KR102149748B1 - Method and apparatus for obtaining heart and lung sounds - Google Patents

Abstract

A method and apparatus for obtaining a cardiopulmonary sound signal are provided. The method is a method performed by a computer, comprising: acquiring a heart sound and/or lung sound according to a measurement position of a probe for measuring a patient's heart sound and/or lung sound, and analyzing the heart sound and/or lung sound according to the measurement position And determining a target measurement location of the probe, and providing a target measurement location of the probe.

Description

Heart and lung sound signal acquisition method and apparatus {METHOD AND APPARATUS FOR OBTAINING HEART AND LUNG SOUNDS}

The present invention relates to a method and apparatus for obtaining a cardiopulmonary sound signal.

Cardiopulmonary sound obtained by auscultation of the heart or lungs can be used as data for diagnosing a patient's disease. However, in the process of acquiring the cardiopulmonary sound, various signals may be mixed or the signal quality may vary depending on the location where the cardiopulmonary sound is measured. In this way, when the cardiopulmonary sound is mixed with noise or the signal quality is poor, there is a problem that the patient's health condition or disease cannot be accurately diagnosed.

Accordingly, there is a need for a method of obtaining cardiopulmonary sound that can be used as accurate analysis data. In addition, there is a need for a method of acquiring cardiopulmonary sound having an accurate signal pattern as pre-processed data and using it as various analysis indicators.

U.S. Patent No.8992435, 2015.03.31

The problem to be solved by the present invention is to provide a method and apparatus for obtaining a cardiopulmonary sound signal.

The problems to be solved by the present invention are not limited to the problems mentioned above, and other problems that are not mentioned will be clearly understood by those skilled in the art from the following description.

A method of acquiring a cardiopulmonary sound signal according to an embodiment of the present invention is a method performed by a computer, comprising: obtaining a heart sound and/or a lung sound according to a measurement position of a probe for measuring a patient's heart sound and/or lung sound, And determining a target measurement position of the probe by analyzing the heart sound and/or lung sound according to the measurement position, and providing a target measurement position of the probe.

In an embodiment of the present invention, the step of obtaining the heart sound and/or the lung sound includes determining at least one measurement position of the probe, and the heart sound and/or the heart sound from each of the determined at least one measurement position. It may include the step of obtaining a lung sound.

In an embodiment of the present invention, the determining of the target measurement position of the probe includes extracting a signal waveform of a heart sound and/or lung sound obtained from each of the at least one measurement position, the heart sound and/or the Analyzing the signal waveform of the lung sound to determine whether it is an appropriate signal waveform, and determining the target measurement position by deriving a measurement position of the probe corresponding to the appropriate signal waveform according to the determination result. .

In an embodiment of the present invention, the step of obtaining the heart sound and/or the lung sound may be obtained by separating the heart sound and the lung sound by filtering by using a difference in frequency bands of the heart sound and/or the lung sound, or Obtaining the heart sound by separating the heart sound into a first heart sound and a second heart sound using a change or magnitude of amplitude from the waveform of the heart sound and/or the lung sound signal, or by using a frequency band of the heart sound and/or the lung sound It may further include the step of separating and obtaining the lung sound into a left lung sound and a right lung sound.

In one embodiment of the present invention, the step of acquiring the heart sound and/or the lung sound includes acquiring the heart sound and/or the lung sound using a microphone, but according to the specific gravity of the heart sound and the lung sound to be acquired, The position can be adjusted.

In an embodiment of the present invention, the determining whether the signal is the appropriate signal waveform comprises analyzing at least one of a change in amplitude and a magnitude of amplitude from the signal waveform of the heart sound and/or the lung sound to determine whether it is a suitable signal waveform. can do.

In one embodiment of the present invention, the step of deriving the target measurement position of the probe and determining the target measurement position as a result of the analysis corresponds to the signal waveform of the heart sound and/or the lung sound determined as the most suitable signal waveform. The measuring position of the probe may be determined as the target measuring position.

In an embodiment of the present invention, in the providing of the target measurement location of the probe, the target measurement location may be compared with the current measurement location of the probe.

In one embodiment of the present invention, obtaining a heart sound and/or lung sound of the patient from a target measurement location of the probe, and constructing a data set using the heart sound and/or lung sound obtained from the target measurement location It may further include a step.

In an embodiment of the present invention, it may further include performing a feature-based analysis or a deep learning-based analysis based on the data set.

In one embodiment of the present invention, based on the patient's heart sound and/or lung sound obtained from the target measurement position of the probe, the pulse pressure is calculated using a correlation between a change in a ventricular systolic time interval and a change in pulse pressure. Alternatively, the method may further include calculating the single stroke amount by using a correlation between the change in the ventricular systolic time interval and the change in stroke volume.

A method for acquiring a cardiopulmonary sound signal according to another embodiment of the present invention is a method performed by a computer, comprising: acquiring a heart sound and/or lung sound from a probe, and whether an abnormal section is included by analyzing the heart sound and/or the lung sound. Determining whether or not, and when it is determined that the heart sound and/or the lung sound includes an abnormal section, correcting the signal waveform of the abnormal section based on a preset normal signal waveform.

In an embodiment of the present invention, the determining whether the abnormal interval is included comprises: acquiring learning data learned from a plurality of patients, and whether to include the abnormal interval based on the learning data It includes determining, and the learning data may be generated by learning based on deep learning by collecting heart sounds and/or lung sounds including abnormal sections from the plurality of patients.

In an embodiment of the present invention, in the step of correcting the signal waveform of the abnormal section, the signal waveform corresponding to the abnormal section may be removed and replaced with the preset normal signal waveform.

In an embodiment of the present invention, the step of correcting the signal waveform of the abnormal section may include, when it is determined that the abnormal section is included in the lung sound, removing the signal waveform of the lung sound to obtain a signal waveform of the heart sound, or When it is determined that the abnormal section is included in the heart sound, the signal waveform of the lung sound may be obtained by removing the signal waveform of the heart sound.

In one embodiment of the present invention, it may further include the step of monitoring the health state of the patient based on the heart sound and/or lung sound including the corrected signal waveform.

In one embodiment of the present invention, tracking a change in the signal waveform of the abnormal section from the heart sound and/or lung sound including the abnormal section, and the disease of the patient based on the change in the signal waveform of the abnormal section. It may further include predicting.

In an embodiment of the present invention, the step of obtaining the heart sound and/or the lung sound includes determining a target measurement position of the probe based on the acquired heart sound and/or lung sound according to the measurement position of the probe, And acquiring the heart sound and/or the lung sound from the target measurement position of the probe.

In an embodiment of the present invention, further comprising the step of measuring the signal quality of the heart sound and/or the lung sound, and determining whether the abnormal section is included, wherein the signal quality is a predetermined threshold value and When it is determined as an abnormal signal by comparison, the analysis of the heart sound and/or the lung sound may not be performed.

In an embodiment of the present invention, it may further include collecting heart and/or lung sounds including the corrected signal waveform, and constructing a data set using the collected heart and/or lung sounds. have.

In an embodiment of the present invention, it may further include performing a feature-based analysis or a deep learning-based analysis based on the data set.

In an embodiment of the present invention, the pulse pressure is calculated from the heart sound and/or lung sound including the corrected signal waveform using a correlation between the change in the ventricular systolic time interval and the change in the pulse pressure, or the ventricular systolic time It may further include calculating the one stroke amount by using a correlation between the change in the interval and the change in the stroke volume.

An apparatus for obtaining a signal of a heart sound and/or a lung sound according to another embodiment of the present invention includes an input unit receiving a signal of the heart sound and/or a lung sound from a probe, and the cardiopulmonary sound signal acquisition according to the embodiment of the present invention A processor performing the method, a memory in which a computer program implementing the method is stored, and an output unit for outputting monitored body information of a patient from signals of the heart sound and/or lung sound.

A computer program according to another embodiment of the present invention is stored in a medium in order to execute the method according to the embodiment of the present invention in a computer.

According to the present invention, it is effective to accurately determine a patient's health status and disease diagnosis by obtaining a cardiopulmonary sound with improved signal quality for each patient. In addition, cardiopulmonary sound of improved quality can be used as pre-processed data, and through this, it can be used as an analysis index to predict various biological signals.

The effects of the present invention are not limited to the effects mentioned above, and other effects not mentioned will be clearly understood by those skilled in the art from the following description.

1 is a flowchart illustrating a method of obtaining a cardiopulmonary sound signal according to an embodiment of the present invention.
2 is a diagram showing a signal waveform of a heart sound and/or a lung sound acquired according to a measurement position of a probe.
3 is a flowchart illustrating a method of obtaining a cardiopulmonary sound signal according to another embodiment of the present invention.
4 to 6 are diagrams for explaining an example of utilizing heart sounds and/or lung sounds obtained according to an embodiment of the present invention.
7 is a block diagram showing an apparatus for performing a method of acquiring a cardiopulmonary sound signal according to an embodiment of the present invention.

Advantages and features of the present invention, and a method of achieving them will become apparent with reference to the embodiments described below in detail together with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below, but may be implemented in a variety of different forms, only the present embodiments are intended to complete the disclosure of the present invention, It is provided to fully inform the technician of the scope of the present invention, and the present invention is only defined by the scope of the claims.

The terms used in the present specification are for describing exemplary embodiments and are not intended to limit the present invention. In this specification, the singular form also includes the plural form unless specifically stated in the phrase. As used in the specification, “comprises” and/or “comprising” do not exclude the presence or addition of one or more other elements other than the mentioned elements. Throughout the specification, the same reference numerals refer to the same elements, and “and/or” includes each and all combinations of one or more of the mentioned elements. Although "first", "second", and the like are used to describe various elements, it goes without saying that these elements are not limited by these terms. These terms are only used to distinguish one component from another component. Therefore, it goes without saying that the first component mentioned below may be the second component within the technical idea of the present invention.

Unless otherwise defined, all terms (including technical and scientific terms) used in the present specification may be used with meanings that can be commonly understood by those of ordinary skill in the art to which the present invention belongs. In addition, terms defined in a commonly used dictionary are not interpreted ideally or excessively unless explicitly defined specifically.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

1 is a flowchart illustrating a method of obtaining a cardiopulmonary sound signal according to an embodiment of the present invention. The method of Fig. 1 is performed by a computer. In the present specification, a computer is used to encompass a computing device including at least one processor. For example, the computing device may include a conventional computing device such as a desktop computer and a laptop computer, as well as a mobile computing device such as a smart phone, a tablet PC, a personal digital assistant (PDA), a notebook, or the like, or a non-mobile computing device. It may also include medical equipment that acquires or observes medical information such as medical images. It may also include a server, and may be composed of one or more devices. The computing device may be configured to include various hardware including a central processing unit (CPU), as well as various software such as an operating system (OS), a program for managing hardware including initialization, and an application program.

Referring to FIG. 1, a computer may acquire a heart sound and/or a lung sound according to a measurement position of a probe (S100).

Here, the probe is a device for measuring heart sound and/or lung sound from a patient, and may be a stethoscope. For example, a probe (eg, a stethoscope) may be inserted into the esophagus of a patient to obtain a heart sound and/or a lung sound at the inserted position. Of course, the heart sound and/or lung sound of the patient may be measured without the probe being inserted into the esophagus, and it may be obtained not only through the esophagus but also through other body parts. In addition, heart sounds and/or lung sounds may be obtained from the probe and then other types of acoustic signal processing may be performed.

In an embodiment, the computer may determine at least one measurement position of a probe for measuring a patient's heart sound and/or lung sound, and obtain a heart sound and/or lung sound from each of the determined measurement positions of the at least one probe. For example, while the probe is inserted into the esophagus of a patient, a plurality of measurement locations for measuring heart and/or lung sounds may be determined based on the insertion path. Alternatively, a plurality of measurement locations may be determined based on the path in the process of gradually removing the probe after being completely inserted into the esophagus of the patient.

In obtaining the heart sound and/or lung sound for each measurement position of the probe, the heart sound and the lung sound may be acquired from the probe at once. At this time, when the proportions of the heart sound and the lung sound to be acquired are different, the acquisition ratio of the heart sound and the lung sound may be adjusted using a microphone. For example, when measuring heart and lung sounds, if the position of the microphone and the number of microphones are adjusted, a difference occurs in the signal shape of the heart sound and lung sound, and the weight of the heart sound and lung sound is configured differently by using the difference between these signal types. can do. Here, the microphone may be configured by being connected to the probe or included in the probe itself. Alternatively, the microphone may be configured separately from the probe, and, if necessary, may be used together with the probe when acquiring heart and/or lung sounds.

According to an embodiment, the heart sound and the lung sound obtained at one time from a probe may be separately obtained by filtering using a difference in frequency band. In addition, the heart sound can be obtained by separating the heart sound into the first heart sound (S1) and the second heart sound (S2) using the amplitude or shape of the signal waveform, and the lung sound can be obtained by using the frequency band of the signal. It can also be obtained by separating it with.

The computer may determine the target measurement position of the probe by analyzing the acquired heart and/or lung sound according to the measurement position of the probe (S110).

In one embodiment, the computer extracts the signal waveform of the heart sound and/or lung sound acquired from each of at least one measurement position, and analyzes the signal waveform of each extracted heart sound and/or lung sound, and the signal of the heart sound and/or lung sound It can be determined whether the waveform is a suitable signal waveform. In this case, the computer may determine whether the heart sound and/or lung sound acquired at the corresponding measurement position is a suitable signal by using a change in amplitude and a magnitude of the amplitude from the signal waveform of the heart sound and/or lung sound. When the signal waveform of the heart sound and/or lung sound is determined to be an appropriate signal waveform according to the determination result, the computer can derive the measurement position of the probe corresponding to the appropriate signal waveform, and based on this, the target measurement position of the probe can be determined. . For example, the computer analyzes the signal waveform of the heart sound and/or lung sound at each measurement location acquired from the probe to determine the heart sound and/or lung sound having the most suitable signal waveform, and measures the probe corresponding to the most suitable signal waveform. The location can be determined as the target measurement location.

The computer may provide information on the target measurement position of the probe (S120). In one embodiment, the computer may output the target measurement position of the probe through the screen, and also measure the heart sound and/or lung sound through various methods of guiding the probe to be positioned at the target measurement position (eg, medical staff) Information can be provided. For example, the computer can guide the target measurement position compared to the current measurement position of the probe.

The computer finally acquires the patient's heart sound and/or lung sound from the probe at the target measurement location, and pre-processes it, and can be used in various ways. That is, the computer may construct a data set by collecting the patient's heart and/or lung sound acquired at the target measurement location, and use the data set of the heart and/or lung sound to build a learning model. When the learning model is constructed in this way, the computer may obtain the heart sound and/or lung sound measured at the target measurement location from the new patient, apply the same to the learning model, and identify and guide the health state of the new patient in real time. In an embodiment, a feature-based analysis or a deep learning-based analysis may be performed based on a heart sound and/or a lung sound acquired from a target measurement location. This will be described later in detail with reference to FIGS. 4 to 6.

Further, the computer may additionally apply the method of FIG. 3 to be described later on the patient's heart sound and/or lung sound from the probe at the target measurement position, and then the finally acquired heart sound and/or lung sound.

2 is a diagram showing a signal waveform of a heart sound and/or a lung sound acquired according to a measurement position of a probe. Each graph shown in FIG. 2 represents an electrocardiogram (top of each graph) and a cardiopulmonary sound signal (bottom of each graph) measured according to the position of a probe inserted into the patient's esophagus.

According to (a) of FIG. 2, the probe is inserted into the esophagus and positioned at a depth of 38 cm from the incisor (the first measurement position), and the measured first signal waveform is shown. In this case, the computer may determine whether the first signal waveform corresponds to a suitable signal based on a change in the amplitude or the amplitude of the first signal waveform. As shown in (a) of FIG. 2, it can be seen that the first heart sound is dominant in the first signal waveform, whereas the second heart sound is very faint and has weak intensity. Accordingly, the computer can determine that the first signal waveform is an inappropriate signal waveform.

According to (b) of FIG. 2, the probe is inserted into the esophagus and positioned at a depth of 32 cm from the maxillary tooth (a second measurement position) to show a second signal waveform measured. In this case, the computer may determine whether the second signal waveform corresponds to a suitable signal based on a change in the amplitude of the second signal waveform or the magnitude of the amplitude. As shown in (b) of FIG. 2, both the first heart sound and the second heart sound are derived from the second signal waveform, but it can be seen that the first heart sound has a large amplitude and superiority compared to the second heart sound. Accordingly, the computer can determine that the second signal waveform is an inappropriate signal waveform.

According to (c) of FIG. 2, the probe is inserted into the esophagus and positioned at a depth of 28 cm from the maxillary tooth (the third measurement position), and the measured third signal waveform is shown. In this case, the computer may determine whether the third signal waveform corresponds to a suitable signal based on a change in the amplitude or the amplitude of the third signal waveform. As shown in (c) of FIG. 2, it can be seen that both the first and second heart sounds are clearly derived from the third signal waveform, and relatively similar patterns in amplitude or change are shown. Accordingly, the computer can determine that the third signal waveform is a suitable signal waveform.

According to (d) of FIG. 2, the probe is inserted into the esophagus and positioned at a depth of 25 cm from the maxillary tooth (the fourth measurement position), and the measured fourth signal waveform is shown. In this case, the computer may determine whether the fourth signal waveform corresponds to a suitable signal based on a change in the amplitude of the fourth signal waveform or the magnitude of the amplitude. As shown in (d) of FIG. 2, although the relationship between the first heart sound and the second heart sound is derived in a similar pattern in the fourth signal waveform, it can be seen that the amplitude is reduced compared to the third signal waveform. Accordingly, the computer can determine that the fourth signal waveform is an unsuitable signal waveform compared to the third signal waveform.

That is, the computer acquires a signal waveform of the heart sound and/or lung sound (first to fourth signal waveform) for each measurement position of the probe (first to fourth measurement position), and the acquired signal waveform of the heart sound and/or lung sound By analyzing each of the (first to fourth signal waveforms), the most suitable signal waveform (eg, the third signal waveform) can be derived. In addition, the computer may determine the measurement position (third measurement position) of the probe corresponding to the derived most suitable signal waveform (eg, the third signal waveform) as the target measurement position. In other words, the computer may determine a position at a depth of 28 cm from the maxillary tooth as a target measurement position, and may provide position information about the position.

3 is a flowchart illustrating a method of obtaining a cardiopulmonary sound signal according to another embodiment of the present invention. The method of Fig. 3 is performed by a computer.

Referring to FIG. 3, the computer may acquire heart sounds and/or lung sounds from the probe (S200).

According to an embodiment, the computer may acquire heart sounds and/or lung sounds according to the method of FIG. 1. That is, the computer can determine the target measurement position of the probe based on the heart sound and/or lung sound acquired according to the measurement position of the probe, and finally obtain and use the heart sound and/or lung sound measured from the probe at the determined target measurement position. have.

The computer may analyze the acquired heart sound and/or lung sound to determine whether or not the abnormal section is included (S210).

In an embodiment, the computer may extract and store heart sounds and/or lung sounds including abnormal sections from a plurality of patients in advance, or learn the heart sounds and/or lung sounds based on deep learning to pre-build learning data. . Thereafter, the computer may determine whether or not an abnormal section exists in the heart sound and/or lung sound acquired in step S200 based on learning data capable of grasping the pattern of the abnormal section previously stored or constructed. Of course, a person (eg, a medical staff) may directly determine whether there is an abnormal section in the heart sound and/or lung sound.

If it is determined that the heart sound and/or the lung sound includes the abnormal section according to the determination result, the computer may correct the signal waveform of the abnormal section based on a preset normal signal waveform (S220).

In an embodiment, the computer may remove a section corresponding to the abnormal section from the signal waveform of the heart sound and/or lung sound, and replace a preset normal signal waveform in the removed section. The preset normal signal waveform may use a signal waveform determined as a normal section from the heart sound and/or lung sound acquired from the corresponding patient, or may use the signal waveform of the normal heart sound and/or lung sound acquired in advance from other patients. Alternatively, a reference signal waveform of a predetermined heart sound and/or lung sound may be used. For example, if there is a signal waveform part caused by a valve abnormality as a result of analyzing the heart sound and/lung sound obtained from a patient by a computer, it is recognized as an abnormal section and correction to remove the signal waveform part caused by a valve abnormality is performed. Can be done. In addition, by inserting signals of normal heart sound and/or lung sound into the removed part, it is possible to provide data that can accurately grasp changes before and after surgery or changes occurring during surgery.

In another embodiment, when it is determined that the lung sound includes an abnormal section from the signal waveform of the heart sound and/or the lung sound, the computer may perform correction in which all signals corresponding to the lung sound are removed and only the signal of the heart sound is obtained. For example, in the case of a patient with bronchiectasis, there may be a problem with the lung sound, but since there is no problem with the heart sound, a normal heart sound signal can be obtained by performing correction to remove only the lung sound. In this case, in the case of a patient with bronchiectasis, a high frequency band is generated according to the breathing cycle, so the computer may obtain a normal signal by performing correction to remove the high frequency band through filtering. Alternatively, when it is determined that the abnormal section is included in the heart sound from the signal waveform of the heart sound and/or the lung sound, the computer may perform correction in which all signals corresponding to the heart sound are removed and only the signal of the lung sound is obtained. For example, in the case that continuous heart noise occurs after the first heart sound in the signal waveform of the heart sound and/or lung sound, the computer may perform a correction to remove all the heart sound to obtain the lung sound signal, or a correction that removes only the heart noise section. It can also be used when analyzing the first heart sound by performing.

According to steps S200 to S220, it is possible to obtain a heart sound and/or lung sound signal in a normal state from which the abnormal section has been removed.

The computer may monitor the health status of the patient based on the heart sound and/or lung sound (signal waveform from which the abnormal section is removed) including the corrected signal waveform obtained in step S220.

In addition, the computer may track a change in a signal waveform for an abnormal section from the heart sound and/or lung sound including the abnormal section, and predict a patient's disease in real time based on the change in the signal waveform for the tracked abnormal section.

In addition, the computer may undergo a pre-processing process based on the heart sound and/or lung sound including the corrected signal waveform obtained in step S220, and use this in various ways. That is, the computer can obtain the normal heart sound and/or lung sound from which the abnormal section has been removed through correction, and build a data set based on this. These datasets can be used to create a learning model through future learning. For example, a feature-based analysis or a deep learning-based analysis may be performed. This will be described later in detail with reference to FIGS. 4 to 6.

According to an embodiment, the computer may measure the signal quality of heart and/or lung sounds. At this time, if the measured signal quality is compared with a preset threshold and determined as an abnormal signal, the computer determines whether the heart sound and/or lung sound contains an abnormal section from the time the abnormal signal is generated. Do not perform analysis on. For example, a signal of a heart sound and/or a lung sound may be measured through a signal quality measuring device (eg, BIS), and a signal quality index (SQI) may be measured as a signal quality of the measured signal. When the SQI value decreases below a preset threshold (e.g., when movement occurs during heart sound and/or lung sound measurement), the computer determines that a problem has occurred in the quality of the heart sound and/or lung sound, and the heart sound and/or Lung sound analysis may not be performed. Therefore, if the heart sound and/or lung sound contains abnormal sections or there is a quality problem, in the present invention, only normal signals can be obtained excluding these abnormal signals, so that more accurately the patient's health status or heart sound and/or lung sound You can get the results of the analysis.

Meanwhile, as described above, the heart sound and/or lung sound (first acquired heart sound and/or lung sound) obtained at the target measurement position, or the heart sound and/or lung sound obtained by preprocessing correction (the second acquired heart sound and/or lung sound) ) Can be used as learning data for feature-based analysis or deep learning-based analysis. In addition, it is possible to construct a learning dataset by performing learning in the form of supervised learning or unsupervised learning on these learning data, and analyzing and predicting the newly added first and second acquired heart sounds and/or lung sounds. can do.

For example, feature-based analysis can use supervised learning methods. When the feature-based analysis is applied, features for the first and second acquired heart sounds and/or lung sounds may be extracted from the learning data set learned by the user.

Deep learning-based analysis can use unsupervised learning methods. In the case of deep learning-based analysis, it is possible to construct a learning dataset using an unsupervised learning method by matching the first and second acquired heart sounds and/or lung sounds of a plurality of patients with invasively acquired reference data (eg Gold standard data). have. In addition, a correlation between specific reference data (eg, Gold standard data) and a specific factor in the first and second acquired heart sounds and/or lung sounds may be learned by applying the built deep learning-based learning data set.

In addition, as described above, when a learning model is constructed through a learning method such as supervised learning or unsupervised learning for the first and second acquired heart sounds and/or lung sounds of patients, the health status of the new patient Can predict change.

For example, when performing surgery on a new patient, first and second acquired heart sounds and/or lung sounds are acquired from the new patient, and applied to the learning model, the change in the patient's state can be recognized. In this case, the change in the patient's condition can be grasped only when data having the same conditions as the heart sound and/or lung sound used in the construction of the learning model are acquired.

In addition, if the patient's first and second acquired heart sounds and/or lung sounds are acquired and applied to the learning model, the data values that must be acquired invasively or non-invasively can be inferred, and through this, it is possible to grasp the change in the state of the critical patient during or after surgery I can.

In addition, it is possible to set a target measurement position before the patient's surgery, and obtain heart and/or lung sounds from this. Therefore, it is possible to determine whether there is a signal waveform requiring correction for the heart sound and/or lung sound acquired before the operation, perform correction, and then separate the heart sound and the lung sound. This can be separated based on the difference between the heart sound cycle and the lung sound cycle. Thereafter, a change in a factor (eg, STI, S2 amplitude, etc.) obtained from the separated heart sound data and lung sound data is applied to the learning model, so that the patient's condition can be grasped in real time.

4 to 6 are diagrams for explaining an example of utilizing heart sounds and/or lung sounds obtained according to an embodiment of the present invention.

The heart sound and/or lung sound acquired according to the embodiment of the present invention may be used as pre-processed data during feature-based analysis or deep learning-based analysis. As an example, in the case of feature-based analysis, the heart sound and/or lung sound acquired according to the embodiment of the present invention may be used as an index for predicting a vital signal that can be acquired invasively during an operation or in a post-operative process. In the case of deep learning-based analysis, it is possible to extract feature elements from heart and/or lung sounds acquired according to an embodiment of the present invention, and generate training data and learning models learned based on deep learning by inputting the extracted feature elements. have. Hereinafter, examples used as an index for predicting other data based on the heart sound and/or lung sound acquired according to an embodiment of the present invention will be described.

4 is a graph showing a ventricular systolic 300 and a pulse pressure 310 included in heart and/or lung sounds acquired according to an embodiment of the present invention. As shown in FIG. 4, it can be seen that there is a correlation between the change trend of the ventricular systolic phase 300 and the pulse pressure 310. Accordingly, the pulse pressure 310 may be calculated using the ventricular systolic 300 obtained according to an embodiment of the present invention. In addition, by deriving a systolic time interval (STI) from the heart sound obtained according to an embodiment of the present invention, a pulse pressure variation (PPV) may be determined.

5 is a graph showing a ventricular systolic 400 and a stroke volume (SV) 410 included in heart and/or lung sounds acquired according to an embodiment of the present invention. As shown in FIG. 5, it can be seen that there is a correlation between the ventricular systolic phase 400 and the SV 410. Accordingly, after calculating the ventricular systolic time interval 400 from the heart sound obtained according to an embodiment of the present invention, the SV 410 may be calculated by candidate-determining the ventricular systolic time interval pattern.

6 is a graph showing a second heart sound 500 and a systemic vascular resistance (SVR) 510 included in the heart sound and/or lung sound obtained according to an embodiment of the present invention. As shown in FIG. 6, it can be seen that there is a correlation between the second heart sound 500 and the SVR 510. Accordingly, the SVR 510 can be non-invasively predicted by performing post-processing on the amplitude of the second heart sound 500 after extracting the second heart sound 500 from the heart sound obtained according to an embodiment of the present invention.

In FIGS. 4 to 6, other data having a correlation with this through ventricular systolic time, systolic time interval (STI), and second heart sound (S2) among various characteristic factors included in the heart sound and/or lung sound Examples for predicting (eg, pulse pressure, change in pulse pressure, SV, SVR) have been described. This is described only as an example, and the present invention is not limited thereto.

That is, according to the present invention, when learning the heart sound and/or lung sound acquired from a patient through a learning model, various characteristic elements included in the heart sound and/or lung sound are extracted, and various characteristics having a correlation with the extracted characteristic elements Data can be derived. At this time, the derived data is invasive data obtained only when measured invasively, but in an embodiment of the present invention, even if only heart sound and/or lung sound is measured from a patient, invasive data can be derived through learning in a non-invasive method.

7 is a block diagram showing an apparatus for performing a method of acquiring a cardiopulmonary sound signal according to an embodiment of the present invention.

Referring to FIG. 7, the device 600 may include an input unit 610, a processor 620, an output unit 630, and a memory 640.

The input unit 610 may receive a heart sound and/or a lung sound from a probe (eg, a stethoscope), and the output unit 630 may output various information, eg, information on a target measurement location of the probe. In one embodiment, the output unit 630 may output the monitored patient's body information from the heart sound and/or lung sound that is performed by the processor 620 and obtained according to an embodiment of the present invention.

The processor 620 performs a method of obtaining a cardiopulmonary sound signal according to an embodiment of the present invention.

The memory 640 may store a computer program in which a method of acquiring a cardiopulmonary sound signal according to an embodiment of the present invention is implemented.

The steps of a method or algorithm described in connection with an embodiment of the present invention may be implemented directly in hardware, implemented as a software module executed by hardware, or a combination thereof. Software modules include Random Access Memory (RAM), Read Only Memory (ROM), Erasable Programmable ROM (EPROM), Electrically Erasable Programmable ROM (EEPROM), Flash Memory, hard disk, removable disk, CD-ROM, or It may reside on any type of computer-readable recording medium well known in the art to which the present invention pertains.

In the above, embodiments of the present invention have been described with reference to the accompanying drawings, but those of ordinary skill in the art to which the present invention pertains can be implemented in other specific forms without changing the technical spirit or essential features. You can understand. Therefore, the embodiments described above are illustrative in all respects, and should be understood as non-limiting.

Claims (24)

As a method performed by a computer,
Obtaining heart and/or lung sounds from the probe;
Analyzing the heart sound and/or the lung sound to determine whether an abnormal section is included; And
And when it is determined that the heart sound and/or the lung sound includes an abnormal section, correcting the signal waveform of the abnormal section based on a preset normal signal waveform. delete delete delete delete delete delete delete delete delete delete The method of claim 1,
The step of obtaining the heart sound and/or the lung sound,
Acquiring a heart sound and/or a lung sound according to the measurement position of the probe;
Determining a target measurement location of the probe by analyzing heart and/or lung sounds according to the measurement location;
Providing a target measurement position of the probe; And
And acquiring a heart sound and/or a lung sound from the target measurement position of the probe. The method of claim 1,
The step of determining whether the abnormal section is included,
Acquiring learning data learned from a plurality of patients; And
And determining whether to include the abnormal section based on the learning data,
The training data,
The heart and lung sound signal acquisition method, characterized in that generated by learning based on deep learning by collecting heart sounds and/or lung sounds including abnormal sections from the plurality of patients. The method of claim 1,
The step of correcting the signal waveform of the abnormal section,
A method of obtaining a cardiopulmonary sound signal, comprising removing a signal waveform corresponding to the abnormal section and replacing it with the preset normal signal waveform. The method of claim 1,
The step of correcting the signal waveform of the abnormal section,
When it is determined that the abnormal section is included in the lung sound, the signal waveform of the heart sound is obtained by removing the signal waveform of the lung sound, or
When it is determined that the abnormal section is included in the heart sound, the signal waveform of the lung sound is obtained by removing the signal waveform of the heart sound. The method of claim 1,
And monitoring a patient's health status based on the heart sound and/or lung sound including the corrected signal waveform. The method of claim 1,
Tracking a change in the signal waveform of the abnormal section from the heart sound and/or the lung sound including the abnormal section; And
Predicting a patient's disease based on a change in the signal waveform of the abnormal section. delete The method of claim 1,
Further comprising the step of measuring the signal quality of the heart sound and/or the lung sound,
The step of determining whether the abnormal section is included,
When the signal quality is determined as an abnormal signal by comparing with a preset threshold value, analysis of the heart sound and/or the lung sound is not performed. The method of claim 1,
Collecting heart sounds and/or lung sounds including the corrected signal waveform; And
And constructing a data set using the collected heart and/or lung sounds. The method of claim 20,
And performing a feature-based analysis or deep learning-based analysis based on the data set. The method of claim 1,
The pulse pressure is calculated from the heart sound and/or the lung sound including the corrected signal waveform using a correlation between the change in the ventricular systolic time interval and the change in the pulse pressure, or
Or calculating the one-time stroke amount by using a correlation between the change in the ventricular systolic time interval and the change in stroke volume. An apparatus for acquiring a signal of heart sound and/or lung sound, comprising:
An input unit receiving a signal of the heart sound and/or lung sound from a probe;
A processor for performing the method of obtaining a cardiopulmonary sound signal according to any one of claims 1, 12 to 17, and 19 to 22;
A memory storing a computer program implementing the method; And
Device comprising an output unit for outputting the body information of the patient monitored from the signal of the heart sound and/or lung sound. A computer program stored in a medium to execute the method of any one of claims 1, 12 to 17, and 19 to 22 on a computer.

Images