CN-121714242-B - Pulmonary artery high pressure identification device and method based on multipath electrocardiographic heart sound signals

CN121714242BCN 121714242 BCN121714242 BCN 121714242BCN-121714242-B

Abstract

The invention discloses a pulmonary artery high-pressure identification device and method based on multipath electrocardiographic heart sound signals, and relates to the technical field of biomedicine, wherein the device comprises the steps of limiting the falling points of aortic valve closing candidate peaks and pulmonary valve closing candidate peaks in a main peak anchor point time boundary of a double-anchor split fragment set and an electrocardiograph fragment key turning point time interval, and executing adjacent beat retrospective substitution and multi-channel consistent voting reservation on cross-beat jump candidate peaks to generate a physiological consistency split fragment set; and converging split sequences of multiple paths of heart sounds according to the physiological consistency split fragment set, performing beat-level splicing and intra-window consistency closure with the form items related to the right heart load of multiple paths of electrocardiograms, and generating a pulmonary artery high pressure identification diagnosis set. The invention realizes the high-precision synchronization of the multi-path electrocardiograph heart sound signals, eliminates equipment difference and noise interference, ensures the reliability of feature extraction and improves the high-pressure recognition stability of pulmonary artery.

Inventors

GAO XIAOFENG
LI ZHENG
CAI LINWEI
QIAO JINGRONG

Assignees

北京麦迪克斯科技有限公司

Dates

Publication Date: 20260508
Application Date: 20260213

Claims (9)

1. The pulmonary artery high pressure identification method based on the multipath electrocardiographic heart sound signals is characterized by comprising the steps of, Receiving digital sampling streams of multi-path electrocardiosignals and multi-path heart sound signals, and carrying out sliding window slicing and window boundary alignment in a unified time reference to generate a synchronous window sequence; According to the synchronous window sequence, main peak interval correction and abnormal interval elimination are carried out to obtain beat-by-beat boundaries, multi-path electrocardiographic fragments and multi-path heart sound fragments are intercepted, cross-channel envelope alignment is carried out at the same time to generate a beat-by-beat alignment fragment set, the steps are as follows, Defining the time boundary of the adjacent corrected main peak sequence as a beat-to-beat boundary, executing window crossing and splicing, and simultaneously intercepting multiple paths of electrocardiographic fragments and multiple paths of heart sound fragments to generate a beat-to-beat fragment interception set; Uniformly aligning start and stop moments of the beat boundaries according to the beat-to-beat segment intercept sets, and performing adjacent beat backtracking replacement on the beat-to-beat boundaries corresponding to the rejection interval list to generate a beat-to-beat segment alignment list; performing envelope extraction on multiple paths of heart sound fragments in the beat-to-beat fragment alignment list, positioning an envelope anchor point sequence, performing time axis translation compensation and peak width stretching compression compensation, and generating a beat-to-beat alignment fragment set; Extracting and fusing the electrocardio slow-change track and the heart sound envelope slow-change track from the beat-to-beat alignment segment set, constructing a respiratory phase track, and carrying out phase tracking to obtain an inspiration segment and an expiration segment; Positioning aortic valve closing candidate peaks and pulmonary valve closing candidate peaks in an inspiration section and an expiration section, solidifying the relative time interval of the candidate peaks, and generating a double-anchoring split fragment set through robust convergence and rollback splicing; Limiting the falling points of the aortic valve closing candidate peaks and the pulmonary valve closing candidate peaks in the time interval between the main peak anchor point time boundary and the key turning point time boundary of the electrocardiographic fragments of the double-anchor split fragment set, and executing adjacent beat backtracking substitution and multi-channel consistent voting reservation on the cross-beat jump candidate peaks to generate a physiological consistency split fragment set; And converging split sequences of multiple paths of heart sounds according to the physiological consistency split fragment set, performing beat-level splicing and intra-window consistency closure with the form items related to the right heart load of multiple paths of electrocardiograms, and generating a pulmonary artery high pressure identification diagnosis set.
2. The pulmonary artery high pressure identification method based on multi-channel electrocardiographic heart sound signals of claim 1, wherein the generating the synchronization window sequence comprises the following steps, Receiving digital sampling streams of multi-channel electrocardiosignals and multi-channel heart sound signals, executing cross-channel arrival time difference verification, converting the cross-channel arrival time difference verification into channel alignment offset, and generating a time alignment acquisition stream; mapping the time alignment acquisition stream to a unified time reference, and executing segment aggregation and segment interpolation filling, and simultaneously reserving an original sampling coverage area to generate a unified sampling caliber alignment stream; and carrying out sliding window slicing on the unified sampling caliber alignment flow to obtain a plurality of paths of electrocardiographic sliding window segments and a plurality of paths of heart sound sliding window segments, and executing window boundary drift compensation alignment and boundary retraction window to generate a synchronous window sequence.
3. The pulmonary artery high pressure identification method based on multi-path electrocardiographic heart sound signals of claim 2, wherein the main peak interval correction and abnormal interval elimination are performed as follows, Band-pass shaping and baseline drift suppression are carried out on multiple paths of electrocardiographic sliding window segments in the synchronous window sequence, and the number of simultaneous hits across channels and the time deviation range are counted, so that a main peak candidate queue is generated; Based on the main peak candidate queue, merging and de-duplication, backtracking and complement and local smoothing correction are executed, abnormal intervals which do not meet the continuity constraint are eliminated, and a corrected main peak sequence and eliminating interval list is generated.
4. The pulmonary artery high pressure identification method based on multi-channel electrocardiographic heart sound signals of claim 1, wherein the steps of obtaining an inspiration segment and an expiration segment are as follows, Extracting a slow-change baseline sequence, an amplitude slow-change sequence and an envelope slow-change sequence from a beat-to-beat alignment fragment set, and performing beat-level alignment and channel position identification aggregation to generate a beat-level slow-change track alignment set; performing same-scale normalization and trend depolarization processing on the beat-level slow-change track alignment set, establishing cross-channel respiratory agent consistency constraint, and simultaneously performing weighted fusion and back-off recalculation to generate a respiratory phase track; And extracting the boundaries of the phase rising section and the phase falling section from the respiratory phase track, and executing phase continuity back-off splicing on burst phase reversal to generate an inhalation section and an exhalation section.
5. The method for pulmonary artery high pressure identification based on multi-path electrocardiographic heart sound signals of claim 4, wherein the generating the set of dual-anchor split fragments comprises the steps of, Reading multi-path heart sound fragments in the beat-by-beat alignment fragment set according to the inspiration section and the expiration section, positioning a second heart sound coverage area, performing multi-scale smoothing and local extremum extraction, and generating a second heart sound candidate peak queue; performing cross-channel candidate peak consistency gating screening on the segmented second heart sound candidate peak queue, positioning aortic valve closing candidate peaks and pulmonary valve closing candidate peaks, solidifying the relative time intervals of the candidate peaks, and converging the relative time intervals into a segmented relative time interval sequence; And performing robust convergence on the segmented relative time-distance sequences, performing rollback splicing on the isolated mutation and the continuous shortage port, and simultaneously supplementing candidate peak pairs to generate a double-anchoring split fragment set.
6. The method for identifying pulmonary artery high pressure based on multi-path electrocardiographic heart sound signals according to claim 5, wherein the method limits the falling points of aortic valve closing candidate peaks and pulmonary valve closing candidate peaks to the time interval between the main peak anchor point time boundary of the double-anchor split segment set and the key turning point time of the electrocardiograph segment, Combining the double-anchoring split fragment set with the beat-by-beat alignment fragment set, extracting a main peak anchor point time boundary and an electrocardio fragment key turning point time interval, constructing an effective time window for allowing falling points, and generating a beat-by-beat falling point constraint list; According to the beat-by-beat falling point constraint list, performing double constraint falling point clipping on the aortic valve closing candidate peak and the pulmonary valve closing candidate peak falling point, and performing nearest rollback on the out-of-range falling point according to the peak position support priority of the multipath heart sound fragments to generate out-of-range rollback clipping candidate peaks; And executing sequence conflict rearrangement rechecking in an effective time window based on the rollback displacement abstract in the out-of-range rollback clipping candidate peak, and generating a double-constraint clipping candidate peak pair set.
7. The method for pulmonary artery high pressure identification based on multi-channel electrocardiographic heart sound signals of claim 6, wherein the generating of the physiologically uniform split fragment set comprises the steps of, Cutting a candidate peak pair set based on double constraint, constructing a candidate peak relative time distance sequence, positioning a cross-beat jump section, executing adjacent beat backtracking replacement and backtracking rescaling, and generating a backtracking replacement candidate peak pair set; And rechecking and summarizing peak position support of the multipath heart sound fragments according to the backtracking substitution candidate peak pair set, and executing multi-channel consistent voting reservation to generate a physiological consistency split fragment set.
8. The method for identifying pulmonary artery high pressure based on multi-path electrocardiographic heart sound signals according to claim 7, wherein the step of generating a pulmonary artery high pressure identification diagnosis set comprises the following steps, Extracting split peak shape abstract and energy abstract from multi-path heart sound fragment according to physiological consistency split fragment set, and executing split sequence cross-channel credible aggregation to generate multi-path heart sound split sequence; Positioning a right heart load related morphology item from a plurality of paths of electrocardiograph fragments, performing beat level alignment and splicing with a plurality of paths of heart sound splitting sequences, and simultaneously performing adjacent beat alignment and channel rollback finishing on the missing items to generate a beat level splicing item sequence; And based on the beat-level spliced item sequence, performing matching check on the relative time and distance directions of the candidate peaks, and performing double-gate rollback arrangement and beat sequence number continuity check on the cross-beat jump residues to generate a pulmonary artery high pressure identification diagnosis set.
9. The pulmonary artery high pressure recognition device based on the multi-path electrocardiographic heart sound signals is characterized by comprising a host machine, a data processing and mapping analysis function and a data processing and mapping analysis function, wherein the host machine can receive signals transmitted by a collection head and is used for executing the pulmonary artery high pressure recognition method based on the multi-path electrocardiographic heart sound signals according to any one of claims 1-8; the screen is used for displaying the acquired data; the indicator lamp feeds back the working state of the device through different colors or flashing states, including power on, normal signal acquisition, abnormal data processing and recognition results; the function buttons comprise a power key and an acquisition start/stop key and are used for controlling the core operation flow of the device; The hub is a middle hub for signal transmission, is connected with 6 acquisition heads and a host, and is used for centralized receiving and stable transmission of multi-channel heart sounds and electrocardiosignals; The acquisition head is used for directly acquiring heart sounds and electrocardiosignals in a specific chest area and is a front end part for acquiring signals; the paper bin cover is internally provided with a protective cover of the printing module, and the printing paper can be replaced after the protective cover is opened, so that the printing output of the identification result is supported.

Description

Pulmonary artery high pressure identification device and method based on multipath electrocardiographic heart sound signals Technical Field The invention relates to the technical field of biomedicine, in particular to a pulmonary artery high pressure identification device and method based on multipath electrocardiographic heart sound signals. Background In the fields of biomedical engineering and health informatics, the multimodal fusion technology of electrocardiogram and phonocardiogram has been developed in recent years as a core means for noninvasive diagnosis of cardiovascular diseases. Along with the maturation of high-precision sensor technology and the iterative optimization of a digital signal processing algorithm, the synchronous acquisition and collaborative analysis capability of the multichannel physiological signals are greatly enhanced, and a richer physiological information support is provided for early screening of pulmonary arterial hypertension. Researchers gradually construct a pathological characterization framework based on time sequence analysis by integrating the electrical activity characteristics of electrocardios and the mechanical vibration characteristics of heart sounds. The technical path is pushing the PAH diagnosis to evolve from single-mode dependence to multi-mode dynamic fusion, the core breakthrough of the PAH diagnosis is concentrated on the improvement of the time sequence consistency of signal processing and the refinement of feature extraction, and an algorithm foundation is laid for accurate medical treatment. However, the prior art has key limitations in achieving multimodal fusion. Firstly, cross-channel time alignment of multi-channel electrocardio and heart sound signals generally depends on static timestamp matching, is easily interfered by equipment sampling rate difference, signal transmission delay and environmental noise, causes characteristic extraction time sequence distortion, and further weakens the stability and repeatability of PAH identification. Secondly, the dynamic modulation effect of the respiratory cycle on the heart sound splitting characteristic is often ignored, and the traditional method lacks a systematic construction mechanism of the respiratory phase track, and cannot effectively capture the physiological fluctuation of the inspiration/expiration stage, so that the deviation of the second heart sound splitting characteristic occurs under the influence of respiration, and the diagnosis sensitivity and specificity are reduced. Disclosure of Invention The present invention has been made in view of the above-described problems occurring in the prior art. Therefore, the pulmonary artery high pressure identification method based on the multipath electrocardiographic heart sound signals solves the problems of cross-channel time alignment deviation and two-heart sound splitting characteristic deviation. In order to solve the technical problems, the invention provides the following technical scheme: In a first aspect, the invention provides a pulmonary artery high pressure identification method based on multi-path electrocardiographic heart sound signals, which comprises the steps of receiving digital sampling streams of the multi-path electrocardiographic signals and the multi-path heart sound signals, and carrying out sliding window slicing and window boundary alignment in a unified time reference to generate a synchronous window sequence; according to the synchronous window sequence, main peak interval correction and abnormal interval elimination are carried out, beat-by-beat boundaries are obtained, multiple paths of electrocardiograph fragments and multiple paths of heart sounds fragments are intercepted, cross-channel envelope alignment is carried out to generate beat-by-beat alignment fragment sets, an electrocardiograph slow-change track and a heart sound envelope slow-change track are extracted from the beat-by-beat alignment fragment sets, respiratory phase tracks are constructed, phase tracking is carried out, an inhalation segment and an exhalation segment are obtained, aortic valve closing candidate peaks and pulmonary valve closing candidate peaks are positioned in the inhalation segment and the exhalation segment, relative time intervals of the candidate peaks are solidified, and meanwhile, a double-anchoring split fragment set is generated through robust convergence and rollback splicing, the falling points of the aortic valve closing candidate peaks and the pulmonary valve closing candidate peaks are limited in main peak anchor point time intervals and heart sound fragment key turning point time intervals of the double-anchoring split fragment sets, adjacent beat back substitution and multiple channels are carried out on the cross-by the candidate peaks, physiological consistency split fragment sets are generated, the split sequences of multiple paths of heart sounds are converged according to the physiological consistency split