Fetal heart rate monitoring device, fetal heart rate monitoring system, and fetal heart rate monitoring program

Abstract

To improve the accuracy of fetal heart rate measurement based on Doppler signals. [Solution] The Doppler signal forming unit 18 has a Doppler deviation frequency f, which is the difference frequency from the transmission frequency. d The Doppler signal TDS is formed to indicate the target Doppler signal. The Doppler signal identification unit 20 identifies whether the target Doppler signal TDS is composed of fetal heart rate components or fetal heart rate components and artifact components by inputting the target Doppler signal TDS into a learning model that has been trained to identify whether the input Doppler signal is composed of fetal heart rate components or fetal heart rate components and artifact components, using Doppler signals composed of fetal heart rate components and Doppler signals composed of fetal heart rate components and artifact components as training data. The heart rate measurement unit 22 measures the fetal heart rate based on the target Doppler signal TDS identified by the Doppler signal identification unit 20 as being composed of fetal heart rate components.

Description

[Technical Field]

[0001] This specification discloses improvements to a fetal heart rate measurement device, a fetal heart rate measurement system, and a fetal heart rate measurement program. [Background technology]

[0002] Traditionally, fetal heart rate has been measured. Methods for measuring fetal heart rate include those using fetal electrocardiographs, but another method has been proposed that involves transmitting ultrasound waves towards the fetal heart and measuring the fetal heart rate based on the Doppler signal obtained from the reflected waves from the fetal heart. Since the Doppler signal detects the movement of the object being examined, measuring fetal heart rate based on the Doppler signal can be said to be a measurement method based on the movement of the fetal heart.

[0003] For example, Patent Document 1 and Non-Patent Document 1 disclose a technique for measuring the fetal heart rate by transmitting ultrasound waves to a fetus, obtaining a Doppler signal from the reflected waves from the fetus, and detecting its periodicity by performing autocorrelation processing on the Doppler signal. [Prior art documents] [Patent Documents]

[0004] [Patent Document 1] Japanese Patent Publication No. 10-028686 Public Relations [Non-patent literature]

[0005] [Non-Patent Document 1] Hamelmann et al., "Doppler ultrasound technology for fetal heart rate monitoring", IEEE transactions on ultrasonics, ferroelectrics, and frequency control, VOL.67, NO.2, FEBRUARY 2020. [Overview of the project] [Problems that the invention aims to solve]

[0006] Due to factors such as the fetus being inside the mother's womb and the fact that sound waves spread as they travel, it is impossible to transmit ultrasound waves only to the fetal heart, even when transmitting ultrasound towards the fetal heart. As a result, an ultrasound probe that transmits ultrasound waves towards the fetal heart receives not only reflected waves from the fetal heart but also reflected waves from the surrounding tissues. Therefore, the Doppler signal used to measure the fetal heart rate includes not only the signal indicating the fetal heart rate but also signals indicating the movement of the surrounding tissues. Movement of surrounding tissues includes, for example, maternal body movement, maternal tissue movement, or the movement of the fetus's entire body. Such signal components other than the target signal component (the fetal heart rate component indicating the fetal heart rate in this specification) (signal components indicating the movement of surrounding tissues in this specification) are called artifact components.

[0007] Furthermore, unlike fetal electrocardiograms (FECG) measured by a fetal electrocardiograph, the fetal heart rate component of the Doppler signal can fluctuate depending on the direction of ultrasound transmission to the fetal heart. For example, if the fetus's position changes during ultrasound transmission, the signal waveform of the fetal heart rate component may fluctuate.

[0008] As mentioned above, the Doppler signal obtained from the reflected ultrasound waves transmitted towards the fetal heart may contain artifact components. Furthermore, even focusing solely on the fetal heart rate component, its signal waveform can fluctuate. Therefore, in conventional methods such as autocorrelation processing, the measurement accuracy of fetal heart rate based on Doppler signals may decrease due to these artifact components and fluctuations in the signal waveform of the fetal heart rate component. In other words, there was room for improvement in the accuracy of fetal heart rate measurement based on Doppler signals.

[0009] The purpose of the fetal heart rate measurement device, fetal heart rate measurement system, or fetal heart rate measurement program disclosed herein is to improve the accuracy of fetal heart rate measurement based on Doppler signals. [Means for solving the problem]

[0010] The fetal heart rate measuring device according to this embodiment measures the fetal heart rate based on a Doppler signal representing the difference frequency between the frequency of ultrasound transmitted toward the fetal heart and the frequency of the reflected ultrasound waves, and is characterized by comprising: a Doppler signal identification unit that identifies whether the target Doppler signal is composed of a fetal heart rate component or composed of a fetal heart rate component and artifact components based on the output data of the learning model when a target Doppler signal, which is the Doppler signal to be processed, is input to a learning model that has been trained to predict and output whether the input Doppler signal is composed of a fetal heart rate component or composed of a fetal heart rate component and artifact components, using the Doppler signal composed of a fetal heart rate component and the Doppler signal composed of a fetal heart rate component as training data; and a heart rate measuring unit that measures the fetal heart rate based on the target Doppler signal identified as composed of a fetal heart rate component.

[0011] The Doppler signal identification unit extracts unit Doppler signals from the target Doppler signal using a time window of a predetermined length, and sequentially inputs the multiple unit Doppler signals obtained by shifting the time window by unit time into the learning model in the order they were acquired, thereby identifying whether each unit Doppler signal consists of a fetal heart rate component or consists of a fetal heart rate component and an artifact component. The heart rate measurement unit then measures the fetal heart rate based on the time difference between the exercise period, which is the period corresponding to a block of unit Doppler signals in the time direction that has been identified by the Doppler signal identification unit as consisting of a fetal heart rate component, and the next exercise period in the time series.

[0012] The Doppler signal identification unit extracts unit Doppler signals from the target Doppler signal using a time window of a predetermined length, and sequentially inputs the multiple unit Doppler signals obtained by shifting the time window by unit time into the learning model in the order they were acquired, thereby identifying whether each unit Doppler signal consists of a fetal heart rate component or consists of a fetal heart rate component and an artifact component. The heart rate measurement unit calculates a correlation coefficient between a reference signal and a comparison target portion of the target Doppler signal which has the same time length as the reference signal. The comparison target portion whose correlation coefficient satisfies predetermined conditions is used as a new reference signal, and autocorrelation processing is performed to calculate the time change of the correlation coefficient obtained by shifting the comparison target portion in the time direction. The fetal heart rate is measured based on the period of the time change of the correlation coefficient. If at least a part of the comparison target portion whose correlation coefficient satisfies predetermined conditions is determined by the Doppler signal identification unit to consist of a fetal heart rate component and an artifact component, that comparison target portion should not be used as a new reference signal.

[0013] The learning model is trained to predict and output the timing of the behavior in the input Doppler signal, using the learning data to which timing labels indicating the timing of the movement of at least one of the heart wall or heart valves in relation to the Doppler signal composed of fetal heart rate components. The Doppler signal identification unit may identify the timing of the behavior in the target Doppler signal based on the output data of the learning model when the target Doppler signal is input, if the target Doppler signal is composed of fetal heart rate components.

[0014] The learning model is trained using the learning data which further includes a heart diameter parameter indicating the diameter of the fetal heart. The Doppler signal identification unit may identify whether the target Doppler signal consists of a fetal heart rate component or consists of a fetal heart rate component and an artifact component, based on the output data of the learning model when the target Doppler signal and the heart diameter parameter corresponding to the target Doppler signal are input.

[0015] The fetal heart rate measuring device may further include an operating mode switching unit that switches the operating mode between a first operating mode in which the heart rate measuring unit measures the fetal heart rate based on the target Doppler signal identified by the Doppler signal identification unit as consisting of fetal heart rate components, and a second operating mode in which the heart rate measuring unit performs autocorrelation processing to calculate the time change of the correlation coefficient obtained by shifting the comparison portion in the time direction, without using the identification result by the Doppler signal identification unit, and measures the fetal heart rate based on the period of the time change of the correlation coefficient.

[0016] The Doppler signal included in the training data is preferably a Doppler signal representing the duration of time during which the heart wall or heart valve is moving in one heartbeat, extracted from the Doppler signal which is not divided by heartbeat, based on a first time interval from which the signal is extracted by applying a low-pass filter to the Doppler signal to extract low-frequency components corresponding to the movement of the fetal heart wall, a second time interval from which the signal is extracted by applying a band-pass filter to the Doppler signal to extract frequency components corresponding to the movement of the fetal heart wall and heart valve, and a third time interval from which the signal is extracted by applying a high-pass filter to the Doppler signal to extract high-frequency components corresponding to the movement of the fetal heart valve.

[0017] The fetal heart rate measurement system according to this embodiment comprises a Doppler signal generating device and a fetal heart rate measurement device, wherein the Doppler signal generating device includes an ultrasound probe that transmits ultrasound waves toward the fetal heart and receives the reflected waves of the ultrasound waves, a Doppler signal generating unit that forms a target Doppler signal which is a Doppler signal to be processed, representing the difference frequency between the frequency of the ultrasound waves transmitted toward the fetus and the frequency of the reflected waves, and a Doppler signal transmitting unit that transmits the target Doppler signal to the fetal heart rate measurement device via a communication line, and the fetal heart rate measurement device comprises the Doppler signal composed of fetal heart rate components, and the fetal heart rate components and artifact components. The device is characterized by including: a Doppler signal identification unit that identifies whether the target Doppler signal consists of a fetal heart rate component or consists of a fetal heart rate component and artifact components, based on the output data of the learning model when the target Doppler signal is input, using the Doppler signal obtained as training data to predict and output whether the input Doppler signal consists of a fetal heart rate component or consists of a fetal heart rate component and artifact components; and a heart rate measurement unit that measures the fetal heart rate based on the target Doppler signal identified as consisting of a fetal heart rate component.

[0018] The fetal heart rate measurement program according to this embodiment is a fetal heart rate measurement program that operates a computer as a fetal heart rate measurement device that measures the fetal heart rate based on a Doppler signal representing the difference frequency between the frequency of ultrasonic waves transmitted toward the fetal heart and the frequency of the reflected waves of the ultrasonic waves. The computer is configured to use, as learning data, the Doppler signal composed of fetal heart rate components and the Doppler signal composed of fetal heart rate components and artifact components, and to learn to predict and output whether the input Doppler signal is composed of fetal heart rate components or is composed of fetal heart rate components and artifact components. Based on the output data of the learning model when the target Doppler signal, which is the Doppler signal to be processed, is input to the learning model, a Doppler signal identification unit that identifies whether the target Doppler signal is composed of fetal heart rate components or is composed of fetal heart rate components and artifact components, and a heart rate measurement unit that measures the fetal heart rate based on the target Doppler signal identified as being composed of fetal heart rate components. It is characterized by operating as such.

Effect of the Invention

[0019] According to the fetal heart rate measurement device, fetal heart rate measurement system, or fetal heart rate measurement program disclosed in this specification, it is to improve the measurement accuracy of the fetal heart rate based on the Doppler signal.

Brief Description of the Drawings

[0020] [Figure 1] It is a schematic configuration diagram of the fetal heart rate measurement device according to this embodiment. [Figure 2] It is a conceptual diagram of a Doppler signal composed of fetal heart rate components. [Figure 3] It is a conceptual diagram of a Doppler signal composed of fetal heart rate components and artifact components. [Figure 4] It is a conceptual diagram showing the generation process of learning data. [Figure 5] It is a graph showing the relationship between the frequency and signal components in the Doppler signal. [Figure 6]This is a conceptual diagram showing the output of the hysteresis processing unit. [Figure 7] This is a conceptual diagram illustrating the process of assigning timing labels to training data. [Figure 8] This is a conceptual diagram showing a unit Doppler signal. [Figure 9] This is a conceptual diagram illustrating the processing of the Doppler signal identification unit. [Figure 10] This is a conceptual diagram illustrating the first method for measuring fetal heart rate. [Figure 11] This is a conceptual diagram showing the comparison portion used in autocorrelation processing. [Figure 12] This is a conceptual diagram showing the reference signal used in autocorrelation processing. [Figure 13] This graph shows the time evolution of the correlation coefficient between the comparison portion and the reference signal. [Figure 14] This is a schematic diagram of the configuration of the fetal heart rate measurement system according to this embodiment. [Modes for carrying out the invention]

[0021] <Overview of Fetal Heart Rate Measurement Devices> Figure 1 is a schematic diagram of the configuration of the fetal heart rate measurement device 10 according to this embodiment. The fetal heart rate measurement device 10 is a medical device installed in a medical institution such as a hospital, and is a device that performs the function of measuring the heart rate of a fetus. The fetal heart rate measurement device 10 may be, for example, a fetal monitoring device that also measures the mother's heart rate and contractions.

[0022] The ultrasound probe 12 transmits ultrasound waves toward the fetal heart and receives the reflected waves of the ultrasound waves. The ultrasound probe 12 is equipped with multiple vibrating elements. In this embodiment, the ultrasound probe 12 has several vibrating elements (10 or fewer).

[0023] The ultrasound probe 12 transmits ultrasound waves from multiple vibrating elements based on a transmission signal, which is an electrical signal from a transmitting unit 14 (described later). These multiple vibrating elements then receive reflected waves from various tissues within the ultrasound transmission area, including the fetal heart, and convert the reflected waves into received signals, which are electrical signals, and output them.

[0024] The ultrasound probe 12 is a portable device that is attached to the surface of the mother's abdomen by a belt or the like. In this state, the ultrasound probe 12 transmits ultrasound waves toward the fetal heart and receives the reflected waves of the ultrasound waves.

[0025] The transmitting unit 14 transmits a transmission signal to the ultrasound probe 12 (more specifically, multiple vibrating elements). As a result, ultrasound waves are transmitted from each vibrating element toward the fetal heart. In this embodiment, the transmitting unit 14 causes the multiple vibrating elements to transmit ultrasound waves having a center frequency of approximately 1 MHz. In this embodiment, the transmitting unit 14 repeatedly causes the multiple vibrating elements to transmit ultrasound waves simultaneously and then stops transmitting ultrasound waves simultaneously. When ultrasound transmission is stopped, the multiple vibrating elements receive reflected waves from the fetal heart, etc. Specifically, the transmitting unit 14 intermittently transmits a transmission signal to the multiple vibrating elements such that the ultrasound transmission period and the transmission stop period (reception period) repeat at a frequency of several kHz. In this embodiment, approximately 100 ultrasound waves are transmitted during one transmission period.

[0026] In this embodiment, the ultrasound transmitted towards the fetal heart is used solely for measuring the fetal heart rate. That is, the ultrasound is not used for forming ultrasound images or for measuring the velocity of tissues or blood flow in the subject.

[0027] The receiving unit 16 receives a received signal from the ultrasonic probe 12 (more specifically, the multiple vibrating elements) based on the reflected waves received by the multiple vibrating elements during the reception period described above. As described above, although the ultrasound from the ultrasonic probe 12 is transmitted toward the fetal heart, the ultrasound can also be reflected from tissues other than the fetal heart. Therefore, the received signal received by the receiving unit 16 may include signals based on reflected waves from various tissues as well as the fetal heart. The frequency of the reflected waves may have a frequency that is shifted from the frequency of the transmitted ultrasound (transmission frequency) due to the Doppler effect caused by the movement of each tissue, including the fetal heart (Doppler shift frequency).

[0028] The receiving unit 16 is equipped with an amplifier, which adds the received signals from multiple vibration elements and then amplifies them. The receiving unit 16 outputs the amplified received signal to the Doppler signal forming unit 18.

[0029] The Doppler signal forming unit 18 has a mixer. The mixer multiplies the received signal received from the receiving unit 16 by a reference signal having the same frequency f0 as the transmission frequency. This results in the Doppler deviation frequency f of the received signal. d (In other words, a signal having the difference frequency from the transmission frequency, and 2f0+f d A signal with the following frequency is obtained.

[0030] Furthermore, the Doppler signal generation unit 18 has a low-pass filter. The low-pass filter filters the higher frequency component 2f0+f of the two signals acquired by the mixer, which have different frequency components. d The signal is removed. This eliminates the Doppler shift frequency f. d A signal possessing this characteristic is extracted.

[0031] Furthermore, the Doppler signal forming unit 18 includes an AD converter. The AD converter uses the Doppler shift frequency f obtained by the low-pass filter. d A signal having [specific characteristics] is converted into a digital signal. In this embodiment, the sampling frequency for digital conversion is approximately 2 kHz.

[0032] The Doppler shift frequency f obtained by the Doppler signal generation unit 18 d The digital signal shown is the Doppler signal as used herein. The Doppler signal can be represented as a waveform in a coordinate space where the horizontal axis is defined by time and the vertical axis by the signal intensity of the reflected wave. As described above, the Doppler signal may contain not only the fetal heart rate component but also artifact components.

[0033] Furthermore, the Doppler signal formation unit 18 also has an FFT (Fast Fourier Transform) processing unit, and the Doppler shift frequency f d Further FFT processing may be performed on the digital signal having the following characteristics. The signal after FFT processing may be used as a Doppler signal. This Doppler signal can be represented as a waveform in a coordinate space where the horizontal axis is time and the vertical axis is frequency (in other words, the velocity of each tissue within the ultrasound transmission region).

[0034] The Doppler signal identification unit 20 uses a pre-trained model 32 (described later) to identify whether the Doppler signal consists of fetal heart rate components or fetal heart rate components and artifact components. The heart rate measurement unit 22 measures the fetal heart rate based on the Doppler signal identified by the Doppler signal identification unit 20 as consisting of fetal heart rate components. Details of the processing performed by the Doppler signal identification unit 20 and the heart rate measurement unit 22 will be described later.

[0035] The output control unit 24 displays the fetal heart rate measured by the heart rate measurement unit 22 on the display 26. The output control unit 24 displays the fetal heart rate on the display 26 as a numerical value, such as beats per minute. Alternatively, the output control unit 24 may display the fetal heart rate as it changes over time in the form of a graph on the display 26. In addition to displaying the heart rate measured by the heart rate measurement unit 22 on the display 26, the output control unit 24 may also output the heart rate as sound from a speaker (not shown). For example, the speaker may output pre-recorded heart sounds in sync with the fetal heart rate.

[0036] The display 26 is comprised of, for example, a liquid crystal display. Various screens are displayed on the display 26 in response to control from the output control unit 24.

[0037] The input interface 28 consists of a touch panel and various buttons. The input interface 28 is used to input instructions from users such as doctors, nurses, or pregnant women to the fetal heart rate measurement device 10.

[0038] Memory 30 is composed of non-volatile memory such as eMMC (embedded Multi Media Card) and ROM (Read Only Memory), and volatile memory such as RAM (Random Access Memory). The non-volatile memory stores the fetal heart rate measurement program for operating each part of the fetal heart rate measurement device 10. The fetal heart rate measurement program can also be stored on a computer-readable non-temporary storage medium such as a USB (Universal Serial Bus) memory or a CD-ROM. The fetal heart rate measurement device 10 can read and execute the fetal heart rate measurement program from such a storage medium.

[0039] Furthermore, as shown in Figure 1, the learning model 32 is stored in memory 30 (more specifically, non-volatile memory). The learning model 32 is not limited to this, but for example, an image recognition model such as CNN (Convolutional Neural Network) or a natural language processing model such as Transformer can be used. The learning model 32 is trained to distinguish whether the input Doppler signal consists of the fetal heart rate component or the fetal heart rate component and artifact component, using Doppler signals composed of the fetal heart rate component and artifact component as training data. The learning method of the learning model 32 and how to use it will be described later.

[0040] The operating mode switching unit 34 changes the operating mode of the fetal heart rate measurement device 10 in response to instructions from the user. Each operating mode of the fetal heart rate measurement device 10 will be described later.

[0041] The configuration outline of the fetal heart rate measurement device 10 is as described above. Of the above-mentioned parts of the fetal heart rate measurement device 10, the transmitting unit 14, receiving unit 16, Doppler signal formation unit 18, Doppler signal identification unit 20, heart rate measurement unit 22, output control unit 24, and operating mode switching unit 34 are composed of a processor. The processor is composed of at least one of a general-purpose processing unit (e.g., a CPU (Central Processing Unit)) and a dedicated processing unit (e.g., a GPU (Graphics Processing Unit), ASIC (Application Specific Integrated Circuit), FPGA (Field Programmable Gate Array), or programmable logic device). The processor may not consist of a single processing unit, but rather of multiple processing units located in physically separate locations working together. Furthermore, each of the above-mentioned parts may be realized through the cooperation of hardware such as a processor and software.

[0042] <Training data and method for generating training data> Next, the training data for training the learning model 32 and the method for generating said training data will be described. In this embodiment, the training data is generated and the training process for training the learning model 32 is executed in a learning processing device, which is a device other than the fetal heart rate measurement device 10. The trained learning model 32 is then stored in the memory 30 of the learning processing device. However, the fetal heart rate measurement device 10 may also be equipped with a learning processing unit that performs at least one of the training data generation and training process described below; that is, the fetal heart rate measurement device 10 may perform at least one of the training data generation and training process.

[0043] Figure 2 is a conceptual diagram of a Doppler signal DSA composed of fetal heart rate components (hereinafter simply referred to as "Doppler signal DSA"), and Figure 3 is a conceptual diagram of a Doppler signal DSB composed of fetal heart rate components and artifact components (hereinafter simply referred to as "Doppler signal DSB"). The training data for training the learning model 32 consists of a Doppler signal DSA labeled to indicate that it is composed of fetal heart rate components, and a Doppler signal DSB labeled to indicate that it is composed of fetal heart rate components and artifact components. In this embodiment, the training data consists of a Doppler signal DSA for one heartbeat and a Doppler signal DSB for one heartbeat. Note that the training data does not necessarily have to be for one heartbeat, and may consist of Doppler signals for several heartbeats.

[0044] In this specification, the term "Doppler signal composed of fetal heart rate components" does not refer only to Doppler signals that contain only fetal heart rate components (in other words, signals that contain no artifact components at all), but also to Doppler signals that contain some artifact components. For example, a Doppler waveform that contains a small amount of artifact components, to the extent that the fetal heart rate components can be sufficiently read when viewed as a waveform, can be considered a "Doppler signal composed of fetal heart rate components." On the other hand, a Doppler waveform in which the fetal heart rate components are buried by artifact components, or in other cases where the fetal heart rate components cannot be read when viewed as a waveform, is considered a "Doppler signal composed of both fetal heart rate components and artifact components."

[0045] The training data is generated based on a set of Doppler signals acquired by transmitting ultrasound to the fetal heart in the past and stored in a Doppler signal database. The Doppler signals stored in the Doppler signal database may be data with various time lengths, but generally, they are data with time lengths corresponding to multiple heartbeats.

[0046] As a method for generating training data, one can be employed in which the training data creator analyzes each Doppler signal stored in the Doppler signal database, identifies the portion corresponding to the heartbeat in each Doppler signal, determines whether the identified portion consists of fetal heartbeat components or fetal heartbeat components and artifact components, extracts the relevant portion, and assigns a label according to the determination, repeating this process. However, a considerable amount of training data is required to adequately train the training model 32, and creating such a large amount of training data would be quite time-consuming using the above method.

[0047] Therefore, in this embodiment, as described below, training data is generated from the Doppler signal group stored in the Doppler signal database by automating at least a portion of the process. Specifically, the process of separating Doppler signals that have not been separated by heart rate into heart rate-specific signals is automated.

[0048] Figure 4 is a conceptual diagram illustrating the process of generating training data. First, the training data creator selects a Doppler signal DS from the Doppler signal group stored in the Doppler signal database that has not yet been segmented by heart rate. The training data creator then examines the Doppler signal DS and roughly segment it into a portion consisting of fetal heart rate components and a portion consisting of fetal heart rate components and artifact components. The Doppler signal DS after rough segmentation does not need to be segmented by heart rate. If the entire selected Doppler signal DS consists of fetal heart rate components, or if the entire selected Doppler signal DS consists of fetal heart rate components and artifact components, rough segmentation is not necessary.

[0049] Next, the roughly segmented Doppler signal DS is input to a low-pass filter 40 that extracts signals below a first frequency (e.g., around 50 Hz), a band-pass filter 42 that extracts signals from the first frequency up to a second frequency higher than the first frequency (e.g., around 250 Hz), and a high-pass filter 44 that extracts signals above the second frequency.

[0050] Fetal heart rate components include a heart wall component corresponding to the movement of the fetal heart wall and a heart valve component corresponding to the movement of the fetal heart valve. FIG. 5 is a graph showing the relationship between the frequency and signal components in the Doppler signal DS. Since the heart wall has a slower movement speed than the heart valve, as shown in FIG. 5, in the Doppler signal DS, the Doppler shift frequency f of the reflected wave from the heart wall d is the Doppler shift frequency f of the reflected wave from the heart valve d will have a smaller frequency. Therefore, the heart wall component, which is a low-frequency component, is extracted from the Doppler signal DS by the low-pass filter 40. On the other hand, the heart valve component, which is a high-frequency component, is extracted from the Doppler signal DS by the high-pass filter 44. Also, the frequency components corresponding to both the heart wall component and the heart valve component (the overlapping part) are extracted from the Doppler signal DS by the band-pass filter 42. The signals extracted from the Doppler signal DS by the low-pass filter 40, the band-pass filter 42, and the high-pass filter 44 are respectively input to the envelope processing unit 46.

[0051] The envelope processing unit 46 detects the envelope of the Doppler signal DS after each filter processing and converts the Doppler signal DS after each filter processing into an envelope signal by performing an absolute value process on the input signal. Each envelope signal is respectively input to the threshold processing unit 48.

[0052] Since the signal intensity of each envelope signal changes over time, the threshold processing unit 48 outputs "1" when the signal intensity of the envelope signal is greater than or equal to a predetermined output threshold, and outputs "0" when the signal intensity of the envelope signal is less than the output threshold. Here, when the signal intensity of the envelope signal changes from being greater than or equal to the output threshold to less than the output threshold, the threshold processing unit 48 does not immediately change the output to "0", but maintains the output at "1" for a predetermined dead time after the signal intensity becomes less than the output threshold, and then changes the output to "0".

[0053] Figure 6 is a conceptual diagram showing the output of the thresholding unit 48. The first time interval in which the thresholding unit 48a, which receives the heart wall component extracted by the low-pass filter 40, outputs "1" represents the time interval in which the heart wall is in motion. The second time interval in which the thresholding unit 48b, which receives the heart wall component extracted by the band-pass filter 42, outputs "1" represents the time interval in which both the heart wall and heart valves are in motion. The third time interval in which the thresholding unit 48c, which receives the heart valve component extracted by the high-pass filter 44, outputs "1" represents the time interval in which the heart valve component is in motion. The outputs of thresholding units 48a to 48c are added together by the adder 50 and input to the hysteresis unit 52.

[0054] The hysteresis processing unit 52, when the summation signal obtained by adding the outputs of threshold processing units 48a to c changes from "1" to "0", does not immediately set the summation signal to "0", but rather maintains the summation signal as "1" for a predetermined dead period from the moment it changes from "1" to "0" before setting it to "0".

[0055] The summed signal after processing by the hysteresis processing unit 52 is a signal like flag F shown in Figure 6. The time interval from the rising to falling edge of flag F represents the period during which the fetal heart wall or heart valves are moving in one heartbeat. However, the heart wall and heart valves are not constantly moving during the duration of one heartbeat; there is a period at the end of that period during which the movement of the heart wall and heart valves stops or becomes very small. However, since this period is short, the time interval indicated by flag F roughly represents the duration of one fetal heartbeat.

[0056] Therefore, the processor of the learning device can obtain a Doppler signal DS equivalent to approximately one heartbeat by extracting a time interval indicated by flag F from the Doppler signal, which is not segmented by heartbeat. If the coarsely segmented Doppler signal S input to the low-pass filter 40, band-pass filter 42, and high-pass filter 44 consists of a portion composed of fetal heartbeat components, the processor of the learning device assigns a label to the acquired Doppler signal DS equivalent to one heartbeat indicating that it consists of fetal heartbeat components. On the other hand, if the coarsely segmented Doppler signal S input to the low-pass filter 40, band-pass filter 42, and high-pass filter 44 consists of a portion composed of fetal heartbeat components and artifact components, the processor of the learning device assigns a label to the acquired Doppler signal DS equivalent to one heartbeat indicating that it consists of fetal heartbeat components and artifact components.

[0057] By repeating the above process, training data consisting of multiple single-heartbeat Doppler signals (DSA) and multiple single-heartbeat Doppler signals (DSB) is generated.

[0058] Preferably, timing labels indicating the timing of movement of at least one of the cardiac wall or heart valves are assigned to the Doppler signal DSA, which is composed of fetal heart rate components and serves as training data. In this case, it is necessary that the fetal electrocardiogram measured by the fetal electrocardiograph at the time the Doppler signal was acquired is associated with and stored in the Doppler signal database for each Doppler signal.

[0059] Figure 7 is a conceptual diagram illustrating the process of assigning timing labels to training data. As described above, the first time interval in which the threshold processing unit 48a, which receives the cardiac wall component extracted by the low-pass filter 40 as input, outputs "1" represents the time interval in which the cardiac wall is in motion. Therefore, the processor of the training processing unit identifies the type of fetal cardiac wall movement in the first time interval based on the relationship between a predetermined timing (e.g., R wave) shown in the fetal electrocardiogram and the timing of the first time interval. For example, the first time interval indicated by the code f1 shown in Figure 7 can be identified as the period in which the atrial wall is contracting. Such identification processing may be performed manually by the training data generator.

[0060] Furthermore, the third time interval in which the threshold processing unit 48c, which receives the heart valve component extracted by the high-pass filter 44, outputs "1" represents the time interval in which the heart valves are in motion. Therefore, the processor of the learning processing unit identifies the type of fetal heart valve movement in the third time interval based on the relationship between the predetermined timing shown by the fetal electrocardiogram and the timing of the third time interval. For example, the third time interval indicated by the code f2 in Figure 7 can be identified as the period in which the mitral valve is open, and the third time interval indicated by the code f3 can be identified as the period in which the mitral valve is closed and the aortic valve is open. Such identification processing may be performed manually by the learning data generator.

[0061] The processor of the learning processing unit assigns timing labels to the Doppler signal DPA, which is used as training data, indicating the timing of the movement of at least one of the heart wall or heart valve, as identified above.

[0062] <Training methods for learning models> The processor of the learning device uses the training data generated as described above to train the learning model 32 to predict and output whether the input Doppler signal consists of fetal heart rate components or fetal heart rate components and artifact components. Specifically, the processor of the learning device inputs a Doppler signal DSA (see Figure 2) or a Doppler signal DSB (see Figure 3) as training data to the learning model 32. The learning model 32 predicts and outputs whether the input Doppler signal DSA or Doppler signal DSB consists of fetal heart rate components or fetal heart rate components and artifact components. The processor of the learning device modifies the parameters within the learning model 32 so that the difference between the prediction result of the learning model 32 and the label attached to the input Doppler signal DSA or Doppler signal DSB becomes smaller. By repeating this process, the learning model 32 is trained. A well-trained model 32 can accurately predict whether the input Doppler signal consists of fetal heart rate components or fetal heart rate components and artifact components.

[0063] As described above, if timing labels are assigned to the Doppler signal DSA as training data, the training model 32, if it predicts that the input Doppler signal DSA consists of fetal heart rate components, further predicts and outputs the timing of the movement of at least one of the heart wall or heart valve in the Doppler signal DSA. The processor of the training processing device modifies the parameters within the training model 32 so that the difference between the training model 32's predicted timing of the movement and the timing labels assigned to the input Doppler signal DSA becomes smaller. By performing this training, the trained training model 32 can accurately predict the timing of the movement of at least one of the heart wall or heart valve in the Doppler signal DS when it predicts that the input Doppler signal DS consists of fetal heart rate components.

[0064] Furthermore, the learning model 32 may be trained using training data that further includes a heart diameter parameter indicating the fetal heart diameter. The heart diameter parameter indicating the fetal heart diameter is, for example, the fetal gestational age or the fetal heart diameter. In this case, it is necessary that the Doppler signal database stores a heart diameter parameter corresponding to each Doppler signal (a parameter indicating the heart diameter of the fetal heart from which ultrasound was transmitted when the Doppler signal was acquired). The processor of the learning processing device inputs the corresponding heart diameter parameter along with the Doppler signal DSA or Doppler signal DSB as training data to the learning model 32. The learning model 32 predicts and outputs whether the input Doppler signal DSA or Doppler signal DSB consists of fetal heart rate components or consists of fetal heart rate components and artifact components, taking the input heart diameter parameter into consideration. The processor of the learning processing device modifies the parameters within the learning model 32 so that the difference between the prediction result of the learning model 32 and the label attached to the input Doppler signal DSA or Doppler signal DSB becomes smaller.

[0065] Generally, the diameter of a fetus's heart increases with gestational age. As the diameter of the fetus's heart increases, its movement speed increases. Therefore, the larger the diameter of the fetus's heart, the stronger the signal intensity of the fetal heart rate component in the Doppler signal DS tends to be, and the smaller the diameter of the fetus's heart, the weaker the signal intensity of the fetal heart rate component tends to be. On the other hand, the signal intensity of the artifact component in the Doppler signal DS does not change with fetal gestational age. For this reason, when the diameter of the fetus's heart is small, the input Doppler signal DS tends to be predicted to consist of the fetal heart rate component and artifact component, and when the diameter of the fetus's heart is large, the input Doppler signal DS tends to be predicted to consist of the fetal heart rate component. By training the learning model 32 with training data that includes the heart diameter parameter, it is possible to predict whether the input Doppler signal DS consists of the fetal heart rate component or the fetal heart rate component and artifact component, taking the diameter of the fetus's heart into consideration. This further improves the prediction accuracy of the learning model 32.

[0066] <Doppler signal identification processing> Next, the Doppler signal identification process by the Doppler signal identification unit 20 will be explained. In the following explanation, the Doppler signal that is the target of the identification process by the Doppler signal identification unit 20 will be referred to as the target Doppler signal. The Doppler signal identification unit 20 inputs the target Doppler signal to the trained training model 32 (hereinafter simply referred to as "trained model 32") and identifies whether the target Doppler signal consists of fetal heart rate components or fetal heart rate components and artifact components based on the output data of the training model 32 for the input.

[0067] Figure 8 is a conceptual diagram showing the unit Doppler signal DSU. First, the Doppler signal identification unit 20 obtains the unit Doppler signal DSU by extracting the target Doppler signal TDS using a time window TW of a predetermined time length. The time length of the time window TW may be set as appropriate.

[0068] The Doppler signal identification unit 20 acquires multiple unit Doppler signals DSU by shifting the time window TW every unit time t1. In the example in Figure 8, unit Doppler signal DSU1 is acquired by extracting the target Doppler signal TDS using time window TW1, unit Doppler signal DSU2 is acquired by extracting the target Doppler signal TDS using time window TW2 which is shifted by unit time t1 from time window TW1, and unit Doppler signal DSU3 is acquired by extracting the target Doppler signal TDS using time window TW3 which is shifted by unit time t1 from time window TW2.

[0069] The unit time t1 may be set as appropriate by the designer, administrator, or user of the fetal heart rate measurement device 10. For example, the unit time t1 may be the time for one sample at the sampling frequency of the AD converter of the Doppler signal generation unit 18.

[0070] As shown in Figure 9, the Doppler signal identification unit 20 sequentially inputs the acquired unit Doppler signals DSUs to the learning model 32 in the order they were acquired. The learning model 32 predicts and outputs whether the input unit Doppler signal DSU consists of fetal heart rate components or consists of fetal heart rate components and artifact components. Based on the output of the learning model 32, the Doppler signal identification unit 20 identifies whether each unit Doppler signal DSU consists of fetal heart rate components or consists of fetal heart rate components and artifact components.

[0071] If the learning model 32 is trained using training data in which timing labels indicating the timing of the behavior of the fetal heart wall or heart valves in relation to the Doppler signal DSA (see Figure 2), then if the learning model 32 predicts that the input unit Doppler signal DSU consists of fetal heart rate components, it further predicts and outputs the timing of the behavior of the fetal heart wall or heart valves in the unit Doppler signal DSU. Based on the output of the learning model 32, the Doppler signal identification unit 20 further identifies the timing of the behavior in the unit Doppler signal DSU if the unit Doppler signal DSU consists of fetal heart rate components.

[0072] Furthermore, if the learning model 32 has been trained using training data that also includes a heart diameter parameter indicating the diameter of the fetus's heart, the Doppler signal identification unit 20 inputs a unit Doppler signal DSU and the corresponding heart diameter parameter (a parameter indicating the diameter of the fetus's heart from which ultrasound was transmitted when the unit Doppler signal DSU was acquired) to the learning model 32. The learning model 32 predicts and outputs whether the input unit Doppler signal DSU consists of fetal heart rate components or consists of fetal heart rate components and artifact components, while referring to the heart diameter parameter. Based on the output of the learning model 32, the Doppler signal identification unit 20 identifies whether each unit Doppler signal DSU consists of fetal heart rate components or consists of fetal heart rate components and artifact components.

[0073] <Measuring fetal heart rate> The heart rate measurement unit 22 measures the fetal heart rate based on the target Doppler signal TDS, which has been identified by the Doppler signal identification unit 20 as consisting of fetal heart rate components.

[0074] As a first method for measuring the fetal heart rate, the heart rate measurement unit 22 directly measures the fetal heart rate from the identification results of the Doppler signal identification unit 20 for each of a plurality of unit Doppler signals DSU arranged in time series. Specifically, the heart rate measurement unit 22 measures the fetal heart rate based on the time difference between a motion period, which is a period corresponding to a block of unit Doppler signals DSU in the time direction that has been identified by the Doppler signal identification unit 20 as consisting of fetal heart rate components, and the next motion period in the time series. This will be explained in detail with reference to Figure 10.

[0075] Figure 10 shows the target Doppler signal TDS and the time window TW set by the Doppler signal identification unit 20. In the example in Figure 10, the Doppler signal identification unit 20 identifies that the unit Doppler signal DSU corresponding to time window TW4 (extracted by time window TW4) consists of fetal heart rate components and artifact components, multiple unit Doppler signal DSUs corresponding to multiple time windows TW5 to TW6 are identified as consisting of fetal heart rate components, and the unit Doppler signal DSU corresponding to time window TW7 is identified as consisting of fetal heart rate components and artifact components. In this case, the period from the start of time window TW5 to the end of time window TW6 becomes the exercise period MP. The exercise period MP may be the period of a single unit Doppler signal DSU, rather than the period corresponding to multiple unit Doppler signal DSUs. That is, a block of unit Doppler signal DSUs in the time direction identified by the Doppler signal identification unit 20 as consisting of fetal heart rate components may be a single unit Doppler signal DSU. In this way, depending on the identification result of the Doppler signal identification unit 20 for each unit Doppler signal DSU, multiple motion periods MP can be defined, as shown in Figure 10.

[0076] A single exercise period MP is the period within a heartbeat during which the heart wall or heart valves are moving, and the period between exercise periods MP is the period within a heartbeat during which the movement of the heart wall and heart valves ceases or decreases. Therefore, the heart rate measurement unit 22 can use the time difference t2 between exercise periods MPs as the time of one heartbeat in the fetus. In this embodiment, the time difference t2 between exercise periods MPs is defined as the time between the start of one exercise period MP and the start of the next exercise period MP in the time series. The heart rate measurement unit 22 can determine time t2 using the time difference between the time windows TW that serve as the start points for both exercise periods MPs. For example, as shown in Figure 10, if the time window TW that serves as the start point for exercise period MP1 is time window TW5, and the time window TW that serves as the start point for the next exercise period MP2 is time window TW8, then the time between the start of time window TW5 and the start of time window TW8 is time t2. The time difference t2 between motion periods MP may also be defined as the time between the end point of one motion period MP and the end point of the next motion period MP in the time series.

[0077] The heart rate measurement unit 22 obtains the fetal heart rate per second by calculating the reciprocal of time t2. The heart rate measurement unit 22 also obtains the heart rate per minute (bpm) by multiplying the fetal heart rate per second by 60. As the Doppler signal identification unit 20 sequentially identifies the unit Doppler signal DUS in the order it is acquired, the heart rate measurement unit 22 obtains time t2 by sequentially identifying the exercise period MP and sequentially calculates the fetal heart rate per second.

[0078] As a second method for measuring the fetal heart rate, the heart rate measurement unit 22 measures the fetal heart rate by autocorrelation processing using the identification result of the unit Doppler signal DSU by the Doppler signal identification unit 20.

[0079] Figure 11 is a conceptual diagram showing the comparison portion CP used in autocorrelation processing, and Figure 12 is a conceptual diagram showing the reference signal RDS used in autocorrelation processing. The heart rate measurement unit 22 calculates the correlation coefficient (hereinafter sometimes simply referred to as "correlation coefficient") between the reference signal RDS and the comparison portion CP, which is the portion of the target Doppler signal TDS with the same time length as the reference signal RDS. As will be described later, the reference signal RDS can be updated, but the initial reference signal RDS is pre-prepared to show the fetal heart rate component.

[0080] The heart rate measurement unit 22 may, from the viewpoint of reducing the influence of noise, not calculate a correlation coefficient between the reference signal RDS itself and the comparison target portion CP itself, but rather compare the portion of the comparison target portion CP that has a signal intensity equal to or greater than a predetermined signal intensity threshold Thr with the portion of the reference signal RDS that has a signal intensity equal to or greater than the signal intensity threshold Thr. In this case, from the viewpoint of reducing the amount of computation required for comparison, the heart rate measurement unit 22 may downsample the extraction of the portion of the comparison target portion CP or the reference signal RDS that has a signal intensity equal to or greater than the signal intensity threshold Thr. For example, if the sampling frequency of the AD converter of the Doppler signal generation unit 18 is 2 kHz, the heart rate measurement unit 22 may extract the portion of the comparison target portion CP or the reference signal RDS that has a signal intensity equal to or greater than the signal intensity threshold Thr at a sampling frequency of 100 Hz.

[0081] The heart rate measurement unit 22 repeatedly calculates the correlation coefficient while gradually shifting the comparison portion CP in the time direction (for example, by the time of one sample at the sampling frequency of the AD converter of the Doppler signal generation unit 18). By calculating the correlation coefficient while shifting the comparison portion CP in the time direction, the time change of the correlation coefficient is calculated.

[0082] Figure 13 is a graph showing the time evolution of the correlation coefficient. Since the reference signal RDS represents the fetal heart rate component, the timing when the correlation coefficient is large (for example, when it takes a maximum value) indicates the timing when the fetal heart wall or heart valves are moving in the same way as indicated by the reference signal RDS. From this, it can be said that the period of the time evolution of the correlation coefficient represents the fetal heart rate. Therefore, the heart rate measurement unit 22 measures the fetal heart rate based on the period of the time evolution of the correlation coefficient. For example, the time t3 between the maximum values ​​of the correlation coefficient can be said to represent the time between one fetal heartbeat, so the heart rate measurement unit 22 obtains the fetal heart rate per second by calculating the reciprocal of time t3.

[0083] Traditionally, the reference signal RDS has been updated during autocorrelation processing. Specifically, when the correlation coefficient satisfies predetermined conditions (for example, when the correlation coefficient reaches a maximum value, or when the correlation coefficient exceeds a predetermined correlation coefficient), the comparison portion CP corresponding to that correlation coefficient is set as the new reference signal RDS. By updating the reference signal RDS, it is possible to suppress the decrease in the accuracy of fetal heart rate calculations, even when the signal waveform of the fetal heart rate component changes gradually.

[0084] In this embodiment, the heart rate measurement unit 22 does not use the comparison target portion CP as a new reference signal RDS if the Doppler signal identification unit 20 determines that at least a portion of the comparison target portion CP whose correlation coefficient satisfies the predetermined conditions is composed of fetal heart rate components and artifact components. In Figure 10, the portion of the target Doppler signal TDS corresponding to the exercise period MP is the portion determined to be composed of fetal heart rate components, and the portion of the target Doppler signal TDS corresponding to periods other than the exercise period MP is the portion determined to be composed of fetal heart rate components and artifact components. Therefore, if at least a portion of the comparison target portion CP whose correlation coefficient satisfies the predetermined conditions includes periods other than the exercise period MP, the heart rate measurement unit 22 does not use the comparison target portion CP as a new reference signal RDS.

[0085] Even if the comparison portion CP contains many artifact components, there may be cases where the correlation coefficient between it and the reference signal RDS happens to be large. In such cases, if the comparison portion CP is used as the reference signal RDS, the signal containing many artifact components becomes the reference signal RDS. Consequently, even if the comparison portion CP is truly composed of fetal heart rate components thereafter, the correlation coefficient between the comparison portion CP and the reference signal RDS will not be large because the reference signal RDS contains many artifact components. This prevents the fetal heart rate from being calculated correctly. According to this embodiment, if at least a portion of the comparison portion CP whose correlation coefficient satisfies predetermined conditions is determined by the Doppler signal identification unit 20 to be composed of fetal heart rate components and artifact components, the comparison portion CP is not used as the new reference signal RDS, thereby reducing the occurrence of such problems.

[0086] In the second measurement method, the heart rate measurement unit 22 uses the comparison portion CP as a new reference signal RD if the Doppler signal identification unit 20 determines that at least a portion of the comparison portion CP whose correlation coefficient satisfies the above predetermined conditions is composed of fetal heart rate components. Therefore, it can be said that in the second measurement method as well, the heart rate measurement unit 22 measures the fetal heart rate based on the target Doppler signal TDS which has been identified by the Doppler signal identification unit 20 as being composed of fetal heart rate components.

[0087] In both of the first and second measurement methods described above, the heart rate measurement unit 22 measures the fetal heart rate based on the target Doppler signal TDS identified by the Doppler signal identification unit 20 as consisting of fetal heart rate components. In other words, the heart rate measurement unit 22 measures the fetal heart rate by excluding the portion of the target Doppler signal TDS identified as consisting of fetal heart rate components and artifact components. This reduces the influence of artifact components when measuring the fetal heart rate, thereby improving the accuracy of fetal heart rate measurement based on the target Doppler signal TDS.

[0088] As described above, the information indicating the fetal heart rate measured by the heart rate measurement unit 22 is output to the display 26 (or speaker) by the output control unit 24. Furthermore, if the Doppler signal identification unit 20 identifies the timing of the behavior of the fetal heart wall or heart valve in the unit Doppler signal DSU composed of fetal heart rate components, the output control unit 24 further displays information indicating that behavior timing on the display 26.

[0089] <Switching operating modes> The operating mode switching unit 34 switches the operating mode of the fetal heart rate measurement device 10 between the first operating mode and the second operating mode in response to instructions from the user. The first operating mode is, as described above, an operating mode in which the heart rate measurement unit 22 measures the fetal heart rate based on the target Doppler signal TDS identified by the Doppler signal identification unit 20 as being composed of fetal heart rate components. On the other hand, the second operating mode is an operating mode in which the fetal heart rate is measured by so-called conventional autocorrelation processing. That is, in the second operating mode, the heart rate measurement unit 22 calculates the correlation coefficient between the reference signal RDS and the comparison target portion CP, which is the portion of the target Doppler signal TDS with the same time length as the reference signal RDS, without using the identification result by the Doppler signal identification unit 20, and performs autocorrelation processing to calculate the time change of the correlation coefficient obtained by shifting the comparison target portion CP in the time direction, and measures the fetal heart rate based on the period of the time change of the correlation coefficient. In the second operating mode, regardless of whether at least a portion of the comparison portion CP that satisfies the predetermined conditions consists of the fetal heart rate component or the fetal heart rate component and artifact component, the comparison portion CP is used as a new reference signal RDS.

[0090] <Overview of the Fetal Heart Rate Measurement System> Figure 14 is a schematic diagram of the configuration of the fetal heart rate measurement system 60 according to this embodiment. In the fetal heart rate measurement device 10 shown in Figure 1, the process of forming the target Doppler signal TDS, the identification process of identifying whether the target Doppler signal TDS consists of fetal heart rate components or fetal heart rate components and artifact components, and the measurement process of measuring the fetal heart rate were performed within a single device, but each of these processes may be performed in different devices.

[0091] The fetal heart rate measurement system 60 shown in Figure 14 comprises a Doppler signal generator 62 and a fetal heart rate measurement device 64. Although only one Doppler signal generator 62 is shown in Figure 14, the fetal heart rate measurement system 60 may have multiple Doppler signal generators 62. The Doppler signal generator 62 and the fetal heart rate measurement device 64 are connected to each other via a communication line 66 such as a LAN (Local Area Network) or WAN (Wide Area Network) to enable communication.

[0092] The Doppler signal generator 62 is installed in a hospital examination room or delivery room, while the fetal heart rate monitor 64 is installed in a different location within the hospital, such as a nurses' center or central control room. Alternatively, the Doppler signal generator 62 may be installed, for example, at the pregnant woman's home, and the fetal heart rate monitor 64 may be located on a server remotely from the pregnant woman's home. In this case, the functions performed by the fetal heart rate monitor 64 may be provided through a cloud service.

[0093] The Doppler signal forming apparatus 62 includes an ultrasonic probe 12, a transmitting unit 14, a receiving unit 16, a Doppler signal forming unit 18, and a communication unit 68. The configuration and function of the ultrasonic probe 12, transmitting unit 14, receiving unit 16, and Doppler signal forming unit 18 in the Doppler signal forming apparatus 62 are the same as those in Figure 1, so a detailed explanation is omitted.

[0094] The communication unit 68 is composed of, for example, a network card. The communication unit 68, acting as a Doppler signal transmission unit, transmits the target Doppler signal TDS formed by the Doppler signal formation unit 18 to the fetal heart rate measuring device 64 via the communication line 66.

[0095] The fetal heart rate measurement device 64 includes a communication unit 70, a Doppler signal identification unit 20, a heart rate measurement unit 22, an output control unit 24, a display 26, an input interface 28, a memory 30, and an operating mode switching unit 34. The configuration and functions of the Doppler signal identification unit 20, heart rate measurement unit 22, output control unit 24, display 26, input interface 28, memory 30, and operating mode switching unit 34 in the fetal heart rate measurement device 64 are the same as those of the Doppler signal identification unit 20, heart rate measurement unit 22, output control unit 24, display 26, input interface 28, memory 30, and operating mode switching unit 34 in Figure 1, so a detailed explanation is omitted.

[0096] The communication unit 70 is composed of, for example, a network card. The communication unit 70 receives the target Doppler signal TDS from the Doppler signal forming device 62 via the communication line 66.

[0097] Thus, in the fetal heart rate measurement system 60, the fetal heart rate measurement device 64 identifies whether the target Doppler signal TDS received from the Doppler signal generator 62 via the communication line 66 consists of fetal heart rate components or consists of fetal heart rate components and artifact components, and then measures the fetal heart rate based on the target Doppler signal TDS identified as consisting of fetal heart rate components.

[0098] The Doppler signal generator 62 may also have a display and a speaker, and information indicating the fetal heart rate measured by the fetal heart rate measuring device 64 may be sent to the Doppler signal generator 62, and said information may be output by the Doppler signal generator 62.

[0099] Furthermore, in the fetal heart rate measurement system 60, the identification process of the target Doppler signal TDS and the measurement process for measuring the fetal heart rate are performed in a single device, but the identification process and the measurement process may be performed in different devices.

[0100] The fetal heart rate measurement system 60 may further have a storage device 72. The Doppler signal generating device 62 and the fetal heart rate measurement device 64 are able to access the storage device 72 via a communication line 66. In this case, the Doppler signal generating device 62 stores the generated target Doppler signal TDS in the storage device 72. The fetal heart rate measurement device 64 reads the target Doppler signal TDS from the storage device 72 and performs identification and measurement processing. This makes it possible to measure the fetal heart rate corresponding to the target Doppler signal TDS based on the target Doppler signal TDS that has been previously formed and stored.

[0101] Although the fetal heart rate measurement device, fetal heart rate measurement system, and fetal heart rate measurement program according to this embodiment have been described above, the fetal heart rate measurement device, fetal heart rate measurement system, and fetal heart rate measurement program according to this embodiment are not limited to the above embodiment, and various modifications are possible without departing from the spirit thereof. [Explanation of Symbols]

[0102] 10,64 Fetal heart rate measurement device, 12 Ultrasound probe, 14 Transmitter, 16 Receiver, 18 Doppler signal generation unit, 20 Doppler signal identification unit, 22 Heart rate measurement unit, 24 Output control unit, 26 Display, 28 Input interface, 30 Memory, 32 Learning model, 40 Low-pass filter, 42 Band-pass filter, 44 High-pass filter, 46 Envelope processing unit, 48 Threshold processing unit, 50 Adder, 52 Hysteresis processing unit, 60 Fetal heart rate measurement system, 62 Doppler signal generation device, 66 Communication line, 68,70 Communication unit, 72 Storage device.

Claims

1. A fetal heart rate measuring device that measures the fetal heart rate based on a Doppler signal representing the difference frequency between the frequency of ultrasound transmitted toward the fetal heart and the frequency of the reflected ultrasound wave, A Doppler signal identification unit identifies whether the target Doppler signal is composed of a fetal heart rate component or composed of a fetal heart rate component and artifact components, based on the output data of the learning model when the target Doppler signal, which is the Doppler signal to be processed, is input to a learning model that has been trained to predict and output whether the input Doppler signal is composed of a fetal heart rate component or composed of a fetal heart rate component and artifact components, using the Doppler signal composed of a fetal heart rate component and the Doppler signal composed of a fetal heart rate component and artifact components as training data. A heart rate measurement unit that measures the fetal heart rate based on the target Doppler signal identified as consisting of fetal heart rate components, A fetal heart rate measuring device characterized by comprising the following:

2. The Doppler signal identification unit extracts unit Doppler signals from the target Doppler signal using a predetermined time window, and sequentially inputs the multiple unit Doppler signals obtained by shifting the time window by unit time into the learning model in the order they were acquired, thereby identifying whether each unit Doppler signal consists of a fetal heart rate component or consists of a fetal heart rate component and an artifact component. The heart rate measurement unit measures the fetal heart rate based on the time difference between a motion period, which is a period corresponding to a block of unit Doppler signals in the time direction that has been identified by the Doppler signal identification unit as consisting of fetal heart rate components, and the next motion period in the time series. The fetal heart rate measuring device according to feature 1.

3. The Doppler signal identification unit extracts unit Doppler signals from the target Doppler signal using a predetermined time window, and sequentially inputs the multiple unit Doppler signals obtained by shifting the time window by unit time into the learning model in the order they were acquired, thereby identifying whether each unit Doppler signal consists of a fetal heart rate component or consists of a fetal heart rate component and an artifact component. The aforementioned heart rate measurement unit is The correlation coefficient between a reference signal and a comparison portion of the target Doppler signal that has the same time duration as the reference signal is calculated. An autocorrelation process is performed to calculate the time change of the correlation coefficient obtained by shifting the comparison portion in the time direction while using the comparison portion whose correlation coefficient satisfies predetermined conditions as the new reference signal, and the fetal heart rate is measured based on the period of the time change of the correlation coefficient. If the correlation coefficient satisfies the predetermined conditions and at least a portion of the comparison target is determined by the Doppler signal identification unit to consist of fetal heart rate components and artifact components, then the comparison target will not be used as a new reference signal. The fetal heart rate measuring device according to feature 1.

4. The learning model is trained to predict and output the timing of the behavior in the input Doppler signal, using the training data which has timing labels indicating the timing of the movement of at least one of the heart wall or heart valves in relation to the Doppler signal composed of fetal heart rate components. The Doppler signal identification unit identifies the behavioral timing in the target Doppler signal based on the output data of the learning model when the target Doppler signal is input, if the target Doppler signal consists of fetal heart rate components. The fetal heart rate measuring device according to feature 1.

5. The aforementioned learning model is trained using the aforementioned learning data, which further includes a cardiac diameter parameter indicating the fetal heart diameter. The Doppler signal identification unit identifies whether the target Doppler signal consists of a fetal heart rate component or a fetal heart rate component and an artifact component, based on the target Doppler signal and the output data of the learning model when the heart diameter parameter corresponding to the target Doppler signal is input. The fetal heart rate measuring device according to feature 1.

6. An operating mode switching unit switches the operating mode of the fetal heart rate measuring device between a first operating mode in which the heart rate measuring unit measures the fetal heart rate based on the target Doppler signal identified by the Doppler signal identification unit as consisting of fetal heart rate components, and a second operating mode in which the heart rate measuring unit performs autocorrelation processing to calculate the time change of the correlation coefficient obtained by shifting the comparison target portion in the time direction, without using the identification result by the Doppler signal identification unit, and measures the fetal heart rate based on the period of the time change of the correlation coefficient, and The fetal heart rate measuring device according to claim 1, further comprising the above.

7. The Doppler signal included in the training data is a Doppler signal representing the duration of time during which the heart wall or heart valve is moving in a single heartbeat, extracted from the Doppler signal which is not divided by heartbeat, based on a first time interval from which the signal is extracted by applying a low-pass filter to the Doppler signal to extract low-frequency components corresponding to the movement of the fetal heart wall, a second time interval from which the signal is extracted by applying a band-pass filter to the Doppler signal to extract frequency components corresponding to the movement of the fetal heart wall and heart valve, and a third time interval from which the signal is extracted by applying a high-pass filter to the Doppler signal which extracts high-frequency components corresponding to the movement of the fetal heart valve. The fetal heart rate measuring device according to feature 1.

8. Doppler signal shaping apparatus, Fetal heart rate monitor and Equipped with, The Doppler signal forming apparatus is An ultrasound probe that transmits ultrasound waves toward the fetal heart and receives the reflected waves of the ultrasound waves, A Doppler signal forming unit that forms a target Doppler signal, which is a Doppler signal to be processed, representing the difference frequency between the frequency of the ultrasound transmitted to the fetus and the frequency of the reflected wave. A Doppler signal transmission unit that transmits the target Doppler signal to the fetal heart rate measuring device via a communication line, Includes, The aforementioned fetal heart rate measuring device is A Doppler signal identification unit identifies whether the target Doppler signal is composed of a fetal heart rate component or composed of a fetal heart rate component and artifact components, based on the output data of the learning model when the target Doppler signal is input, using the Doppler signal composed of a fetal heart rate component and the Doppler signal composed of a fetal heart rate component and artifact components as learning data, A heart rate measurement unit that measures the fetal heart rate based on the target Doppler signal identified as consisting of fetal heart rate components, including, A fetal heart rate measurement system characterized by the following:

9. A fetal heart rate measurement program that operates a computer as a fetal heart rate measurement device to measure the fetal heart rate based on a Doppler signal representing the difference frequency between the frequency of ultrasound transmitted toward the fetal heart and the frequency of the reflected ultrasound wave, The aforementioned computer, A Doppler signal identification unit identifies whether the target Doppler signal is composed of a fetal heart rate component or composed of a fetal heart rate component and artifact components, based on the output data of the learning model when the target Doppler signal, which is the Doppler signal to be processed, is input to a learning model that has been trained to predict and output whether the input Doppler signal is composed of a fetal heart rate component or composed of a fetal heart rate component and artifact components, using the Doppler signal composed of a fetal heart rate component and the Doppler signal composed of a fetal heart rate component and artifact components as training data. A heart rate measurement unit that measures the fetal heart rate based on the target Doppler signal identified as consisting of fetal heart rate components, A fetal heart rate measurement program characterized by operating as such.

Images