JP2013034670A - Cardiac sound information processor, cardiac sound information processing method, and cardiac sound information processing program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a cardiac sound information processor capable of accurately detecting all characteristic parts from cardiac sound signals without omission, without being affected by noise or the like.SOLUTION: The cardiac sound information processor includes: a peak position detection part 311 for detecting a plurality of peak positions from the cardiac sound signals for which sound of the heart is collected; an accuracy determination part 312 for determining whether or not the detection of the peak positions by the peak position detection part 311 is accurate; and a correction processing part 313 for performing correction processing of correcting the detection of the peak positions when it is determined that the detection of the peak positions by the peak position detection part 311 is not accurate by determination by the accuracy determination part 312. The accuracy determination part 312 determines the accuracy of the peak position detection on the basis of the result of clustering the peak positions on the basis of a time length between the detected peak positions. The correction processing part 313 repeatedly executes the correction processing according to a determination result by the accuracy determination part 312.

Description

  The present invention relates to a technique for analyzing a heart sound and detecting a characteristic part from the heart sound.

  Various information is contained in the heart sound, and the doctor diagnoses the presence and type of abnormality by listening to the heart sound using a stethoscope. However, diagnosis using a stethoscope requires special skills and training, and it is extremely difficult for ordinary people to perform (for example, Non-Patent Document 1). Therefore, if it is possible to analyze heart sounds using a computer and determine the presence or absence and type of abnormality, it is considered that this can be used to expand home medical care.

  Here, in Patent Document 1 (Japanese Patent Publication No. 2004-529720), an audio signal from the heart is recorded to detect a peak position, and the peak position is classified into various heart sounds and heart noises by cluster analysis. A system is described in which the presence / absence of an abnormality and the type thereof are determined from combinations of types of heart sounds and heart noises included in the audio signal.

  Patent Document 2 (Japanese Patent Publication No. 2009-535106) describes a device that assists clinicians in diagnosis by displaying the results of analyzing sound signals from the heart on the screen.

Special Table 2004-529720 Special table 2009-535106

Auscultation training by CD <heart sound edition> 2nd revised edition, Toshimi Sawayama (Author), Nanedo; (1994/02)

When heart sounds are recorded in an actual environment, the effects of noise and the like cannot be avoided. Even if such a heartbeat with noise is simply subjected to frequency analysis, it is greatly influenced by the noise, and as a result, not all correct peak positions can be detected without omission.
In Patent Document 1, all types of heart sounds are detected and classified by a single clustering, but in reality, the influence of noise cannot be ignored.
Noise reduction may be possible if a specially crafted operator with special expertise performs special filter adjustments, etc., but if such complicated operations are required, home medical care for ordinary people Can't contribute.

Further, the apparatus of Patent Document 2 displays the result of analyzing the heart sound on the screen, and the heart sound is displayed as it is as a time waveform. If you have medical expertise, you will be able to see the meaning of the heart waveform.
However, for ordinary people, even if only a complex waveform of heart sounds is displayed on the screen, it is difficult to judge what is normal and what is abnormal. Therefore, a user who does not have specialized knowledge regarding medical care cannot read the content displayed on the screen and cannot grasp the physical condition of the subject.

Accordingly, an object of the present invention is to provide a heart sound information processing apparatus capable of accurately detecting all correct peak positions without omission even from a heart sound signal including noise.
A second object of the present invention is to provide a heart sound information processing apparatus that displays the characteristics of heart sounds in an easy-to-understand manner so that even people without medical knowledge can understand.

The heart sound information processing apparatus of the present invention
A peak position detector for detecting a plurality of peak positions from a heart sound signal obtained by collecting heart sounds;
An accuracy determination unit that determines whether or not the detection of the peak position by the peak position detection unit is accurate;
A correction processing unit that performs correction processing to correct the detection of the peak position when it is determined by the determination by the accuracy determination unit that the detection of the peak position by the peak position detection unit is not accurate,
The accuracy determination unit determines the accuracy of peak position detection based on a result of clustering peak positions based on a time length between detected peak positions,
The correction processing unit repeatedly performs correction processing according to a determination result by the accuracy determination unit.

In the present invention,
The accuracy determination unit
It is preferable to determine the accuracy of peak position detection based on a value indicating the variation in the histogram of each cluster.

In the present invention,
The accuracy determination unit
It is preferable to determine the accuracy of peak position detection based on the difference or ratio of the number of elements in each cluster.

In the present invention,
The accuracy determination unit
It is preferable to determine the accuracy of peak position detection based on whether the clusters to which adjacent peak positions belong are the same or different.

In the present invention,
It is preferable that an upper limit number is set for the accuracy determination by the accuracy determination unit and the correction processing by the correction processing unit.

In the present invention,
The correction processing unit
When two or more peak positions that are not mode values of the elements of each cluster are consecutive, peak positions whose time intervals before and after both deviate from the mode value of the cluster may be deleted as false detection peak positions. preferable.

In the present invention,
The correction processing unit
When two or more peak positions belonging to the same cluster are continuous and the peak positions that are not the mode values of the elements of each cluster are adjacent to each other before and after that, the same adjacent to the peak position that is not the mode value It is preferable to delete the peak positions belonging to the cluster.

In the present invention,
The correction processing unit
It is preferable to re-search the peak position in a section between the peak position having a longer time length than the mode value of the cluster having the longer time length and the next peak position.

In the present invention,
It is preferable that the image processing apparatus further includes a feature location detection unit that further detects a characteristic location of the heart sound based on the detected peak position after the accuracy determination unit determines that the peak position detection is accurate.

In the present invention,
A display information generating unit for generating display information for displaying the characteristic location of the heart sound detected by the peak position detection unit and the characteristic location detection unit on a display unit;
The display information generation unit, a shape determination unit for classifying the shape of heart noise when the heart sound element is heart noise,
It is preferable to include a schematic unit that generates schematic display information based on the classification result.

In the present invention,
Preferably, the shape determination unit classifies the shape of the heart noise based on a differential coefficient of the heart sound in the heart noise section.

In the present invention,
It is preferable that the display information generation unit further classifies the heart noise based on a relative positional relationship between the heart noise and the I sound / II sound of the heart sound.

In the present invention,
It is preferable that the schematic unit further generates schematic display information based on a shape classification of heart noise and a volume of heart noise.

In the present invention,
Furthermore, it is preferable to further include a heart sound classification unit that classifies the feature locations of the heart sounds detected by the peak position detection unit and the feature location detection unit into normal sounds and abnormal sounds.

In the present invention,
The heart sound classification unit further classifies normal sound and abnormal sound by adding at least one of the subject's age information, the subject's body shape information, and the information of the auscultation position when auscultating the subject to the judgment criteria. Is preferred.

In the present invention,
It is preferable that the display information generation unit counts the number of detected feature points of the heart sound for each type.

In the present invention,
The display information generation unit includes at least one piece of information on the type, size, length of sounding time, and shape of the heart and the positional relationship with the sound I and sound II for each feature location of the heart sound. It is preferable to set a coefficient based on the above and perform counting using this coefficient.

In the present invention,
The schematic unit preferably performs color coding for each type of heart sound and heart noise.

In the present invention,
When the display information generating unit reproduces and outputs the heart sound, it is preferable that the display unit displays a reproduction part indicating symbol indicating a characteristic part of the heart sound corresponding to the heart sound being reproduced.

In the present invention,
The display information generation unit preferably displays a symbol indicating the repeated section on the display unit when the specific section of the heart sound is repeatedly reproduced and output.

In the present invention,
The display information generation unit preferably displays a description of the feature location of the heart sound being displayed when displaying the feature location of the heart sound on the display unit.

In the present invention,
When the display information generating unit reproduces and outputs a heart sound, the display information generating unit preferably displays on the display unit a schematic diagram schematically illustrating a characteristic portion of the heart sound corresponding to the heart sound being reproduced.

The heart sound information processing method of the present invention comprises:
Detecting a plurality of peak positions from a heart sound signal obtained by collecting heart sounds;
Clustering the peak positions based on the time length between the detected peak positions;
Determining the accuracy of peak position detection based on the clustered results;
Correcting the detection of the peak position when the determination of the accuracy determines that the detection of the peak position is not accurate, and
The step of correcting the detection of the peak position is repeatedly performed according to the accuracy determination result.

The heart sound information processing program of the present invention includes:
Computer
A peak position detector for detecting a plurality of peak positions from a heart sound signal obtained by collecting heart sounds;
An accuracy determination unit that determines the accuracy of peak position detection based on the result of clustering the peak positions based on the time length between detected peak positions,
According to the determination result by the accuracy determination unit, the correction unit is configured to function as a correction processing unit that repeatedly executes correction processing for correcting detection of peak positions.

In the present invention, after the peak position is detected from the heart sound signal by the peak position detection unit, the accuracy determination and the correction process are repeatedly executed.
As a result, even if the heart sound signal includes noise or the like, it is possible to correct the erroneous detection or the omission of detection by repeated correction processing and accurately detect the peak position regardless of the influence of the noise.
Such signal analysis makes it possible to accurately detect the characteristics of the heart sound, so that the physical condition of the subject can be grasped, for example, by home medical care without relying on special medical knowledge or skill.

The figure which shows the structure of 1st Embodiment which concerns on a heart sound information processing apparatus. 6 is a flowchart illustrating a processing procedure according to the first embodiment. The figure which shows the example of a threshold value calculation in peak position detection. The figure which shows the example of a threshold value calculation in peak position detection. 6 is a flowchart showing a processing procedure of iterative correction processing (ST130) in the first embodiment. 6 is a flowchart showing a processing procedure of iterative correction processing (ST130) in the first embodiment. The figure which shows the example of detection of a peak position. The figure which shows an example of a clustering result. The figure which shows an example of a clustering result. The figure for demonstrating the deletion conditions of a false detection peak position. The figure for demonstrating the re-search condition of a detection omission peak position. The figure which shows the example of a detection of II sound division. The figure which shows the example of a III sound detection. The figure which shows the example of a detection of IV sound. The figure which shows the example of a detection of click sound. The figure which shows the example of a heart noise detection. The figure which shows the structure of 2nd Embodiment. 10 is a flowchart showing a processing procedure of the second embodiment. 9 is a flowchart showing a processing procedure of a display information generation step (ST200) in the second embodiment. The figure which shows the shape pattern of a heart murmur. 9 is a flowchart showing a processing procedure of heart noise shape determination (ST235) in the second embodiment. 9 is a flowchart showing a processing procedure of heart noise shape determination (ST235) in the second embodiment. The figure which shows the "increase / decrease switching part" of heart noise. The figure which shows the example of schematic. The figure which shows the example of schematic. The figure which shows the example of schematic. The figure which shows the example of a display screen. The figure which shows the structure of 3rd Embodiment. 10 is a flowchart showing a processing procedure of the third embodiment. 9 is a flowchart showing a processing procedure of heart sound classification (ST150) in the third embodiment. 9 is a flowchart showing a processing procedure of a display information generation step (ST260) in the third embodiment. The figure which shows an example of a screen display. The figure which shows an example of a screen display. The figure which shows an example of a screen display.

Embodiments of the present invention will be described below with reference to the drawings.
An embodiment of a heart sound information processing apparatus according to the present invention will be described below. First, as a first embodiment, a configuration and processing steps for correctly detecting a peak position from a heart sound signal will be described.
Subsequently, as a second embodiment, a configuration and processing steps for displaying features of heart sounds in an easy-to-understand manner will be described. The third embodiment is a modification of the second embodiment.

(First embodiment)
FIG. 1 is a diagram showing a configuration of a first embodiment according to the heart sound information processing apparatus of the present invention.
The heart sound information processing apparatus 100 includes a signal input unit 110, an arithmetic processing unit 120, a display unit 150, and an input unit 160.

The arithmetic processing unit 120 includes a frequency analysis unit 200, a heart sound detection unit 300, and a display information generation unit 400.
The frequency analysis unit 200 includes a target signal extraction unit 210 as its internal element.

The heart sound detection unit 300 includes an I sound / II sound detection unit 310 and a characteristic location detection unit 370.
The I / II sound detection unit 310 includes a peak position detection unit 311, an accuracy determination unit 312, and a correction processing unit 313.

  The feature location detection unit 370 includes an II sound division detection unit 320, a III sound detection unit 330, an IV sound detection unit 340, a click sound detection unit 350, and a heart noise detection unit 360.

  The operation of each functional unit will be described later with reference to flowcharts and examples.

In addition, each functional unit of the arithmetic processing unit 120 may be configured by dedicated hardware including various logic elements or the like, or each heartbeat information processing program is incorporated in a computer having a CPU, ROM, and RAM. It may be operated as a functional unit.
To install the program, a memory card, CD-ROM or the like may be directly inserted into the computer, or a device for reading these storage media may be connected externally. Furthermore, the program may be supplied to the computer and installed by wired or wireless communication. It should be noted that the program supplied by various recording media, communication means, and the like is only required to include the heart sound information processing program according to the present invention, and may include a program for performing other control.

The operation of the heart sound information processing apparatus in this embodiment will be described with reference to the flowcharts of FIGS.
A heart sound signal collected by a predetermined microphone or the like is input to the signal input unit 110 and sequentially input from the signal input unit 110 to the frequency analysis unit 200.

When the heart sound signal is input to the frequency analysis unit 200, the frequency analysis unit 200 extracts the analysis target signal s (ST100). That is, the frequency analysis unit 200 divides the input signal into frames having a predetermined time length (for example, 30 ms), and converts the input signal into the frequency domain for each frame. Then, the sum of power in a predetermined band (0 to 300 Hz) is calculated for each frame. These values are sequentially extracted as the analysis target signal s by the target signal extraction unit 210 (ST101).
The target signal s extracted in this manner is sent to the heart sound detection unit 300 as the analysis target signal s.

  Here, the time length of the frame is 30 ms as an example, but it is needless to say that the time length is not limited thereto. Moreover, the sum of the power of each frequency component of 0 to 300 Hz is mainly extracted and used as the analysis target signal s. However, the analysis target signal s does not necessarily use the above-described frequency component (0 to 300 Hz). Alternatively, a plurality of frequency components may be used.

In heart sound detection unit 300, first, I sound / II sound detection unit 310 detects I sound / II sound from the target signal (ST110).
The I sound / II sound detection step (ST110) includes peak position detection (ST120) and repetitive correction processing (ST130).

In the peak position detection (ST120), the peak position detection unit 311 detects the peak position from the analysis target signal s.
Here, there are a plurality of methods for detecting the peak position, and these can be used alone or in combination.
Below, the example of the method of detecting a peak position is given.

(First method to detect peak position)
As a first method for detecting the peak position, a frame having a maximum value is obtained based on the differential coefficient, and this is used as the peak position.
For example, the differential coefficient is calculated by smoothing the analysis target signal s using a known Savitzky-Golay method. Then, a frame having a maximum value is obtained from the differential coefficient and detected as a peak position.

Of course, the smoothing method may be other than the Savitzky-Golay method. Furthermore, the smoothing process itself may not be performed. The detection method of the maximum value frame is not limited.
For example, a power difference may be calculated for each adjacent frame of the analysis target signal s to obtain a differential coefficient, and a maximum value of the coefficient may be detected as a peak position.

(Second method to detect peak position)
As a second method for detecting the peak position, a frame having a power larger than a predetermined threshold is used as the peak position. For example, the average value AVE and the standard deviation SD of the power are obtained for every predetermined number of frames. Then, a product SD ′ (= C × SD) of the predetermined value C and the standard deviation SD is obtained.
The sum (= AVE + SD ′) of the average value AVE and the product SD ′ is used as a threshold value, and a frame having a power larger than this sum is detected as a peak position.

The threshold may be determined based on a power histogram as follows. That is, as shown in FIGS. 3A and 3B, a power histogram is obtained for each predetermined number of frames of the analysis target signal s. Then, a threshold value is determined based on this histogram.
FIG. 3A is an example in which a value obtained by dividing the area of the histogram by a predetermined ratio is used as the threshold value.
FIG. 3B is an example in which the threshold value is a value at which the histogram has a predetermined slope.

The threshold value determination method based on the histogram is not limited to the above example.
The “predetermined number of frames” may be the number of frames corresponding to several seconds or the total number of frames, and the number of frames is not limited.

Here, the first method for detecting the peak position and the second method for detecting the peak position may be used in combination.
For example, the peak position is a frame that is the maximum value obtained by the first method and exceeds the threshold value obtained by the second method, but the method of using the combination is not limited to this.

  Information on the peak position detected in this way is sent to the accuracy determination unit 312.

Once the peak position has been detected, the iterative correction process (ST130) for correctly correcting the peak position detection is next executed.
Details of the iterative correction process (ST130) will be described with reference to the flowcharts of FIGS. 4A and 4B.
As shown in the flowcharts of FIGS. 4A and 4B, the iterative correction process (ST130) includes an accuracy determination (ST131) and an error correction process (ST140).
This will be specifically described below.

First, referring to FIG. 4A, accuracy determination (ST131) is performed.
In the accuracy determination (ST131), first, a peak interval is calculated (ST132). That is, for each peak position, a time interval (ms) from the next peak position is calculated and set as the time interval of each peak position.
For example, it is assumed that the peak position is detected as shown in FIG.
The vertical thick line in the figure is the detected peak position.
At this time, the length of the double-ended arrow on the right side of each peak position is the time interval. That is, for the peak position P1, the time interval t1 to the peak position P2 next to P1 is taken.
Similarly, for the peak position P2, a time interval t2 to the peak position P3 next to P2 is taken.
Hereinafter, the “time interval between peak positions” indicates the time interval with the next peak position.

Next, in ST133, the calculated time intervals are clustered and classified into two clusters. An example of a clustering method is a known k-means method, but other clustering methods may be used.
FIG. 6 shows an example of the clustering result.
The clustering result can be visualized as shown in FIG. 6 using a time interval histogram.
In FIG. 6, class elements 511 and 521 with short time intervals on the left side of the alternate long and short dash line are elements belonging to cluster 1, and similarly, class elements 522, 512 and 513 with long time intervals on the right side of the alternate long and short dash line are included. It is an element belonging to cluster 2.

  Next, in ST134, the degree of variation is obtained for each cluster. That is, the degree of variation of the histogram of cluster 1 is obtained, and the degree of variation of the histogram of cluster 2 is obtained.

  Here, typical examples of the degree of variation include taking a variance value, but in addition, the difference between the average value and the median value, the difference between the maximum value and the minimum value, and the frequency of the histogram Any value can be used as long as it indicates the variation of the histogram for each cluster, such as the number of classes of 1 or more.

Next, in ST135, each obtained variation degree is compared with a predetermined threshold value, and it is determined whether or not each variation degree is a predetermined value or more.
For example, in FIG. 6, a plurality of classes exist in each cluster.
Although each cluster should have only one class in nature, multiple clusters are included in one cluster because the noise is picked up as a peak position. There is a high possibility that the position is not detected.
In this case, it is necessary to delete the erroneously detected peak position or to search again for the detection-missing peak position.
Here, when a plurality of classes exist in each cluster, the degree of variation obtained for each cluster increases.
On the other hand, as shown in FIG. 7, when the elements are distributed in only one class in each cluster, it is considered that there is no error in the detection of the peak position. In this case, the degree of variation for each cluster is small. Therefore, it can be determined whether or not the peak position is correctly detected based on the variation degree.

When the variation degree for each cluster is less than a predetermined value (ST135: NO), a difference in the number of elements between the clusters is further obtained (ST136), and this difference is compared with a predetermined threshold (ST137).
If the I and II sounds are correctly detected in the peak position detection, the number of elements in each cluster should be equal. Therefore, when there is a large difference in the number of elements between clusters, it is considered that the peak position is not correctly detected.
For example, when the number of elements in cluster 1 is less than half of the number of elements in cluster 2 or more than twice, it can be determined that peak position detection is not correct.

Further, if the difference in the number of elements between the clusters is less than the predetermined threshold (ST137: NO), the cluster to which the continuous peak position belongs is determined (ST138).
If the I and II sounds are correctly detected in the peak position detection, the clusters to which the continuous peak positions belong should be alternated. Therefore, if consecutive peak positions exist in the same cluster (ST138: YES), it is considered that the peak positions are not correctly detected.

Whether the variation degree is greater than or equal to a predetermined value (ST135: YES), whether the difference in the number of elements between clusters is greater than or equal to a predetermined value (ST137: YES), or if consecutive peak positions are included in the same cluster (ST138: YES) The processing shifts to error correction processing (ST140) by the processing unit 313.
In the error correction process (ST140), first, one element (peak position) to be examined is selected (ST141). Then, the time interval to be examined is compared with the mode value of cluster 2 (the cluster having the longer time interval) (ST142). This is because the necessary correction processing differs depending on whether the time interval to be examined is longer or shorter than the mode value of cluster 2.

  When the time interval to be examined is shorter than the mode value of cluster 2 (ST142: YES), there is a possibility that there is a peak position erroneously detected in the vicinity, and this is deleted. That is, whether or not to delete the peak position is determined based on the peak position whose time interval is shorter than the mode value of cluster 2 and the time intervals of the surrounding peak positions.

This will be described with reference to FIG.
In FIG. 8, a peak position P631, a peak position P632, and a peak position P633 are peak positions that are erroneously detected between correct peak positions.
Here, the immediately preceding peak positions P641, P642, and P643 are considered. Then, even if these are detected correctly, the next peak position is detected in an extra place, so when calculating the "time interval" of the peak positions P641, P642, P643, these time intervals Is out of the mode of the cluster.

Therefore, when two or more peak positions deviating from the mode value of the cluster continue, the first deletion condition is that a peak position whose time interval before and after deviates from the mode value of the cluster is regarded as a false detection peak position. To delete.
For example, when focusing on the peak position P632, the time interval t632 of the peak position P632 deviates from the mode value of the cluster.
Furthermore, the time interval t642 of the peak position P642, which is the previous time interval, also deviates from the mode value of the cluster. Therefore, the peak position P632 is deleted as an erroneously detected peak position.
If judged similarly, the peak position P631 and the peak position P633 are also to be deleted.

Consider a case where the peak position P634 is erroneously detected between the correct peak position P644 and the peak position P645.
Furthermore, it is assumed here that the time interval t634 of the erroneously detected peak position P634 and the time interval t645 of the next correct peak position P645 are coincidentally coincident. In this case, the time interval t634 of the erroneously detected peak position P634 is the same as the time interval t645 of the correct peak position P645, which means that the time interval t634 of the erroneously detected peak position P634 is included in the mode value of the cluster. It will be.
On the other hand, the time interval t644 of the correctly detected peak position P644 is out of the mode value of the cluster.
In this case, the time interval t644 of the peak position P644 is out of the mode value of the cluster, but the peak position P634 should be deleted.

Therefore, when the peak positions belonging to the same cluster are consecutive before and after the peak position deviating from the mode value of the cluster (referred to as PeakErr), the second deletion condition is that PeakErr of the continuous peak positions is set to PeakErr. The adjacent peak position is deleted.
In the case of FIG. 8, the peak position P644 is PeakErr, and the peak position P634 and the peak position P645 belong to the same cluster behind the peak position P644.
Therefore, the peak position P634 adjacent to the peak position P644 that is PeakErr is set as the deletion target.
As a result, the false detection peak position P634 is deleted.

Here, when the peak positions are deleted according to the first deletion condition and the second deletion condition, the correct peak positions of the I sound and the II sound may be deleted.
However, in this embodiment, since the peak position is re-searched after this and repeated error correction processing is performed, the exact positions of the I and II sounds can be finally detected without omission. ,No problem.
Although there is a problem that the number of repetitive processes increases, the advantage that accurate heart sound analysis that is not affected by noise can be reliably achieved by reliably deleting the false detection peak position is much more beneficial.
Further, the division of the II sound and the III sound and the IV sound are also temporarily deleted as false detection peaks. However, in this embodiment, there is no problem because the peak position is further searched after this.

  In the present embodiment, the first deletion condition and the second deletion condition described above are exemplified, but other deletion conditions may be used.

On the other hand, in ST142, when the time interval of the selected element is equal to or greater than the mode value of cluster 2 (ST142: NO), there may be a peak position leaking from the detection in this vicinity. Perform (ST144).
This will be described with reference to FIG.
The peak position re-search (ST144) is executed between a peak position having a time interval equal to or greater than the mode value of cluster 2 and the next peak position.
When there is a detection omission, the time interval before and after the peak position that was not detected becomes longer.

  For example, assume that peak positions P651, P652, P653, and P654 are not detected in FIG. Then, the time interval t661 of the peak position P661, the time interval t663 of the peak position P663, and the time interval t665 of the peak position P665 become longer. Accordingly, the peak position is searched again in the section between the peak position P661, the peak position P663, the peak position P665, and the peak position immediately after these.

For the peak position re-search (ST144), the first method and the second method for detecting the peak position described in the peak position detection step of ST120 may be used.
However, when the second method is used, the “predetermined number of frames” is a frame within the search range.
Also, the conditions may be changed from those in ST120.
For example, the length of smoothing may be changed in the first method.
In calculating the threshold value in the second method, the value of the predetermined value C multiplied by the standard deviation SD may be changed.
Further, in the second method, when the threshold value is calculated based on the power histogram, the ratio value for dividing the area of the histogram may be changed.

  Alternatively, the search range may be narrowed based on the cluster to which the front and rear peak positions belong. For example, the peak position P662 in front of the peak position P663 in FIG. Since the sound I and sound II should alternate, it can be estimated that the peak position P663 belongs to the cluster 2. Accordingly, the search is limited to the vicinity of the peak position P663 with the aim that there should be a correct peak position in the vicinity of the position of the mode 2 time interval separated from the mode 2.

After deleting the false detection peak position (ST143) or re-searching the detection omission peak position (ST144), determine whether error correction processing has been performed for all elements (ST145). If there are elements that have not been corrected yet (ST145: NO), returning to ST141, the correction process for the next element is executed.
When error correction processing has been performed for all elements (ST145: YES), the process returns to the beginning of accuracy determination (ST131) once again and calculation is performed from the calculation of the peak interval (ST132). That is, the determination of ST135, ST137, and ST138 is performed again. If all the determination conditions of ST135, ST137, and ST138 have been cleared, it is determined that the I / II sound has been correctly detected, and the process proceeds to the next step.
In this way, by repeating the error correction process (ST140) until all the determination conditions of ST135, ST137, and ST138 are cleared, detection omissions and false detections are corrected, and peak positions of I and II sounds are accurately detected. be able to.

  In addition, although the case where error correction processing (ST140) is repeated until all the determination conditions of ST135, ST137, and ST138 are cleared is given as an example, an upper limit is set for the number of executions of error correction processing (ST140). If the value reaches, the repeated correction process (ST130) may be terminated.

When the I and II sounds can be accurately detected without omission, other heart sounds are detected based on the peak positions of these I and II sounds. That is, first, in ST150, II sound division detection unit 320 detects II sound division.
The detection of II sound splitting (ST150) is performed based on the already detected I and II sounds.
In FIG. 10, P701 to P705 indicated by broken lines are examples of the division of the II sound and are often heard immediately after the II sound. Thus, for example, searching for the peak position in the section from the position of the II sound to 0.12 seconds later can be cited as an example, but the length of such a search section can be appropriately set as necessary. Alternatively, the search interval may be calculated each time based on the time interval between the sound I and the sound II without fixing the time of the search interval.

  If the peak position is detected in this search, it is determined as “II sound split”. As a peak position search method for detecting II sound splitting, a method similar to the peak position re-search described in ST144 may be used.

Next, in ST160, the III sound detection unit 330 detects the III sound.
The detection of the III sound (ST160) is performed based on the already detected I and II sounds.
In FIG. 11, the position indicated by the broken line is the III sound, which is often heard about 0.15 seconds after the II sound. Thus, for example, searching for the peak position in the section of 0.12 seconds to 0.17 seconds after the II sound can be cited as an example. The length of such a search section is appropriately set as necessary. sell. Alternatively, the search interval may be calculated each time based on the time interval between the sound I and the sound II without fixing the time of the search interval.

  If the peak position is detected in this search, the sound is III. The peak position search method for detecting the III sound may be the same method as the peak position re-search described in ST144.

Subsequently, in ST170, the IV sound detection unit 340 performs detection of the IV sound.
The detection of the IV sound (ST170) is performed based on the already detected I and II sounds.
In FIG. 12, the position indicated by the broken line is the IV sound, which is often heard immediately before the I sound. Thus, for example, searching for a peak position in a section from 0.10 seconds before the I sound to the I sound can be cited as an example, but the length of such a search section can be appropriately set as necessary. Alternatively, the search interval may be calculated each time based on the time interval between the sound I and the sound II without fixing the time of the search interval.

  If the peak position is detected in this search, the sound is an IV sound. The peak position search method for detecting the IV sound may be the same method as the peak position re-search described in ST144.

Subsequently, in ST180, the click sound detection unit 350 detects the click sound. The click sound is detected (ST180) based on the already detected I and II sounds. In FIG. 13, the position indicated by the broken line is the click sound, which is often heard immediately before the II sound. Thus, for example, searching for the peak position in the section from 0.10 seconds before the II sound to the II sound is given as an example, but the length of such a search section can be appropriately set as necessary.
Alternatively, the search interval may be calculated each time based on the time interval between the sound I and the sound II without fixing the time of the search interval.

  When a peak position is detected in this search, a click sound is set. The peak position search method for detecting the click sound may be the same method as the peak position re-search described in ST144.

Finally, in ST190, the heart noise detection unit 360 detects heart noise.
Detection of heart noise (ST190) is performed based on the already detected I and II sounds.
The heart noise is often heard between the I sound and the II sound or after the II sound, as exemplified by the hatched lines in FIGS. 14 (A) and (B). Therefore, a frame having a power higher than a predetermined threshold is detected between the I sound and the II sound and between the II sound and the I sound.
If a predetermined number or more of frames having power greater than the predetermined threshold continues, it is determined as cardiac noise.

Examples of the predetermined threshold value include a value obtained by multiplying an average value of the power of the I sound and the power of the II sound by a predetermined value (such as 0.5), but the threshold value is not limited to this and is appropriately set. Can be changed.
An example of the “predetermined number” in the case where “a predetermined number or more of frames whose power is greater than the predetermined threshold continues” is given as a length corresponding to 0.1 seconds, but this is also limited. is not.

Through the above process, I sound, II sound, II sound splitting, III sound, IV sound and heart noise were all accurately detected from the heart sound signal.
This detection result is sent to the display information generation unit 400 to generate image information so that each heart sound can be displayed on the screen. The generated image information is displayed on the display unit 130.
The user can grasp the state of the body by looking at the heart sound information displayed on the display unit. Since the heart sounds are classified and detected for each type by the analysis by the arithmetic processing unit 120, it is possible to grasp the body condition by looking at the analysis results of the heart sounds without medical expertise.

According to such a first embodiment, the following effects can be achieved.
In the present embodiment, iterative correction processing (ST130) is performed after peak position detection (ST120).
Thereby, even if there is an error in the detection of the peak position due to the influence of noise or the like, it is possible to correct the erroneous detection and the omission of detection by repeated correction processing, and accurately detect the I sound and the II sound. Then, other characteristic portions can be accurately detected based on the accurately detected I / II sounds.
Such signal analysis makes it possible to accurately detect the characteristics of the heart sound, so that the physical condition of the subject can be grasped, for example, by home medical care without relying on special medical knowledge or skill.

In this embodiment, there is an aspect in which processing steps increase, such as performing repeated correction processing, or deleting characteristic portions such as II sound splits and III sounds once in the process of correction processing, but there are noises etc. However, the accuracy of heart sound analysis is greatly improved because I and II sounds can be detected accurately.
Thus, by making it possible to reliably achieve accurate heart sound analysis, it becomes possible to accurately grasp the physical condition of the subject, which can greatly contribute to the expansion of home medical care.

(Modification)
In the first embodiment, in the accuracy determination, all the conditions of ST135, ST137, and ST138 are combined. However, the accuracy may be determined only by, for example, ST135 (variation degree). .

(Second embodiment)
A second embodiment according to the heart sound information processing apparatus of the present invention will be described.
The second embodiment is characterized by the configuration and processing steps for displaying the features of heart sounds in an easy-to-understand manner.
The configuration of the second embodiment is shown in FIG.
In FIG. 15, in the second embodiment, the display information generating unit 400 includes a shape determining unit 410 and a schematic unit 420. Then, the display information generation unit divides the heart sound detected by the heart sound detection unit 300 into patterns, and generates image information for performing a typical display according to each feature. The image information for the schematic display is, for example, the display position, size, shape, color, and the like. Specific examples will be described later.

  The frequency analysis unit 200 and the heart sound detection unit 300 may be the same as those in the first embodiment, or may detect heart sounds such as I sound and II sound by another method.

FIG. 16 is a flowchart showing the processing procedure of the second embodiment, but the frequency analysis ST100, the analysis target signal extraction ST101, and the heart sound detection ST102 are the same as those of the first embodiment, and thus the description thereof is omitted.
Although it is desirable to use ST110-ST190 of the first embodiment as the heart sound detection step (ST102), the specific means (step) is not limited as long as heart sounds can be detected and classified.

The display information generation process (ST200) by the display information generation unit 400 will be described.
When the heart sounds included in the analysis target signal s are classified and detected for each type in ST102, display information is generated from the classified results in ST200.
Details of the display information generation step (ST200) are shown in the flowchart of FIG.
First, one heart sound element to be displayed is selected in order (ST231). The heart sound elements may be selected one by one along the time axis. For example, in order, I sound, II sound, I sound, II sound, II sound splitting ...

If the selected heart sound element is not heart noise (ST232: NO), display information may be generated as it is.
Here, when the selected heart sound element is heart noise (ST232: YES), the process proceeds to a special step of further classifying the heart noise by shape determination (ST235). Specifically, as shown in FIG. 18, the heart noise is classified into five types of shapes: a constant shape 401, a gradually increasing and decreasing shape 402, a gradually increasing shape 403, a gradually decreasing shape 404, and a gradually decreasing and gradually increasing shape 405.

(Method for determining the shape of heart noise, part 1)
The process of heart noise shape determination (ST235) by the shape determination unit 410 will be described with reference to the flowchart of FIG.
First, a section classified as cardiac noise is smoothed using a moving average or the like (ST301), and a differential coefficient of that portion is calculated (ST302). In calculating the differential coefficient, the difference between the previous and next frames may be used as the differential coefficient, or may be calculated using a known Savitzky-Golay method, and the method is not limited.

Next, it is determined whether or not a derivative obtained by connecting the calculated differential coefficients of each frame intersects the time axis (ST303).
If the derivative does not intersect the time axis (ST303: NO), then the sign of the derivative is viewed (ST304). When the derivative is always positive (ST304: YES), the heart noise has a gradually increasing shape 403 (ST305).
When the derivative is always negative (ST304: NO, ST306: YES), the heart noise is a gradually decreasing shape 404 (ST307).

  On the other hand, if the derivative is not negative in ST306 (ST306: NO), that is, that the derivative is always zero, the shape of the heart noise is a constant shape 401 (ST308).

If the derivative crosses the time axis in ST303 (ST303: YES), it is further determined whether or not the derivative crosses the time axis only at one place (ST309).
When there is only one intersection between the derivative and the time axis (ST309: YES), the intersection is a switching point between increase and decrease.
Therefore, it is determined whether or not the derivative is smaller than 0 before the intersection of the time axis and the derivative (ST310). If the derivative changes from decreasing to increasing, the heart noise has a gradually increasing shape 405 (ST311).
On the other hand, when the frequency is changed from increasing to decreasing, the heart noise has a gradually increasing and decreasing shape 402 (ST312).
If the derivative crosses the time axis in multiple places instead of one, the derivative is assumed to be zero many times and transition around zero, so the heart noise is a constant shape 401 in this case (ST308).

(Method for determining the shape of heart noise, part 2)
Here, the heart noise shape determination (ST235) may be performed as follows instead of the flowchart of FIG. That is, the ratio of the area where the derivative is positive and the ratio of the area where the derivative is negative are considered, not whether or not the derivative intersects the time axis.
This will be described with reference to the flowchart of FIG.
The smoothing process (ST301) and the calculation of the differential coefficient (ST302) are the same.

  Here, next, in ST321, the ratio between the portion (region) where the differential coefficient is greater than 0 and the portion (region) where the differential coefficient is less than 0 is obtained. And if this ratio is extremely less than 0.1 or 0.9 or more (ST322: YES), even if there is a slight rise and fall, the shape increases almost monotonically as a whole Or it can be said that it is monotonously decreasing. Therefore, if the ratio is 0.9 or more (ST323: YES), the shape is gradually increased 403 (ST305), and if the ratio is 0.1 or less (ST323: NO), the shape is gradually decreased 404 (ST307).

In ST322, when the ratio is between 0.1 and 0.9, it is considered that the shape is either constant 401, gradually increasing 405, or gradually increasing 402. Therefore, it is determined whether or not the number of intersections between the derivative and the time axis is a predetermined number (for example, 3 times) or less (ST324).
If the number of crossings between the derivative and the time axis exceeds a certain number (eg, 3 times) (ST324: NO), the derivative will be zero many times and estimated to transition near zero. Is a fixed shape 401 (ST308).

On the other hand, when the number of intersections between the derivative and the time axis is equal to or less than a predetermined number (for example, 3 times) in ST324 (ST324: YES), next, “increase / decrease switching location” is specified.
The “increase / decrease switching location” is the boundary (end point) of the section where the differential coefficient with the same sign is the longest and continuous, and in FIG. 21, it is the location indicated by the arrow 1001 or the arrow 1002.
The shape is a gradual increase 405 (ST311) when the increase / decrease change point is changed, and the shape is a gradual increase 405 (ST311), and the shape is a gradual increase 402 (ST312). ).

As described above, in the method 1 and 2 of determining the shape of the heart noise, the case where the shape of the heart noise is classified into five types has been illustrated, but it is further classified, for example, by including the unevenness of the function. May be.
For example, whether the heart noise is after the I sound and before the II sound, or after the II sound and before the I sound, the positional relationship between the heart noise and the I sound / II sound The heart noise may be classified by paying attention.

Returning to the flow of FIG. 17, next, display information is generated (ST233).
This is based on the type of heart sound (I sound, II sound, etc.) detected by heart sound detection (ST102) and the shape of heart noise determined and classified by the shape determination (ST235). Generate information for display on
In the case of sounds other than heart noise, that is, I sound, II sound, II sound splitting, III sound, IV sound, click sound, etc., for example, the display position is exemplified as the start position (start time) and volume. However, other information such as the length of the pronunciation time may be added.

In the case of cardiac noise, display information is generated as follows based on the sound generation time, volume, and shape.
When the heart noise has a constant shape 401 (see FIG. 18), the average value of the volume is obtained and is assumed to be constant at the average value during the sound generation time.
In the case of the shape 402 in which the heart noise gradually increases and decreases, the sound volume is the minimum value at both ends, and further, the sound volume is the maximum value at the switching portion obtained in ST309 or ST325.
When the heart noise is a gradually increasing shape 403, the minimum value of the sound volume is assumed at the left end, and the maximum value of the sound volume is assumed at the right end.
When the heart noise is a gradually decreasing shape 404, the maximum value of the sound volume is assumed at the left end and the minimum value of the sound volume is assumed at the right end, and the interval is interpolated.
In the case of the shape 405 in which the heart noise is gradually decreased and gradually increased, it is assumed that the volume is the maximum value at both ends and the volume is the minimum value at the switching point, and the interval is interpolated.
As an interpolation method, linear interpolation is given as an example, but other interpolation methods may be used.

  When display information is generated for all elements (heart sound, heart noise) (ST234: YES), display information generation (ST200) is completed.

  Returning to the flow of FIG. 16, the schematic unit 420 performs schematic display information (ST400). That is, the display information generated in ST200 is schematically shown. Examples of the schematic are shown in FIG. 22, FIG. 23, and FIG.

22A shows a case where there is a II sound split, FIG. 22B shows a case where there is a III sound, FIG. 22C shows a case where there is an IV sound, and FIG. 22D shows a case where there is a click sound. . In this example, a line segment having a length (height) corresponding to the volume of each sound is displayed at the start position of each sound.
In addition, it may replace with a line segment and may be a strip-like rectangle with a little width, and it should just be displayed so that it may be easy to see. Moreover, you may change a color and a line type for every kind of heart sound.

FIGS. 23 (A)-(D) and FIGS. 24 (A)-(D) are schematic examples in the case where cardiac noise is included.
23 (A) and FIG. 24 (A) show the case where the heart noise has a constant shape, and FIG. 23 (B) and FIG. 24 (B) show the case where the heart noise has a gradually increasing and decreasing shape. 24 (C) is when the heart noise has a gradually increasing shape, and FIGS. 23 (D) and 24 (D) are when the heart noise has a gradually decreasing shape.
As described above, the shape (form) of the schematic view is not limited as long as it is a display in which the cardiac noise interval and its type can be distinguished.

The information modeled as described above is displayed on the screen (ST500).
FIG. 25 is an example of a display screen.
An example of the operation will be described with an example of this display screen.
First, a heart sound signal is selected from the open button 801. A recorded heart sound file may be selected, or a heart sound input in real time may be selected.
By pressing the play button 802, the analyzed heart sounds can be heard. When the stop button 803 is pressed, the heart sound reproduction is stopped.

  A repeat playback button 804 is pressed, and then a repeat playback section 807 is designated on the heart sound display area 805. Then, the designated section is reproduced repeatedly. A reproduction position display bar 806 shows a reproduction position at the time of heart sound reproduction.

The enlargement / reduction button 808 changes the scale of the heart sound display area 805.
The information display area 809 displays detailed information about a schematic display portion of heart sounds and heart noises.
Examples of the detailed information include heart sounds, names and shapes of heart sounds, normal sounds and abnormal sounds, and the like. Of course, other contents may be displayed.
Examples of the operation for displaying the detailed information include mouse over, click, screen touch, and the like, but any operation suitable for the device may be used.
The display method and position for displaying the detailed information may also be indicated by using, for example, an arrow in addition to the balloon as shown in FIG. 25, or may be displayed by providing a specific display area at the bottom of the screen.

According to such a second embodiment, the following effects are obtained.
That is, it is possible to accurately detect characteristic portions from the heart sounds and further display them in a schematic manner so that the user can understand them. Therefore, even a user with insufficient medical knowledge can easily grasp the characteristics included in the heart sound by looking at the schematic display, and can grasp the physical condition of the subject.

(Third embodiment)
Next, a third embodiment will be described.
The basic configuration of the third embodiment is the same as that of the second embodiment, but the third embodiment is characterized in that normal sounds and abnormal sounds are classified in addition to the classification of heart sound types.
FIG. 26 is a diagram showing a configuration of a heart sound information processing apparatus according to the third embodiment.
In the third embodiment, a heart sound classification unit 500 is provided to classify whether the heart sound detected by the heart sound detection unit is normal or abnormal.

FIG. 27 is a flowchart showing overall steps of the third embodiment.
In the third embodiment, after the heart sound is detected in ST102, heart sound classification (ST150) for classifying whether the heart sound is normal or abnormal is performed.
Heart sound classification (ST150) will be described with reference to the flowchart of FIG.

First, one heart sound element is selected in order (ST151), and it is determined whether it is an I sound or an II sound (ST152).
If the sound is an I sound or II sound (ST152: YES), and if it does not fall into any of the division sound, attenuation, or enhancement (ST153: NO), the heart sound element is a normal sound (ST154).
Here, whether or not it is a split sound has already been determined at the stage of heart sound detection (ST102).

  As for the attenuation and enhancement, for example, judgment is made by comparing the power with the other heart sound. That is, when the selected heart sound is I sound, it is compared with the II sound, and when the selected heart sound is II sound, it is compared with the I sound, and when the power is extremely small (for example, less than 50%). Is determined to be attenuated, and conversely, if the power is extremely high (for example, 150% or more), it is determined to be enhanced. Alternatively, attenuation / enhancement may be determined based on a predetermined threshold, and the method is not limited.

  Further, since the volume balance of the I sound and the II sound changes depending on the auscultation position, the user may input the auscultation position at the time of use, and add the position information to the determination of attenuation / enhancement. Further, since it varies depending on the body shape, information on height and weight may be added to the determination of attenuation / enhancement.

On the other hand, in ST153, if it is an I sound or an II sound, and it falls into any one of splitting sound, attenuation and enhancement (ST153: YES), the heart sound element is an abnormal sound (ST158).
If it is not the I sound or the II sound (ST152: NO), it is determined whether or not it is the III sound (ST156). If it is not an I, II, or III sound (ST156: NO), the heart sound is considered abnormal (ST158).
If it is a III sound (ST156: YES), if the subject is under a predetermined age (ST157), for example, under 40 years of age, it is not abnormal to have a III sound, so the heart sound element is a normal sound (ST154) .

  The selected heart sound elements are classified into normal sounds (ST154) or abnormal sounds (ST158), and when all the heart sound elements have been classified (ST155: YES), the heart sound classification process (ST150) ends.

Returning to FIG. 27, since the heart sound classification (ST150) is completed, next, display information is generated (ST260).
With reference to FIG. 29, the display information generation step (ST260) will be described.
A heart sound element is selected (ST231), and the shape of the heart noise is determined (ST235). The shape determination (ST235) is the same as in the first embodiment.

Next, in ST261, normal sounds and abnormal sounds are counted. That is, if the selected heart sound (ST231) is classified as a normal sound in ST150, the number of normal sounds is counted up by one.
If the selected heart sound (ST231) is classified as an abnormal sound in ST150, the number of abnormal sounds is counted up by one.
By this count, the number of normal sounds and the number of abnormal sounds in the heart sound signal to be analyzed are counted.

  For abnormal sounds, set specific coefficients for the type, size, duration of sound generation, shape of heart noise, positional relationship with I and II sounds, and so on. Count (ST261).

  Display information is generated based on the information so far (ST262). Display information is generated for all the heart sound elements (ST234: Yes), modeled (ST400), and then displayed on the screen (ST500). Examples of screen display will be described with reference to FIG. 30, FIG. 31, and FIG.

In FIGS. 30 to 32, reference numerals 1101 to 1106 denote examples of graph display indicating the count numbers of abnormal sounds and normal sounds.
In FIG. 30 and FIG. 31, the horizontal axis represents the number of counts, and abnormal sound counts 1101 and 1103 and normal sound counts 1102 and 1104 are displayed on the screen.
In FIG. 32, an abnormal sound ratio 1105 and a normal sound ratio 1106 are displayed on the screen.
As shown in Fig. 30 to Fig. 32, the color and painting method (diagonal lines, dots, etc.) are changed according to the type of sound (I-IV sound, click sound, heart noise) so that the sound distribution is easy to understand. It is preferable.

  The screen display method can be changed in various ways, such as changing the direction of the axes, adding axes such as power and pronunciation time, and using graphs of other shapes (such as pie charts and bubble charts). is there.

When a characteristic sound is being reproduced on the display screen, the sound may be schematically displayed in the areas 1107, 1109, and 1111.
As the display content, the original waveform may be displayed as it is, or a schematic display as shown in FIGS.
Further, a description of the characteristic sound schematically displayed and whether it is an abnormal sound or a normal sound may be shown in regions 1108, 1110, and 1112.

  According to such a third embodiment, similarly to the second embodiment, even a user with insufficient medical knowledge can easily grasp the characteristics included in the heart sound by looking at the schematic display, and the physical condition of the subject Can be grasped.

DESCRIPTION OF SYMBOLS 100 ... Heart sound information processing apparatus, 110 ... Signal input part, 120 ... Operation processing part, 150 ... Display part, 160 ... Input means, 200 ... Frequency analysis part, 210 ... Target signal extraction part, 300 ... Heart sound detection part, 310 ... I / II sound detection unit, 311 ... peak position detection unit, 312 ... accuracy determination unit, 313 ... correction processing unit, 320 ... II sound division detection unit, 330 ... III sound detection unit, 340 ... IV sound detection unit, 350 ... click sound detection unit, 360 ... heart noise detection unit, 370 ... characteristic location detection unit, 400 ... display information generation unit, 410 ... shape determination unit, 500 ... heart sound classification unit.

Claims (24)

A peak position detector for detecting a plurality of peak positions from a heart sound signal obtained by collecting heart sounds;
An accuracy determination unit that determines whether or not the detection of the peak position by the peak position detection unit is accurate;
A correction processing unit that performs correction processing to correct the detection of the peak position when it is determined by the determination by the accuracy determination unit that the detection of the peak position by the peak position detection unit is not accurate,
The accuracy determination unit determines the accuracy of peak position detection based on a result of clustering peak positions based on a time length between detected peak positions,
The heartbeat information processing apparatus, wherein the correction processing unit repeatedly executes correction processing according to a determination result by the accuracy determination unit. In the heart sound information processing apparatus according to claim 1,
The accuracy determination unit
A heart sound information processing apparatus, wherein accuracy of peak position detection is determined based on a value indicating a variation in a histogram of each cluster. In the heart sound information processing apparatus according to claim 1 or claim 2,
The accuracy determination unit
A heart sound information processing apparatus, wherein accuracy of peak position detection is determined based on a difference or ratio of the number of elements in each cluster. In the heart sound information processing apparatus according to any one of claims 1 to 3,
The accuracy determination unit
A heart sound information processing apparatus, wherein accuracy of peak position detection is determined based on whether clusters to which adjacent peak positions belong are the same or different. In the heart sound information processing apparatus according to any one of claims 1 to 4,
An upper limit number of times is set for the accuracy determination by the accuracy determination unit and the correction processing by the correction processing unit. In the heart sound information processing apparatus according to any one of claims 1 to 5,
The correction processing unit
When two or more peak positions that are not mode values of the elements of each cluster are continuous,
A heart sound information processing apparatus, wherein a peak position where both of the preceding and following time intervals deviate from the mode value of the cluster is deleted as an erroneously detected peak position. In the heart sound information processing apparatus according to any one of claims 1 to 6,
The correction processing unit
When two or more peak positions belonging to the same cluster are continuous, and peak positions that are not the mode values of the elements of each cluster are adjacent to each other before and after that,
A heart sound information processing apparatus, wherein a peak position belonging to the same cluster adjacent to a peak position that is not the mode value is deleted. In the heart sound information processing apparatus according to any one of claims 1 to 7,
The correction processing unit
A heart sound information processing apparatus characterized by re-searching a peak position in a section between a peak position having a longer time length than a mode value of a cluster having a longer time length and the next peak position. In the heart sound information processing apparatus according to any one of claims 1 to 8,
And a feature location detection unit that further detects a characteristic location of the heart sound based on the detected peak location after the accuracy determination unit determines that the peak location detection is accurate. Heart sound information processing device. In the heart sound information processing apparatus according to claim 9,
A display information generating unit for generating display information for displaying the characteristic location of the heart sound detected by the peak position detection unit and the characteristic location detection unit on a display unit;
The display information generation unit, a shape determination unit for classifying the shape of heart noise when the heart sound element is heart noise,
A heart sound information processing apparatus comprising: a schematic unit that generates schematic display information based on a classification result. In the heart sound information processing apparatus according to claim 10,
The heart sound information processing apparatus, wherein the shape determination unit classifies the shape of heart noise based on a differential coefficient of heart sounds in a heart noise section. In the heart sound information processing apparatus according to claim 10 or claim 11,
The display information generating unit further classifies heart noise based on a relative positional relationship between heart noise and heart sounds I and II sounds. In the heart sound information processing apparatus according to any one of claims 10 to 12,
The schematic unit further generates schematic display information based on a shape classification of heart noise and a volume of heart noise. In the heart sound information processing apparatus according to any one of claims 10 to 13,
The heart sound information processing apparatus further comprises a heart sound classification unit that classifies the characteristic locations of the heart sounds detected by the peak position detection unit and the characteristic location detection unit into normal sounds and abnormal sounds. In the heart sound information processing apparatus according to claim 14,
The heart sound classification unit further classifies normal sound and abnormal sound by adding at least one of the subject's age information, the subject's body shape information, and the information of the auscultation position when auscultating the subject to the judgment criteria. A heart sound information processing apparatus characterized by In the heart sound information processing apparatus according to any one of claims 10 to 15,
The said display information generation part counts the number of the characteristic location of the detected heart sound for every kind. The heart sound information processing apparatus characterized by the above-mentioned. The heart sound information processing apparatus according to claim 16,
The display information generation unit includes at least one piece of information on the type, size, length of sounding time, and shape of the heart and the positional relationship with the sound I and sound II for each feature location of the heart sound. A heart sound information processing apparatus characterized in that a coefficient is set based on, and counting is performed using the coefficient. In the heart sound information processing apparatus according to any one of claims 10 to 17,
The heart sound information processing apparatus, wherein the schematic unit performs color coding for each type of heart sound and heart noise. In the heart sound information processing apparatus according to any one of claims 10 to 18,
The heartbeat information processing apparatus, wherein when the heartbeat sound is played back and output, the display information generation section displays a playback location symbol indicating a feature location of the heartbeat corresponding to the heartbeat being played on the display portion. In the heart sound information processing apparatus according to any one of claims 10 to 19,
The display information generating unit displays a symbol indicating a repeated section on the display unit when repeatedly reproducing and outputting a specific section of the heart sound. In the heart sound information processing apparatus according to any one of claims 10 to 20,
The display information generating unit displays a description of the feature location of the heart sound being displayed when the feature location of the heart sound is displayed on the display unit. In the heart sound information processing apparatus according to any one of claims 10 to 21,
The heartbeat information processing apparatus, wherein when the heartbeat sound is reproduced and output, the display information generation portion causes the display portion to display a schematic diagram schematically showing a feature portion of the heartbeat corresponding to the heartbeat being reproduced. Detecting a plurality of peak positions from a heart sound signal obtained by collecting heart sounds;
Clustering the peak positions based on the time length between the detected peak positions;
Determining the accuracy of peak position detection based on the clustered results;
Correcting the detection of the peak position when the determination of the accuracy determines that the detection of the peak position is not accurate, and
The heart sound information processing method, wherein the step of correcting the detection of the peak position is repeatedly executed according to the accuracy determination result. Computer
A peak position detector for detecting a plurality of peak positions from a heart sound signal obtained by collecting heart sounds;
An accuracy determination unit that determines the accuracy of peak position detection based on the result of clustering the peak positions based on the time length between detected peak positions,
A heart sound information processing program that functions as a correction processing unit that repeatedly executes correction processing that corrects detection of a peak position in accordance with a determination result by the accuracy determination unit.

Images