, such as inverted R wave peaks. The sliding integral is then used to obtain the R wave peak position to be determined. (2b) Threshold screening After multi-channel filtering, the peaks detected may be R waves or mixed noise. The algorithm uses a series of threshold conditions to filter peaks, ignoring all peaks within 200ms before and after the larger peak. Comparison is then made to determine whether there are double peaks: If the signal peak occurs 360ms after the detected R wave, the peak is considered an R wave. The second threshold in the dual-threshold detection method is half the first. If the peak is greater than the first threshold, it is considered an R wave; otherwise, it is interference. (2c) Retrospective determination of the R wave location According to the Pan-Tompkins R-wave peak detection algorithm, the analog ECG signal is searched and the R-wave is effectively detected through slope, amplitude and width information. After the R-wave peak is found, the corresponding cardiac ultrasound image at the R-wave peak moment is detected according to time synchronization.
[0017] This patent reduces user operation time by combining multiple ultrasound images obtained with simulated ECG electrocardiogram signals, effectively screens the obtained two-dimensional ultrasound images, and improves the accuracy of constructing three-dimensional models, which is of great significance for clinical teaching.
[0018] The following experimental scenarios were tested: Table 1 shows the data collected using the heart simulation model and ECG signal generator according to the process.
[0019] Table 1 Beneficial effects: 1. The simulated ECG signal is combined with the ultrasound image collected by the probe. The time synchronization between the two facilitates the subsequent screening of the ultrasound image; 2. After detecting the image at the peak moment of the R wave using the PT algorithm, dynamically calculate the isovolumetric time window and save the ultrasound image of the same phase (isovolumetric contraction or isovolumetric relaxation) throughout the cardiac cycle; 3. Image similarity comparison: Affine transformation registration and NCC screening are performed on the acquired ultrasound image and the image at the R-wave peak moment, and the required two-dimensional ultrasound image is finally determined for three-dimensional model reconstruction, which greatly improves the accuracy of the subsequent three-dimensional model reconstruction.
[0020] Finally, it should be noted that the above embodiments are merely examples for the purpose of illustrating the present invention and are not intended to limit the embodiments. Those skilled in the art will readily appreciate that other variations or modifications based on the above description are possible. It is not necessary and impossible to provide an exhaustive list of all embodiments. However, obvious variations or modifications arising therefrom remain within the scope of protection of the present invention.
Claims
1. A method for ECG-based ultrasound 3D modeling, characterized in that: The specific steps are as follows: Step (1) Synchronous acquisition: Periodically scan the ultrasound simulation heart model, synchronously acquire simulated ECG signals and two-dimensional B-ultrasound images, and achieve time synchronization between the heartbeat signal and the ultrasound image; Step (2) R wave detection: using the PT algorithm to detect the R wave peak of the ECG signal, and determining the ultrasound image corresponding to the R wave peak moment according to time synchronization; Step (3) Dynamic time window: dynamically determining the isovolumetric contraction and isovolumetric relaxation time windows based on the R wave peak and the RR interval, and saving the ultrasound image within the time window; Step (4) Image screening: Select a reference image, perform similarity calculation and registration processing on all ultrasound images with the same phase, and retain images that meet the threshold for three-dimensional modeling.
2. The method for ECG-based ultrasound 3D modeling according to claim 1, characterized in that: The PT algorithm in step (2) includes: (2a) Preprocessing: The ECG signal is subjected to bandpass filtering, differential filtering, squaring, and sliding integration in sequence to enhance the R-wave characteristics and suppress noise; (2b) Threshold screening: A dual-threshold condition is used to screen peaks, where the second threshold is half the first threshold. If the peak value is greater than the first threshold, it is judged as an R wave, and peaks within 200 ms before and after the R wave are ignored. If the peak value occurs 360 ms after the detected R wave and is greater than the first threshold, it is judged as a new R wave. (2c) Retrospective judgment: Locate the peak of the R wave based on the slope, amplitude, and width information.
3. The method for ECG-based ultrasound 3D modeling according to claim 2, wherein: The step (3) dynamic time window, The starting point of the isovolumetric contraction period is the peak of the R wave, and the duration is set to 5%-8% of the current RR interval; The starting point of the isovolumetric relaxation period is the end of the T wave, the duration is set to 7%-10% of the RR interval, and the end time is determined by 40ms after the peak of the R wave.
4. The method for ECG-based ultrasound 3D modeling according to claim 3, characterized in that: The image screening step (4) specifically includes: The similarity with the benchmark image is calculated based on the improved normalized cross-correlation algorithm; Image registration via affine transformation; An adaptive threshold was set according to the signal-to-noise ratio (SNR) of the reference image: when SNR>30, the threshold was 0.98; when SNR≤30, the threshold was 0.
95. Images with similarity ≥ the threshold were retained.
5. The ECG-based ultrasound 3D modeling method according to claim 4, characterized in that: The reference image in step (4) is an ultrasound image at the peak moment of the R wave or the end moment of the T wave.
6. The ECG-based ultrasound 3D modeling method according to claim 4, characterized in that: The affine transformation includes translation and rotation compensation to eliminate probe motion artifacts.