Blood pressure information estimation device, blood pressure information estimation method, and blood pressure information estimation program
Patent Information
- Authority / Receiving Office
- JP · JP
- Patent Type
- Patents
- Current Assignee / Owner
- CITIZEN WATCH CO LTD
- Filing Date
- 2022-03-18
- Publication Date
- 2026-07-31
Smart Images
Figure 0007898431000003 
Figure 0007898431000004 
Figure 0007898431000005
Abstract
Description
[Technical Field]
[0001] This invention relates to a blood pressure information estimation device, a blood pressure information estimation method, and a blood pressure information estimation program. [Background technology]
[0002] The most common method for measuring blood pressure in humans is the cuff-type blood pressure monitor, which involves wrapping a cuff around the upper arm. However, forcing subjects to wrap a cuff is undesirable when simply checking blood pressure. Therefore, so-called cuffless blood pressure monitors, which allow for easy blood pressure measurement, have been reported.
[0003] A known cuffless blood pressure estimation method involves detecting pulse waves from images of different parts of the human body, calculating the pulse wave propagation velocity based on the time difference between the pulse waves, and estimating blood pressure from the pulse wave propagation velocity (for example, Patent Document 1). In the method described in Patent Document 1, in order to make it easier to identify the time difference in pulse wave propagation, parts of the body that can be obtained at a predetermined distance, such as the face and hand, are used, and blood pressure is estimated from a time difference of about 100 milliseconds. [Prior art documents] [Patent Documents]
[0004] [Patent Document 1] International Publication No. 2014 / 136310 [Overview of the project]
[0005] However, conventional methods using facial images for measurement have drawbacks, such as the inability to accurately read color information from the skin surface due to makeup, or the limited area of skin surface exposed due to the face being covered by a mask, making it difficult to obtain an accurate pulse wave waveform.
[0006] The invention according to the embodiments of this disclosure aims to provide a blood pressure information estimation device that can easily estimate blood pressure information.
[0007] The blood pressure information estimation device according to the embodiment of this disclosure is characterized by comprising: a video acquisition unit that acquires a video image of a predetermined area at a first frame rate in order to detect the delay time of pulse waves in a first area and a second area separated by a predetermined distance within a predetermined area of a living body; a video playback unit that plays back the acquired video image at a second frame rate lower than the first frame rate; a pulse wave extraction unit that extracts a first pulse wave in the first area and a second pulse wave in the second area based on the played-back video image; a pulse wave propagation velocity calculation unit that calculates the pulse wave propagation velocity from the time difference and predetermined distance between the peaks of the first pulse wave and the second pulse wave; a blood pressure information estimation unit that estimates blood pressure information based on the calculated pulse wave propagation velocity; and an output unit that outputs the estimated blood pressure information.
[0008] In the blood pressure information estimation device according to the embodiment of this disclosure, it is preferable to further include a video conversion unit that converts acquired video images into uncompressed video images.
[0009] In the blood pressure information estimation device according to the embodiment of this disclosure, it is preferable to further include a filter control unit that controls the bandwidth of a rectangular wave correlation filter for extracting a first pulse wave and a second pulse wave from an acquired video image.
[0010] In the blood pressure information estimation device according to the embodiment of the present disclosure, it is preferable that the video acquisition unit further includes an image recognition unit that recognizes an image of a predetermined part from a video image acquired by the video acquisition unit at a second frame rate before the video acquisition unit acquires a video image captured at a first frame rate.
[0011] In the blood pressure information estimation device according to the embodiment of this disclosure, it is preferable that the image recognition unit recognizes an image of a predetermined area from the acquired video image using a trained image recognition model.
[0012] In the blood pressure information estimation device according to the embodiment of the present disclosure, it is preferable that the video acquisition unit further includes an image determination unit that determines whether or not a pulse wave can be detected from the video image acquired by the video acquisition unit at a second frame rate, before the video acquisition unit acquires the video image captured at a first frame rate.
[0013] In the blood pressure information estimation device according to the embodiment of this disclosure, it is preferable that the image determination unit issues a warning if it cannot detect a pulse wave from the acquired video image.
[0014] In the blood pressure information estimation device according to the embodiment of this disclosure, it is preferable that the blood pressure information includes information on at least one of blood pressure, vascular age, and arteriosclerosis.
[0015] In the blood pressure information estimation device according to the embodiment of this disclosure, the predetermined site is preferably the hand.
[0016] In the blood pressure information estimation device according to the embodiment of this disclosure, it is preferable to further include an input unit for inputting information about the gender and height of a subject whose blood pressure is to be measured, and a predetermined distance calculation unit for calculating a predetermined distance based on the arm length estimated from the gender information and height.
[0017] The blood pressure information estimation method according to the embodiment of the present disclosure is characterized in that a video acquisition unit acquires a video image of a predetermined area at a first frame rate in order to detect the delay time of pulse waves in a first area and a second area separated by a predetermined distance within a predetermined area of a living body; a video playback unit plays back the acquired video image at a second frame rate lower than the first frame rate; a pulse wave extraction unit extracts a first pulse wave in the first area and a second pulse wave in the second area based on the played-back video image; a pulse wave propagation velocity calculation unit calculates the pulse wave propagation velocity from the time difference and predetermined distance between the peaks of the first and second pulse waves; a blood pressure information estimation unit estimates blood pressure information based on the calculated pulse wave propagation velocity; and an output unit outputs the estimated blood pressure information.
[0018] The blood pressure information estimation program according to an embodiment of the present disclosure causes a computer to acquire a moving image obtained by imaging a predetermined site at a first frame rate in order to detect a delay time of a pulse wave in a first region and a second region spaced apart by a predetermined distance within the predetermined site of a living body, reproduce the acquired moving image at a second frame rate lower than the first frame rate, extract a first pulse wave in the first region and a second pulse wave in the second region based on the reproduced moving image, calculate a pulse wave propagation speed from a time difference between peaks of each of the first pulse wave and the second pulse wave and the predetermined distance, estimate blood pressure information based on the calculated pulse wave propagation speed, and output the estimated blood pressure information, characterized by executing each step.
[0019] According to the blood pressure information estimation device according to an embodiment of the present disclosure, blood pressure information can be easily estimated.
Brief Description of Drawings
[0020] [Figure 1] It is a diagram for explaining an outline of measurement using the blood pressure information estimation device according to the first embodiment of the present disclosure. [Figure 2] It is a block diagram showing the configuration of the blood pressure information estimation device according to the first embodiment of the present disclosure. [Figure 3] It is a flowchart for explaining an operation procedure of the blood pressure information estimation device according to the first embodiment of the present disclosure. [Figure 4] It is a graph showing an example of a pulse wave obtained by the blood pressure information estimation device according to the first embodiment of the present disclosure. [Figure 5] It is a graph showing the relationship between the pulse wave propagation speed and blood pressure. [Figure 6] It is a diagram for explaining an outline of measurement using the blood pressure information estimation device according to the second embodiment of the present disclosure. [Figure 7] It is a block diagram showing the configuration of the blood pressure information estimation device according to the second embodiment of the present disclosure. [Figure 8] It is a flowchart for explaining an operation procedure of the blood pressure information estimation device according to the second embodiment of the present disclosure. [Figure 9]This is a block diagram showing the configuration of a blood pressure information estimation device according to a third embodiment of this disclosure. [Figure 10A] This graph shows the relationship between height and arm length for Japanese men. [Figure 10B] This graph shows the relationship between height and arm length for Japanese women. [Figure 11] This figure shows the relationship between the subject's arm length and a predetermined distance. [Modes for carrying out the invention]
[0021] The blood pressure information estimation device, blood pressure information estimation method, and blood pressure information estimation program according to the present invention will be described below with reference to the drawings. However, it should be noted that the technical scope of the present invention is not limited to these embodiments, but extends to the invention described in the claims and its equivalents.
[0022] [First Embodiment] First, a blood pressure information estimation device according to the first embodiment of this disclosure will be described. Figure 1 shows a diagram illustrating the overview of measurement using the blood pressure information estimation device according to the first embodiment of this disclosure. In the blood pressure information estimation device 100 according to the first embodiment, the case in which blood pressure information such as blood pressure is estimated using an image of the hand 301 of a subject 300 will be described as an example. The hand 301 can be easily imaged using the camera of a mobile terminal 200, and unlike a person's face, it is not difficult to observe the color of the skin surface due to the effect of makeup or covered by a mask, etc. Furthermore, it is considered that it is less difficult to observe the color of the palm of the hand due to discoloration of the skin due to sunburn, etc. However, it is not limited to such an example, and if the skin color can be easily observed, blood pressure information may be estimated using a predetermined part other than the hand.
[0023] In the example shown in Figure 1, first, a video 400 of the subject's hand 301 is captured using a mobile terminal 200. The captured video 400 is displayed on the display unit 201 of the mobile terminal 200. The mobile terminal 200 transmits the data of the captured video 400 to another mobile terminal, a blood pressure information estimation device 100, via a communication network 1000. The blood pressure information estimation device 100 estimates the blood pressure information of the subject 300 from the received video data. Here, "video" refers to a sequence of images played back at regular time intervals. Therefore, even a collection of still images is included in a video if it is played back at regular time intervals.
[0024] A video image 400 of the subject's hand 301, captured by the mobile terminal 200, is transmitted to the blood pressure information estimation device 100 via wired or wireless communication. Figure 1 shows an example where the received video image 500 of the hand is displayed on the display unit 40 of the blood pressure information estimation device 100. However, the device is not limited to this example, and the video image 500 received by the blood pressure information estimation device 100 may be played back (processed) without being displayed on the display unit 40. The blood pressure information estimation device 100 plays back the received video image 500 of the hand and extracts the first pulse wave 601 from the temporal change in color of the video image of the first region 501 at the base of the hand. Furthermore, the same video image is played back again and the second pulse wave 602 is extracted from the temporal change in color of the video image of the second region 502 at the fingertips of the hand. The first region 501 and the second region 502 are separated by a predetermined distance d. The predetermined distance d can be the distance between the center of the first region 501 and the center of the second region 502. In the example shown in Figure 1, the first region 501 is defined as the area at the base of the hand, and the second region 502 is defined as the area near the first joint of the middle finger. However, the example is not limited to this, and the first region 501 and the second region 502 may be defined as a predetermined part of another finger and a part of the palm other than the area near the base of the hand, or they may be defined as two regions in a part of the body other than the hand.
[0025] In the RGB color signals of the moving images in the first region 501 and the second region 502, the intensity of the green (G) signal changes in accordance with the pulse wave. This is based on the fact that the intensity of the G signal changes in accordance with the amount of hemoglobin contained in the blood flowing through the arteries.
[0026] If the second region 502 is defined as the fingertip region and the first region 501 as the base of the hand region, then because the second region 502 is further from the heart than the first region 501, the second pulse wave 602 observed in the second region 502 propagates later than the first pulse wave 601 observed in the first region 501. Therefore, the pulse wave propagation velocity can be calculated from the time delay between the first pulse wave 601 and the second pulse wave 602, and a predetermined distance d between the first region 501 and the second region 502. There is a relationship between pulse wave propagation velocity and blood pressure, where an increase in systolic blood pressure increases the tension of the blood vessel wall, reducing the elasticity of the blood vessel and thus increasing the pulse wave propagation velocity. By utilizing this relationship, blood pressure information can be estimated based on the pulse wave propagation velocity.
[0027] In the example shown in Figure 1, the blood pressure information estimation device 100 according to the first embodiment receives video data of the subject's hand 301 from an external source. Therefore, the blood pressure information estimation device 100 itself does not need to capture video images of the subject 300. As a result, regardless of the subject's location, the blood pressure information estimation device 100 can estimate the subject's blood pressure information by receiving video data of a predetermined part of the subject 300 from the mobile terminal 200.
[0028] Alternatively, the mobile terminal 200 may transmit video data directly to the blood pressure information estimation device 100 without going through the communication network 1000. Furthermore, the subject 300 may pre-capture and save video images of their hand 301 using the mobile terminal 200, and then transmit the saved video data to the blood pressure information estimation device 100.
[0029] Figure 2 shows a block diagram representing the configuration of a blood pressure information estimation device 100 according to the first embodiment of this disclosure. The blood pressure information estimation device 100 includes a control unit 10, a communication unit 20, a storage unit 30, and a display unit 40, which are connected by an internal bus 60. A smartphone, a mobile terminal such as a tablet, or a notebook PC can be used as the blood pressure information estimation device 100.
[0030] The control unit 10 includes a video acquisition unit 1, a video conversion unit 2, a video playback unit 3, a filter control unit 4, a pulse wave extraction unit 5, a pulse wave propagation velocity calculation unit 6, and a blood pressure information estimation unit 7. Each element included in the control unit 10 is implemented as software (program) by a computer in the blood pressure information estimation device 100, which includes a CPU, ROM, and RAM.
[0031] The communication unit 20 is equipped with a transceiver module for communicating with external devices of the blood pressure information estimation device 100. The communication unit 20 receives dynamic image data of a predetermined part of the subject 300 from an external source.
[0032] The memory unit 30 is, for example, a semiconductor memory and includes a video memory unit 31, a frame rate memory unit 32, a first pulse wave peak time memory unit 33, and a second pulse wave peak time memory unit 34.
[0033] The display unit 40 is an example of an output unit. The display unit 40 is composed of a liquid crystal display device or the like and can display moving images of a predetermined part of the subject, detected pulse waves, estimated blood pressure information, etc. An audio output device may also be provided as an output unit to output the estimated blood pressure information etc. by voice.
[0034] The video acquisition unit 1 acquires a video image of a predetermined area at a first frame rate in order to detect the delay time of pulse waves in a first region 501 and a second region 502 separated by a predetermined distance d within a predetermined area of a living body. The first frame rate is a higher frame rate than the second frame rate used to play back the video. For example, if the second frame rate is 30 FPS (Frames Per Second), the first frame rate can be 240 FPS. However, the system is not limited to this example, and other frame rates may be used for the first and second frame rates. Furthermore, in the second embodiment described later, the video acquisition unit 1 can also acquire moving images captured at a second frame rate.
[0035] The positions of the first region 501 and the second region 502 may be stored in advance in the memory unit 30. For example, the memory unit 30 may store the coordinate positions of the lowest point on the wrist side of the palm and the highest point of the fingertips as the coordinate positions of the measurement frames in the first region 501 and the second region 502.
[0036] The video storage unit 31 stores video images of a predetermined part of a living organism captured at a first frame rate. For example, the first frame rate can be set to 240 FPS. This first frame rate value can be stored in the frame rate storage unit 32, and the video acquisition unit 1 can determine whether or not the video image acquired from the communication unit 20 was captured at the first frame rate by referring to the frame rate storage unit 32.
[0037] The video conversion unit 2 converts the acquired video into an uncompressed video. For example, the uncompressed format can be Base64. However, the unit is not limited to this example, and it may convert to other uncompressed video formats. If the acquired video is in MP4 format, playback may be interrupted due to the compression of the video data, potentially causing waveform distortion. Therefore, it is preferable to convert the video captured at the first frame rate to an uncompressed format before slow-motion playback.
[0038] The video playback unit 3 plays back the acquired video at a second frame rate lower than the first frame rate. "Playback" means acquiring each image as it changes in time at the second frame rate, and it is not necessary to display each acquired time-series image on the display unit. This is because the video acquired at the second frame rate is not intended for viewing, but rather for use by the pulse wave extraction unit 5 (described later) to extract pulse waves. The video playback unit 3 can play back the video using a circuit for reproducing operations that is typically found in mobile terminals. However, it is sufficient for the video playback unit 3 to process the video without displaying it on the display unit 40; it is not necessary to perform playback that displays the video on the display unit 40. However, the video playback unit 3 may display the video on the display unit 40 and process the video. In other words, in this specification, "playing back" a video includes both displaying the video on the display unit 40 and not displaying the video on the display unit 40.
[0039] Here, it is preferable to process the video data of the first region 501 and the second region 502 using a square wave correlation filter to remove noise from the video data of the first region 501 and the second region 502. Furthermore, it is preferable to switch the bandwidth of the square wave correlation filter according to the frame rate. Therefore, it is preferable that the blood pressure information estimation device 100 according to the first embodiment further has a filter control unit 4 that controls the bandwidth of the square wave correlation filter for extracting the first pulse wave and the second pulse wave from the acquired video data. Here, the square wave correlation filter is basically a type of digital filter. The waveform of a pulse wave is a waveform that transmits signals to both the positive and negative sides, and has a shape in which the center rises significantly to the positive side. The square wave correlation filter extracts the time series shape of such a pulse wave. For example, when playing back at 30 FPS, the pulse wave depends on the person and changes somewhat depending on the conditions. Therefore, to cover the temporal changes in the pulse wave, it is preferable to multiply and play back using about three different square wave correlations: one with 28 samples that captures the overall shape, another with 24 samples, and yet another with 20 samples. However, these sample counts are examples for matching the time width at 30 FPS; when playing back slowly at 240 FPS, the time density increases, so it is preferable to multiply each by 8 and switch the range accordingly. In other words, it is preferable to switch the time relationship of the square wave correlations used for noise reduction on the playback side at the same time as controlling the frame rate.
[0040] The pulse wave extraction unit 5 extracts the first pulse wave in the first region 501 and the second pulse wave in the second region 502 based on the regenerated video image. As shown in Figure 1, if the first region 501 is the region near the base of the hand and the second region 502 is the region near the fingertips, the second region 502 is located at a distance d further from the heart than the first region 501. Therefore, the second pulse wave is detected with a delay compared to the first pulse wave, corresponding to this distance d. Accordingly, the pulse wave velocity pwv can be calculated from this delay time Δt and distance d.
[0041] The pulse wave propagation velocity calculation unit 6 calculates the pulse wave propagation velocity pwv (=d / Δt) from the time difference Δt between the peaks of the first pulse wave and the second pulse wave, and a predetermined distance d.
[0042] The blood pressure information estimation unit 7 estimates blood pressure information based on the calculated pulse wave velocity. Here, it is preferable that the blood pressure information includes information on at least one of the following: blood pressure, vascular age, and degree of arteriosclerosis. The method for estimating blood pressure from pulse wave velocity will be described later. Since a faster pulse wave velocity is considered to indicate stiffer blood vessels, pulse wave velocity can be used as an indicator of the degree of arteriosclerosis. Furthermore, vascular age can be calculated by comparing the calculated pulse wave velocity with the average value of pulse wave velocity for healthy individuals of each age.
[0043] The display unit 40 outputs estimated blood pressure information. For example, the display unit 40 may display the estimated blood pressure value.
[0044] (Reasons for setting the first frame rate as the high frame rate) Here, we will explain why we use moving images captured at a first frame rate higher than the second frame rate to detect pulse waves. A larger temporal delay between the pulse wave detected in the first region 501 and the pulse wave detected in the second region 502 is advantageous for accurately measuring pulse wave propagation velocity. Conventional non-contact blood pressure detection devices calculated pulse wave propagation velocity from the delay time and distance between two points separated by approximately 1 m, such as the face and hand.
[0045] However, if we define the designated area as the hand, with the first region 501 as the base of the hand and the second region 502 as the fingertip region, the distance between the first region 501 and the second region 502 is approximately 15 cm, and the pulse wave delay time is approximately 2 msec. At 30 FPS, the playback frame rate typically used for MP4 video recording, the sampling time is 33 msec, and even with interpolation, the resolution is only about 10 msec. Therefore, it is difficult to accurately determine the delay time of the two pulse waves by processing video captured at 30 FPS.
[0046] Therefore, in the blood pressure information estimation device 100 according to this embodiment, in order to shorten the sampling time, the delay time of the pulse wave is accurately detected by capturing moving images at a first frame rate that is faster than the second frame rate. For example, by setting the first frame rate to 240 FPS, the sampling time can be set to 4.17 [msec], and by performing interpolation, a resolution of about 1 [msec] can be obtained, which corresponds to the delay time of the pulse wave of about 2 [msec].
[0047] (Reasons for setting the second frame rate lower than the first frame rate) When attempting to implement the blood pressure information estimation device according to this embodiment using a mobile terminal such as a smartphone, even if imaging can be performed at the first frame rate, there is a problem in that it is difficult to perform pulse wave analysis at two points simultaneously at the first frame rate due to the low image processing capabilities of the mobile terminal.
[0048] Therefore, instead of performing analysis in real time, it is preferable to use a first frame rate for capturing the video image and a second frame rate (for example, 30 FPS) lower than the first frame rate for playing back the pulse wave. To this end, in the blood pressure information estimation device 100 according to this embodiment, the video format is converted (footage conversion) from, for example, MP4 format to Base64 format and saved, and the video image captured at the first frame rate is played back in slow motion at the second frame rate.
[0049] It is said that 30 FPS is the optimal frame rate for video playback that does not cause discomfort due to the characteristics of human vision, and most mobile devices are equipped with the function to play videos at 30 FPS. If the second frame rate is set to 30 FPS, then by playing back video footage captured at the first frame rate of 240 FPS at 30 FPS, 1 / 8 slow motion playback can be achieved. By performing slow motion playback in this way, minute fluctuations in the pulse wave can be precisely detected, and the delay time of the pulse wave occurring between the first region 501 and the second region 502 can be accurately detected.
[0050] Here, we have provided an example where the second frame rate is 30 FPS and the first frame rate is 240 FPS, but we are not limited to this example. In other words, even with frame rates other than those exemplified here, the first and second frame rates may be set to other values as long as the delay time between the two pulse waves can be accurately detected.
[0051] Next, a blood pressure information estimation method according to an embodiment of this disclosure will be described. Figure 3 shows a flowchart illustrating the operation procedure of the blood pressure information estimation device 100 according to the first embodiment of this disclosure.
[0052] First, in step S101, the communication unit 20 receives video data from outside the blood pressure information estimation device 100, thereby inputting video images of a predetermined body part, namely the hand, from the outside. That is, the video acquisition unit 1 acquires video images of the predetermined body part at a first frame rate in order to detect the delay time of the pulse wave in a first region 501 and a second region 502 separated by a predetermined distance d within the predetermined body part. The video acquisition unit 1 acquires the input video image file and saves it to the video storage unit 31.
[0053] Here, we will explain using the example where the video acquisition unit 1 acquires video in MP4 format, but it may be in other formats. Furthermore, the video of a predetermined part of the subject may not only be captured by the camera built into the mobile terminal 200, but may also be a video file captured by another terminal. We will also explain using the example where the predetermined part is the hand.
[0054] Here, we assume that the previously captured video footage is captured at a first frame rate (e.g., 240 FPS) that is higher than the second frame rate (e.g., 30 FPS). However, this example is not limited to this, and video footage captured at other frame rates, such as 960 FPS, may also be used.
[0055] Next, in step S102, the acquired MP4 format video of a predetermined region is converted into an uncompressed video by the video conversion unit 2. The uncompressed format can be, for example, Base64 format video. However, the MP4 format video may also be converted into another uncompressed video format.
[0056] The reason for converting MP4 format video to an uncompressed format such as Base64 format is as follows: MP4 format is a format where video quality is reduced during playback depending on the terminal's operating environment. Therefore, in normal MP4 streaming video playback, the playback interval is not constant depending on the operating environment, and when the captured video is played back, there is a tendency for a time lag to occur in the pulse wave. In the blood pressure information estimation device 100 according to this embodiment, it is necessary to precisely detect the pulse wave from the video, but if the MP4 format video is played back as is, there is a risk that the time when the pulse wave peak appears cannot be accurately detected. Therefore, for example, the MP4 format video is converted and saved to Base64 format, which is an uncompressed video format using ASCII code data blocks, and played back at regular intervals. Base64 format converts video into ASCII code character information, enabling playback without time lag for each ASCII code block. By utilizing such accurate block playback intervals, the saved video is played back twice to detect the time difference in the pulse wave between two points. Even when recording video by switching frame rates in this way, the time the subject is held captive during the recording itself is only about 5 to 6 seconds, so the burden on the subject is considered to be small.
[0057] Next, in step S103, the video playback unit 3 plays back the video captured at the first frame rate in slow motion at the second frame rate. That is, the video playback unit 3 plays back the acquired video at a second frame rate lower than the first frame rate. Specifically, the video playback unit 3 plays back the video recorded at the first frame rate of 240 FPS at a second frame rate in slow-speed mode (for example, 30 FPS). Therefore, in this case, the playback speed becomes 30 / 240 = 1 / 8 times, and it is played back as ultra-slow motion video.
[0058] Next, in step S104, the pulse wave extraction unit 5 extracts the pulse wave waveform of the first region 501, which is the area at the base of the hand, from the slow-motion playback image at 1 / 8 speed. That is, the pulse wave extraction unit 5 extracts the first pulse wave in the first region 501 and the second pulse wave in the second region 502 based on the playback video. Here, the position coordinates of the first region 501, which is the area at the base of the hand, are stored in the storage unit 30 of the video image of the palm. In the video image of the first region 501, the brightness change of G (green component) among the RGB components is extracted as the pulse wave. The waveform of the extracted pulse wave is shown in Figure 4.
[0059] At this time, the waveform of the video converted from the first frame rate (240 FPS) to the second frame rate (30 FPS) contains eight times the amount of information as a normal 30 FPS video, and the detected waveform will have eight times the period. Therefore, the filter control unit 4 controls the bandwidth of the square wave correlation filter to correspond to a waveform that is eight times longer than normal. The pulse wave extraction unit 5 extracts pulse waves using the waveform from which noise has been removed by the square wave correlation filter. The bandwidth control of the square wave correlation filter will be described later.
[0060] Next, in step S105, the pulse wave extraction unit 5 detects the pulse wave peak time of the first region from the pulse wave of the first region 501 and stores it in the first pulse wave peak time storage unit 33. For example, as shown in the upper pulse wave graph of the first region in Figure 4, if three peaks appear in the detected waveform, the time at which each peak appeared is stored in the first pulse wave peak time storage unit 33. 11 , t 12 , t 13 This is stored in the first pulse wave peak time memory unit 33.
[0061] Next, in step S106, the video playback unit 3 resets the video to position 0 [sec] and repeats playback from the beginning. During repeat playback, the pulse wave extraction unit 5 switches the measurement frame to the second region 502. At this point, the position coordinates of the fingertip region, which is the second region 502 of the video image of the palm, are stored in the storage unit 30.
[0062] Next, in step S107, the pulse wave extraction unit 5 detects the pulse wave peak time of the second region from the pulse wave of the second region 502 and stores it in the second pulse wave peak time storage unit 34. For example, as shown in the graph of the pulse wave in the lower second region of FIG. 4, when three peaks appear in the detected waveform, the times when each peak appears are t 21 、t 22 、t 23 and are stored in the second pulse wave peak time storage unit 34 as such.
[0063] Next, in step S108, the pulse wave propagation speed calculation unit 6 calculates the pulse wave propagation speed from the average value of the pulse wave peak time differences between the first region and the second region and the average value of the length of the palm. That is, the pulse wave propagation speed calculation unit 6 calculates the pulse wave propagation speed from the time difference between the peaks of the first pulse wave and the second pulse wave and a predetermined distance d. Specifically, the pulse wave propagation speed calculation unit 6 reads the peak times t 11 、t 12 、t 13 of the pulse wave in the first region 501 from the first pulse wave peak time storage unit 33, and reads the peak times t 21 、t 22 、t 23 of the pulse wave in the second region 502 from the second pulse wave peak time storage unit 34, and calculates the time differences Δt1 = t 11 -t 21 、Δt2 = t 12 -t 22 、Δt3 = t 13 -t 23 of these respective pulse waves, and calculates the average value (Δt) of these three values. However, when calculating the average value of the time differences of a plurality of pulse waves, it is not limited to calculating the average value of three values, and the average value of two or four or more values may be calculated.
[0064] Furthermore, the predetermined distance d between the first region 501 and the second region 502 can be set to an average value of 15 [cm]. However, this is not the only example; as will be described later, the predetermined distance d may be adjusted based on the average palm length of men and women according to the subject's gender, or based on the actual measured length of the subject's palm. The pulse wave velocity pwv can be calculated by dividing the predetermined distance d by the average value Δt of the calculated pulse wave peak time difference (d / Δt).
[0065] Next, in step S109, the blood pressure information estimation unit 7 estimates the systolic blood pressure value from the correlation between pulse wave velocity and the systolic blood pressure value. That is, the blood pressure information estimation unit 7 estimates blood pressure information based on the calculated pulse wave velocity. It is preferable that the memory unit 30 stores information regarding the statistical correlation between pulse wave velocity and the systolic blood pressure value. Alternatively, the minimum blood pressure (diastolic blood pressure) may be calculated from the pulse wave velocity. Figure 5 shows the relationship between pulse wave velocity and blood pressure ("Development and Application of Blood Pressure Measurement Methods for Blood Pressure Biofeedback", Biofeedback Research, 1982, Vol. 9, pp. 28-31). The graph shown in Figure 5 can be broadly divided into three regions.
[0066] The first region is the region where the pulse wave velocity (pwv) is less than 5.0 [m / s]. In this region, blood pressure P0 can be calculated using the following formula (1). P0 = (50.0 × pwv) - 150 (1)
[0067] The second region is the area where the pulse wave velocity (pwv) is 5.0 to 17.0 [m / s]. In this region, blood pressure P1 can be calculated using the following formula (2). P1 = 9.4 × pwv (2)
[0068] The third region is the region where the pulse wave velocity (pwv) is 17.0 [m / s] or higher. In this region, blood pressure P2 can be calculated using the following formula (3). P2 = (17.5 × pwv) - 150 (3)
[0069] Here, we will explain why we use slow-motion playback of high-speed video footage for pulse wave propagation speeds over short distances, such as the palm of the hand. In the case of normal shooting at a frame rate of about 30 FPS, the sampling time is about 33 [msec], and when interpolated, it becomes about 10 [msec].
[0070] On the other hand, from the graph of the relationship between pulse wave velocity and blood pressure shown in Figure 5, the pulse wave velocity corresponding to the normal blood pressure measurement range of 90-180 [mmHg] is approximately 6-24 [m / s]. In this case, assuming the average length of the palm is 15 [cm], the discriminable range of the peak time difference of the pulse wave is 6.25-25 [msec]. In this case, if the target resolution is 5 [mmHg] for blood pressure, it is required to detect a time difference of approximately 2 [msec] in the pulse wave.
[0071] However, even with interpolation, a resolution of only about 10 msec can be obtained at the normal frame rate of 30 FPS. Therefore, by using the first frame rate shooting function that is pre-installed on smartphones, high-speed shooting at 240 FPS is possible on iPhones (registered trademark), resulting in a sampling time of 4.17 msec, and with interpolation, a resolution of about 1 msec can be obtained.
[0072] Similarly, with Android® mid-range or higher devices, high-speed shooting at 960 FPS is possible, resulting in a sampling time of 1.04 msec, and with interpolation, a resolution of approximately 0.3 msec can be obtained.
[0073] Furthermore, certain models are capable of high-speed shooting at 7680 FPS, achieving a sampling time of 0.13 msec, and with interpolation, a resolution of approximately 0.04 msec can be obtained.
[0074] Thus, with video images captured at a frame rate of around 30 FPS, it is difficult to accurately detect the delay time of pulse waves in two regions separated by about 15 cm, such as the palm of a hand. However, even for small areas like the palm of a hand, by using video images captured at the first frame rate, it is possible to accurately detect the delay time of pulse waves in two regions and accurately calculate the pulse wave propagation velocity.
[0075] Next, we will explain the switching of the square wave correlation filter between the first frame rate, 240 FPS, and the second frame rate, 30 FPS. A typical pulse rate is 50-100 beats / minute, and to measure this, it is preferable to superimpose square wave correlation filters with window widths of 20, 24, and 28 samples to create a bandpass type for the 30 FPS video. When the signal passes through the square wave correlation filter, the DC component is cut off because it is a bandpass filter, and the pulse wave fluctuates positively and negatively with 0 as the reference.
[0076] In contrast, when playing back video footage captured at the first frame rate of 240 FPS at the second frame rate of 30 FPS, the period is expanded eightfold, so it is preferable to switch each window width of the square wave correlation filter to eight times that amount. Specifically, the above window widths (20, 24, 28) for 30 FPS are switched to (160, 192, 224) respectively. By switching the window width in accordance with the change in frame rate in this way, an appropriate square wave correlation filter can be used for the slow-played video footage, and noise can be effectively removed.
[0077] As described above, the blood pressure information estimation device according to the first embodiment can easily estimate blood pressure information using a video image of a predetermined part of a subject input from an external source.
[0078] [Second Embodiment] In the blood pressure information estimation device 100 according to the first embodiment described above, an example was given in which blood pressure information is estimated using a video file input from an external source. However, the blood pressure information estimation device according to the second embodiment differs in that the blood pressure information estimation device itself captures video images of a predetermined part of the subject.
[0079] Figure 6 shows a diagram illustrating the overview of measurement using the blood pressure information estimation device according to the second embodiment of this disclosure. The blood pressure information estimation device 102 according to the second embodiment captures a video image of the hand 301 of the subject 300 and estimates blood pressure information using the captured video image 500.
[0080] Figure 7 shows a block diagram illustrating the configuration of a blood pressure information estimation device 102 according to the second embodiment of this disclosure. The blood pressure information estimation device 102 according to the second embodiment includes, in addition to the blood pressure information estimation device 100 according to the first embodiment shown in Figure 2, an image recognition unit 8, an image determination unit 9, a shooting speed control unit 11, a trained model storage unit 35, and a camera 50.
[0081] Figure 8 shows a flowchart illustrating the operation procedure of the blood pressure information estimation device 102 according to the second embodiment of this disclosure. First, in step S201, the application program (app) is launched, and the camera 50 captures a video image of the subject's hand 301 at a second frame rate. The camera 50 first captures the image at a second frame rate (30 FPS) in order to recognize the hand image. Here, it is preferable to record the video image at a slow second frame rate when performing hand image recognition. This is because, although it is preferable to perform hand image recognition in real time, if recording is performed at a high first frame rate, it is difficult to process in real time on the mobile device, the blood pressure information estimation device 102, considering the CPU processing power.
[0082] Next, in step S202, the image of the hand is recognized. When the hand 301 is held in front of the camera 50, the image recognition unit 8 uses the image recognition trained model stored in the trained model storage unit 35 to recognize the image of the hand, which is a predetermined part, from the acquired video. That is, in this embodiment, the image recognition unit 8 recognizes the image of a predetermined part from the video acquired at the second frame rate by the video acquisition unit, before the video acquisition unit 1 acquires the video captured at the first frame rate (step S208), as will be described later.
[0083] Next, in step S203, it is determined whether or not the recognition of the hand 301 is complete. If the recognition of the hand is not complete, the system returns to step S202 and performs the recognition of the hand image again. If the recognition of the hand 301 is complete, the measurement of the pulse wave is started.
[0084] Next, in step S204, based on the hand recognition result, the coordinate positions in the image of the feature points of the first region 501, which is the lowest part of the palm on the wrist side, and the second region 502, which is the highest part of the fingertips, are stored in the storage unit 30 as measurement frames.
[0085] Next, in step S205, the image determination unit 9 determines whether the image of the hand was properly captured. That is, before the video acquisition unit 1 acquires the video image captured at the first frame rate, the image determination unit 9 determines whether a pulse wave can be detected from the video image acquired by the video acquisition unit 1 at the second frame rate. The determination items include, for example, the presence or absence of specular reflection and the presence or absence of a pulse wave. Specifically, skin images within the two measurement frames, the first region 501 and the second region 502, are acquired as is at the second frame rate (30 FPS), and the image determination unit 9 briefly checks for the presence or absence of specular reflection and the presence or absence of a pulse wave.
[0086] Here, if the sum of RGB values within the measurement frame exceeds a predetermined value, it is determined that specular reflection is occurring. When specular reflection occurs, it is difficult to detect changes in the color of the skin surface, so the result is set to "NG".
[0087] Furthermore, regarding the presence or absence of a pulse wave, the system determines whether or not a periodic change in the brightness of the green component (G) of the RGB color signal can be detected. If it cannot be detected, the system determines that a pulse wave cannot be detected and sets the result to "NG".
[0088] Here, since the presence or absence of a pulse wave is determined using video images captured at 30 FPS, the filter control unit 4 controls the number of band stages of the square wave correlation filter to match the sampling time of 33.3 [msec] and extracts the pulse wave.
[0089] In step S206, the system determines whether the result is "NG" or "OK". If it is "NG", in step S207, it prompts the user to take the picture properly by making an announcement (warning) such as, "Please change the position relative to the lighting and take the picture."
[0090] If the image determination unit 9 determines that it is OK, in step S208, the shooting speed control unit 11 reads the first frame rate setting value stored in the frame rate storage unit 32 and performs the capture of a moving image of the palm at the first frame rate of 240 FPS.
[0091] Next, in step S209, a video of the palm for a predetermined period of time is saved. Specifically, the camera 50 is used to capture a video in MP4 format at the first frame rate for about 5 seconds, and the video is recorded in the video storage unit 31.
[0092] Using the moving image of the palm captured as described above, the pulse wave propagation velocity is calculated, and blood pressure information is estimated from the calculated pulse wave propagation velocity. The procedure for estimating blood pressure information using the moving image captured by camera 50 is the same as that of the blood pressure information estimation device 100 according to the first embodiment, so a detailed explanation is omitted.
[0093] [Third Embodiment] In the blood pressure information estimation device according to the above embodiment, an example was shown in which the predetermined distance d, which is the distance between the center of the first region 501 and the center of the second region 502, is set to an average value for Japanese men (for example, 15 cm). However, the predetermined distance d changes depending on the height of the subject. Therefore, if the hand is used as the site for capturing a dynamic image of the pulse wave to measure blood pressure, in order to improve the absolute accuracy of the pulse wave propagation velocity pwv in the hand, it is necessary to accurately determine the predetermined distance which is the propagation path of the pulse wave, along with the accurate propagation time t.
[0094] While it is difficult to easily measure hand size using a mobile device, subjects usually know their height, and a predetermined distance can be estimated from their height. For example, the correlation between height and hand length is known from statistical data from AIST, etc. (e.g., Makiko Kawachi, 2012: AIST Japanese Hand Dimensions Data. https: / / www.airc.aist.go.jp / dhrt / hand / index.html). Here, "hand length" refers to the straight-line distance from the wrist crease to the tip of the middle finger with the hand (fingers and palm) extended.
[0095] Therefore, the blood pressure information estimation device according to the third embodiment is characterized by inputting the subject's height and gender, estimating the arm length from the height, and calculating blood pressure using a predetermined distance estimated from the arm length.
[0096] Figure 9 shows a block diagram illustrating the configuration of the blood pressure information estimation device 103 according to the third embodiment of this disclosure. The difference between the blood pressure information estimation device 103 according to the third embodiment and the blood pressure information estimation device 102 according to the second embodiment is that it further includes an input unit 70 for inputting information on the gender and height of the subject whose blood pressure is to be measured, and a predetermined distance calculation unit 12 for calculating a predetermined distance based on the arm length estimated from the gender information and height. The other configurations of the blood pressure information estimation device 103 according to the third embodiment are the same as those of the blood pressure information estimation device 102 according to the second embodiment, so a detailed explanation is omitted.
[0097] The input unit 70 may include icons for inputting gender (male or female) and for inputting height, as displayed on the display unit 40. As a method of inputting gender, for example, the user may directly input the words "male" or "female" into the icons displayed on the display unit 40. Alternatively, the user may select either male or female by touching the icon for selecting male or female displayed on the display unit 40.
[0098] Furthermore, as a method for inputting height values, for example, the height value may be directly entered into the icon displayed on the display unit 40. Alternatively, the user may select a height value by touching the icon displayed on the display unit 40 to scroll through the available height values.
[0099] Alternatively, the input unit 70 may be configured to identify and input the subject's gender and height values through voice recognition.
[0100] Furthermore, when using externally inputted video footage of the subject's hands, the video file name may include information to identify the subject's gender and their height. For example, if the subject is male, the letter "M" indicating maleity may be included in the video file name to identify the subject as male. Similarly, if the subject's height is 170cm, the number "170" may be included in the video file name. However, the characters to be included in the video file are not limited to these examples; other characters may be included in the file name to input information regarding gender and height.
[0101] The input unit 70 must input the subject's gender and height before the blood pressure information estimation unit 7 calculates the subject's blood pressure. For example, the input unit 70 may be configured to input the subject's gender and height before capturing a video image of the subject's hand. Therefore, for example, before the camera 50 captures a video image of the subject's hand, the display unit 40 may display a prompt to input the subject's gender and height into the input unit 70, or voice guidance may be output to prompt the user to input the gender and height into the input unit 70.
[0102] (Relationship between height and arm length) Table 1 below shows an example of the relationship between height and arm length for Japanese men, and Table 2 shows an example of the relationship between height and arm length for Japanese women.
[0103] [Table 1]
[0104] [Table 2]
[0105] Figure 10A shows a graph illustrating the relationship between height and arm length for Japanese men. Figure 10B shows a graph illustrating the relationship between height and arm length for Japanese women. Figures 10A and 10B plot the data shown in Tables 1 and 2, respectively. As can be seen from Figures 10A and 10B, arm length has a linear relationship with height across the range from the minimum to the maximum height. For example, the straight line shown in Figure 10A can be approximated by the least squares method using the following equation (4), where y [mm] is arm length and x [mm] is height. y = 0.1217x - 26.068 (4)
[0106] Therefore, using equation (4), arm length can be calculated from any height value from the minimum to the maximum. However, the formula for calculating arm length from height is not limited to equation (4) above.
[0107] The above equation (4) can be stored in the memory unit 30. The predetermined distance estimation unit 12 can obtain the height value from the input unit 70 and use the equation (4) read from the memory unit 30 to calculate the arm length.
[0108] However, the method by which the predetermined distance estimation unit 12 estimates arm length from height is not limited to using a mathematical formula like equation (4). For example, a database of height values and arm lengths may be stored in the storage unit 30, and the predetermined distance estimation unit 12 may select an arm length value corresponding to the height value obtained from the input unit 70.
[0109] (Calculation of pulse wave propagation path based on arm length) As described above, arm length can be determined from height. However, the predetermined distance used to calculate the pulse wave propagation velocity in this embodiment does not coincide with arm length. This is because a predetermined area is required to capture an image of the skin that changes according to blood flow, but the wrinkled areas are not flat, and the fingertips do not have enough area to detect pulse waves. Therefore, the positions of the first region 501 and the second region 502 are set to be inside the wrinkles and fingertips.
[0110] Figure 11 shows the relationship between the subject's hand length y and a predetermined distance d'. As shown in Figure 11, the center position of the first region 501 is set to a position shifted by a first correction value a from the wrinkle 500a toward the fingertips. The center position of the second region 502 is set to a position shifted by a second correction value b from the fingertips 500b toward the base of the hand. Therefore, the relationship between the predetermined distance d' and the hand length y is given by equation (5) below. d'=yab (5)
[0111] Using equations (4) and (5) above, a predetermined distance d' can be calculated from the height value entered into the input unit 70. As an example, let's consider the case where the subject is male and his height is 1700 [mm]. Also, for example, let's set the first correction value a to 20 [mm] and the second correction value b to 10 [mm]. Then, first, the arm length y can be calculated using equation (4) as follows. y=(0.1217×1700)-26.068=180.822[mm]
[0112] Furthermore, the predetermined distance d' can be calculated using equation (5) as follows. d' = 180.822 - 20 - 10 = 150.822 [mm]
[0113] As described above, the blood pressure information estimation device 103 according to the third embodiment can accurately calculate a predetermined distance based on the subject's gender and height, thereby enabling more accurate calculation of blood pressure information.
[0114] In the above explanation, the example of estimating human blood pressure information was used, but the blood pressure information estimation device according to this embodiment can be used not only for humans but also for other organisms that circulate blood from the heart throughout the body.
[0115] According to the blood pressure information estimation device of this embodiment, blood pressure can be easily checked for the purpose of screening for the health management of elderly people residing in facilities or heavy laborers working at construction sites, etc.
[0116] Conventional cuff-type blood pressure monitors require subjects to roll up their sleeves to wrap a cuff around their upper arm or wrist each time they take a measurement, which is a cumbersome process. However, with the blood pressure information estimation device according to this embodiment, it is only necessary to point a predetermined part of the body, such as the hand, towards a camera on a mobile device, and there is no need to wrap a cuff, so blood pressure can be measured easily and without contact.
[0117] Furthermore, while conventional blood pressure monitors required the subject to directly attach the device for measurement, the blood pressure information estimation device according to this embodiment allows the receiving end to measure blood pressure simply by transmitting a video file of the palm of the hand taken for 5 to 10 seconds. This makes it possible to check the health of subjects who do not have a blood pressure monitor, or subjects in disaster areas or remote locations.
[0118] Furthermore, brain training exercises can increase the risk of stroke in elderly individuals with hypertension due to increased blood pressure caused by tension. The blood pressure information estimation device according to this embodiment allows for easy blood pressure checks to be performed simultaneously with brain training, enabling appropriate brain training while managing blood pressure.
[0119] Furthermore, while the effect of improving the balance of the autonomic nervous system brought about by mindfulness can be easily seen in changes in blood pressure, measuring blood pressure with a cuff after practicing mindfulness may cause the improved state to revert to its original state. With the blood pressure information estimation device according to this embodiment, blood pressure can be measured without burdening the subject, so the effects of mindfulness can be correctly evaluated from the perspective of changes in blood pressure.
Claims
1. A video acquisition unit acquires moving images of a first region and a second region separated by a predetermined distance from a predetermined part of a living organism at a first frame rate, A storage unit that stores the moving image captured at the first frame rate, A pulse wave extraction unit plays back the video stored in the memory unit at a second frame rate lower than the first frame rate to extract a first pulse wave in the first region and a second pulse wave in the second region. A pulse wave propagation velocity calculation unit calculates the pulse wave propagation velocity based on the first pulse wave and the second pulse wave, A blood pressure information estimation unit that estimates blood pressure information based on the calculated pulse wave propagation velocity, An output unit that outputs the estimated blood pressure information, A blood pressure information estimation device characterized by having the following features.
2. The blood pressure information estimation device according to claim 1, further comprising a video conversion unit that converts a video image captured at the first frame rate into an uncompressed video image converted to the second frame rate.
3. The blood pressure information estimation device according to claim 1 or 2, further comprising a filter control unit for controlling the bandwidth of a rectangular wave correlation filter for extracting the first pulse wave and the second pulse wave from the acquired video image.
4. The blood pressure information estimation device according to any one of claims 1 to 3, further comprising an image recognition unit that recognizes an image of a predetermined part from a moving image at a second frame rate before the video acquisition unit acquires the moving image captured at the first frame rate.
5. The blood pressure information estimation device according to claim 4, wherein the image recognition unit recognizes an image of the predetermined region from the acquired video image using a pre-trained image recognition model.
6. The blood pressure information estimation device according to any one of claims 1 to 5, further comprising an image determination unit that determines whether or not a pulse wave can be detected from a moving image at a second frame rate before the video acquisition unit acquires the moving image captured at the first frame rate.
7. The blood pressure information estimation device according to claim 6, wherein the image determination unit issues a warning if it cannot detect a pulse wave from the acquired video image.
8. The blood pressure information estimation device according to any one of claims 1 to 7, wherein the blood pressure information includes information on at least one of blood pressure, vascular age, and degree of arteriosclerosis.
9. The blood pressure information estimation device according to any one of claims 1 to 8, wherein the predetermined part is the hand.
10. An input section for inputting information about the gender and height of the subject whose blood pressure is to be measured, A predetermined distance calculation unit calculates the predetermined distance based on the information regarding gender and the hand length estimated from the height value, The blood pressure information estimation device according to claim 9, further comprising:
11. The video acquisition unit acquires moving images of a first region and a second region separated by a predetermined distance from a predetermined part of a living organism at a first frame rate and stores them in the storage unit. The pulse wave extraction unit plays back the moving image captured at the first frame rate at a second frame rate lower than the first frame rate, and extracts the first pulse wave in the first region and the second pulse wave in the second region. The pulse wave propagation velocity calculation unit calculates the pulse wave propagation velocity based on the first pulse wave and the second pulse wave. The blood pressure information estimation unit estimates blood pressure information based on the calculated pulse wave propagation velocity. The output unit outputs the estimated blood pressure information. A method for estimating blood pressure information, characterized by the following features.
12. On the computer, A moving image is acquired and saved at a first frame rate, capturing a first region and a second region separated by a predetermined distance from a predetermined part of a living organism. The video image captured at the first frame rate is played back at a second frame rate lower than the first frame rate to extract the first pulse wave in the first region and the second pulse wave in the second region. The pulse wave propagation velocity is calculated based on the first pulse wave and the second pulse wave. Based on the calculated pulse wave propagation velocity, blood pressure information is estimated. Output the estimated blood pressure information. A blood pressure information estimation program characterized by executing each step.