Measurement method, program, measurement device, imaging device, and system
The method uses frequency analysis and band-pass filtering to separate motion and biological components in video data, enabling accurate real-time measurement of biological information.
Patent Information
- Application Number
- PCT/JP2026/009130
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2025-05-13
- Filing Date
- 2026-03-10
- Publication Date
- 2026-11-19
AI Technical Summary
Existing methods struggle to accurately measure biological information, such as respiratory rate and pulse rate, from video data due to low frame rates and small subject movements, which blur the distinction between motion components and biological information components.
A method involving frequency analysis to identify peak frequencies, generation of a band-pass filter based on these frequencies, and application of the filter to time-series data to separate and measure biological information components while suppressing motion components.
Enables accurate real-time measurement of biological information by distinguishing between motion and biological components, improving measurement accuracy and reliability.
Smart Images

Figure JP2026009130_19112026_PF_FP_ABST
Abstract
Description
Measurement method, program, measurement apparatus, imaging apparatus, and system
[0001] The present disclosure relates to a method and apparatus for acquiring biological information of a subject.
[0002] Biological information such as respiratory rate and pulse rate can be acquired from video data obtained by imaging a subject. Patent Document 1 discloses a method for accurately measuring a pulse rate even when the frame rate of video data fluctuates. Further, Patent Document 2 discloses a method for accurately measuring a pulse rate even while a subject is exercising.
[0003] Japanese Patent Laid-Open No. 2016-77539 Japanese Patent Laid-Open No. 2021-83783
[0004] It is desired to acquire biological information more favorably from video data.
[0005] A measurement method as one aspect of the present disclosure is a method for acquiring biological information of a subject using video data obtained by imaging the subject, the method comprising: a step of performing frequency analysis on first time-series data generated from the video data; a step of acquiring two different peak frequencies that are included in a specific frequency band and each have a signal intensity reaching a peak value, based on a result of the frequency analysis; a generation step of generating a band-pass filter whose passband is set based on the two peak frequencies; and a step of applying the band-pass filter to the first time-series data to generate second time-series data including the biological information. Note that a program for executing processing according to the above method, and a measurement apparatus for performing the processing also constitute another aspect of the present disclosure.
[0006] According to the present disclosure, it is possible to provide a method and apparatus that can favorably acquire biological information from video data.
[0007] A block diagram showing the configuration of the measuring device in Example 1. A diagram showing frequency information obtained by frequency analysis of time-series data including biological information. A flowchart showing the processing procedure in Example 1. A time chart of the above processing. A flowchart showing the filter generation process in Example 1. A flowchart showing the filter generation process in Example 2. A flowchart showing the filter generation process in Example 3. A flowchart showing the filter generation process in Example 4. A flowchart showing the filter generation process in Example 5. A flowchart showing the filter generation process in Example 6.
[0008] The embodiments of this disclosure will be described below with reference to the drawings. While several features are described in the embodiments, not all of these features are necessarily required, and the features may be combined in any way.
[0009] By irradiating a part of a living organism with light and taking an image, biological information can be acquired (measured) by detecting the amount of reflected light from the organism in the image data. For example, hemoglobin in the blood absorbs green visible light with wavelengths of 495 to 570 nm very well. The G (green) signal obtained through a color filter in an imaging device contains many signals that transmit in the 495 to 570 nm wavelength band, improving the accuracy of respiratory rate and pulse rate measurements. Specifically, there are capillaries on the surface of the skin of a living organism, and as the blood flow in the capillaries changes due to the heartbeat, the amount of light absorbed by the blood also changes in accordance with the heartbeat. Therefore, the change in brightness obtained by reflection from the skin also changes with the heartbeat. Although the amount of change in brightness is small, if the average brightness value within the region of interest, including the skin area, is determined, the time-series brightness data contains signals that indicate biological information (hereinafter referred to as biological information signals). Therefore, by using time-series data of three signals—namely, the G signal, the R (red) signal, and the B (blue) signal—each representing a different amount of light absorption by hemoglobin in the blood, biological information signals can be detected.
[0010] Incidentally, when a subject moves, the time-series data will contain components of the subject's movement (noise component; hereinafter referred to as the motion component) and components of biological information signals (hereinafter referred to as the biological information component). The subject's movement here refers to, for example, the subject repeatedly moving their head during an examination in a hospital. In this case, it is necessary to distinguish between the motion component and the biological information component. The frequency information of the motion component and the biological information component can be confirmed by converting the time-series data into the frequency domain using frequency analysis. By performing frequency analysis, the frequency and signal intensity of the motion component and the biological information component can be determined.
[0011] However, a low frame rate for imaging (video data) reduces the frequency resolution when performing frequency analysis on time-series data. As a result, it becomes difficult to distinguish between motion components and biological information components, making it impossible to accurately measure biological information in real time.
[0012] The method disclosed in Patent Document 1, mentioned above, generates time-series data with a small motion component of the subject by applying multiple bandpass filters in parallel to the time-series data. However, in some cases, the time-series data passed through some bandpass filters may not contain the subject's biological information components, which may prevent accurate real-time measurement of biological information.
[0013] Furthermore, the method described in Patent Document 2 distinguishes between motion components and biological information components by identifying the motion component of the subject from the displacement of the movement of the subject's body surface. However, the subject's movements referred to in Patent Document 2 are large movements such as those performed when using a training machine, and it is difficult to distinguish small motion components of the subject, such as shaking the head, from biological information components.
[0014] Each embodiment provides a method and apparatus that can accurately measure biological information in real time, distinguishing even small movement components of the subject from biological information components.
[0015] First, the underlying technologies for each embodiment will be explained. Figures 2(a) to (c) show frequency information obtained by acquiring video data of a subject with a pulse rate of approximately 66 bpm and performing frequency analysis (Fourier transform) on the time-series data of the brightness of the region of interest. The vertical axis represents the intensity of the biosignal, and the horizontal axis represents frequency (Hz). The region of interest is the skin region within the facial region of the subject. The time-series data shows the average brightness value of the G signal over a 15-second period. The frequency at which the intensity of the biosignal, which represents the pulse component of the subject, peaks is approximately 1.1 Hz.
[0016] Figure 2(a) shows the frequency information when the subject is at rest, and Figure 2(b) shows the frequency information when the subject is moving. Figure 2(c) shows the frequency information when the subject's pulse rate fluctuates (hereinafter referred to as pulse rate fluctuation).
[0017] In Figure 2(a), a high signal intensity is observed only at a frequency of 1.1 Hz within the measurement frequency band of 0.5 to 2.0 Hz. The peak frequency, which is the frequency with the greatest signal intensity, is 1.1 Hz. In other words, the frequency information shown in Figure 2(a) contains only the frequency information of the subject's biological information components.
[0018] In Figure 2(b), the signal strength is high in the range of 0.7 to 1.2 Hz within the measurement frequency band, and this range includes frequency information of the subject's motion component and biological information component. The peak frequency with the highest signal strength is 0.8 Hz, and the peak frequency with the second highest signal strength is 1.1 Hz. Frequencies around 1.1 Hz indicate pulse rate fluctuations as a biological information component. In other words, even if a peak frequency is included in the measurement frequency band, it does not necessarily mean that it is a biological information component of the subject. Furthermore, it is difficult to distinguish between the subject's motion component and the biological information component. In addition, the subject's motion component is lower than 1.0 Hz.
[0019] In Figure 2(c), the signal strength is high in the range of 1.0 to 1.2 Hz within the measurement frequency band, with the peak frequency having the highest signal strength at 1.1 Hz and the second highest peak frequency at 1.2 Hz. Of these, the peak frequency of 1.1 Hz represents pulse rate variability, which is a biological information component.
[0020] When measuring biological information in real time using video data obtained through imaging, it may not be possible to acquire a sufficient amount of time-series data due to limitations in imaging time and the frame rate of the imaging device. When the amount of time-series data is small, it is difficult to distinguish between the motion component of the subject and the biological information component of the subject associated with pulse and respiration in frequency analysis of the time-series data. For this reason, it is desirable to generate a bandpass filter based on the results of frequency analysis in order to measure biological information in real time.
[0021] Figure 1 shows the configuration of a measurement system 1000, which includes a measurement device 10 as the measuring device of Example 1. The measurement system 1000 includes an imaging device 100 that images the skin surface of a subject and generates video data, and a measurement device 10 that continuously acquires information about reflected light from the video data over time to measure biological information. The measurement device 10 and the imaging device 100 are connected to each other so as to be able to communicate with each other by wire or wireless.
[0022] The imaging device 100 has an optical system and an image sensor, and can image a subject at a frame rate of 30 fps (frames per second). In other words, it can acquire video data as a video by continuously generating 30 images per second. Note that the frame rate may be other than 30 fps.
[0023] The image sensor includes pixels that are sensitive to one or more wavelengths. At least one of the pixels is sensitive to biological information components. The image sensor may also have pixels that are sensitive to two or more wavelengths, such as an image sensor with three RGB color filters. Photoelectric conversion elements such as CCD sensors and CMOS sensors are used as image sensors.
[0024] The measuring device 10 includes an analysis unit 11, an acquisition unit 12, a filter generation unit 13, a waveform generation unit 14, and a measurement unit 15. The measuring device 10 is composed of a personal computer or the like, and one or more processors execute a program acquired via a network or storage medium to measure biological information using video data acquired from the imaging device 100.
[0025] The measuring device 10 may be provided within the imaging device 100, or the imaging device 100 may be provided within the measuring device 10. Furthermore, multiple imaging devices 100 may be provided.
[0026] The analysis unit 11 performs frequency analysis on time-series data containing biological information. The time-series data containing biological information is obtained from the subject's body region (region of interest such as the face region) within the video data acquired by the imaging device 100.
[0027] First, the analysis unit 11 sets a region of interest. The region of interest can be set to any size and shape on the body region. Alternatively, the region of interest may be set to a skin region that includes facial features and satisfies skin conditions represented in a color space such as HSV. By setting such a region of interest, the accuracy of biological information measurement can be improved.
[0028] Next, the analysis unit 11 acquires luminance values from each frame image of the video data generated during imaging of the region of interest. The luminance value referred to here is the average luminance value of multiple G pixels (G signals) included in the frame image. By measuring biological information using the average luminance value, the influence of noise can be reduced, and the accuracy of biological information measurement can be improved.
[0029] Note that the average luminance value is not limited to the G signal; the average luminance values of the R signal or B signal may also be used. Alternatively, an average luminance value calculated using arithmetic operations from the luminance values of two or more color signals may also be used. In the following explanation, the average luminance value of the G signal will be denoted as Gave, the average luminance value of the R signal as Rave, and the average luminance value of the B signal as Bave. Furthermore, the average luminance value calculated from the luminance values of two color signals, for example, the G signal and the R signal, will be indicated as Gave / Rave. In addition, if the image sensor can generate video data that includes an IR (infrared) signal, the IR signal may be included in the two or more color signals mentioned above. Because IR light has a deep penetration depth into living tissue, information on blood flow deep within living tissue can be obtained by using an IR signal. Furthermore, by using an IR signal, biological information can be measured from video data generated by imaging in a dark-field environment invisible to the human eye.
[0030] When acquiring the average brightness value for each frame image, time-related information may also be acquired. This time-related information includes the time count from the start of imaging and the number of frames generated by the imaging. By acquiring time-related information, it is possible to detect the subject's biological information at specific points in time during imaging.
[0031] The analysis unit 11 then generates first time-series data that includes the average brightness for each frame image, that is, biological information. Frequency analysis is then performed on this first time-series data, and frequency information is generated as a result.
[0032] The acquisition unit 12 acquires frequency information from the analysis unit 11. The frequency information includes the first peak frequency with the greatest signal strength and its signal strength (peak value) within the measurement frequency band, and the second peak frequency with the second greatest signal strength and its signal strength (peak value). The measurement frequency band may be set to 0.5 to 2.0 Hz as described above, or it may be set to 0.6 to 2.0 Hz, given that the typical resting pulse rate of a subject is 40 to 120 bpm. Furthermore, the measurement frequency band may be set to any other frequency band. The analysis means is composed of the analysis unit 11 and the acquisition unit 12.
[0033] In this embodiment, the first and second peak frequencies are set as the frequency with the highest (first) signal strength and the frequency with the second highest signal strength, but other combinations of the first and second peak frequencies, such as the first and third or the second and third, may also be used.
[0034] The filter generation unit (filter generation means) 13 generates a bandpass filter with a passband set based on the frequency information acquired by the acquisition unit 12.
[0035] The waveform generation unit (data generation means) 14 generates and outputs second time-series data (waveform) containing biological information by applying the bandpass filter generated by the filter generation unit 13 to the first time-series data. By setting (changing) the passband width of the bandpass filter in the filter generation unit 13 based on frequency information, the waveform generation unit 14 can generate second time-series data (waveform) containing the biological information component of the subject while suppressing the motion component of the subject.
[0036] The measurement unit (information acquisition means) 15 measures (acquires) biological information from the second time-series data generated by the waveform generation unit 14. Examples of biological information include the subject's pulse rate, respiratory rate, and oxygen saturation. Specifically, it detects peak values (maximum and minimum values) included in the second time-series data, and calculates and outputs the pulse rate value as biological information based on the interval between frames containing these peak values and the frame rate of the imaging device 100.
[0037] For example, if the frame rate of the imaging device 100 is 30 fps, and the interval between the closest frames containing peak values (hereinafter referred to as adjacent peak frames) is 30 frames, the pulse rate will be 60 × 30 ÷ 30 = 60 beats / minute. In this case, the pulse rate may be calculated from the average peak interval obtained from the interval between adjacent peak frames, or from the number of peak values that appear within a predetermined time. The respiratory rate may also be calculated using a similar method.
[0038] Furthermore, the change in oxygen saturation can be calculated by sequentially calculating the amplitude change ratio in each time series interval of two different second time series data. The change in oxygen saturation can be obtained, for example, by sequentially calculating the amplitude change ratio in each time series interval of Rch and Gch.
[0039] The method for acquiring biometric information is not limited to the method described above. Furthermore, it may be possible to select which biometric information to acquire from the second time-series data.
[0040] The flowchart in Figure 3 shows the process (measurement method) that the measuring device 10 executes according to the program in this embodiment.
[0041] In step S11 (analysis step), the analysis unit 11 performs frequency analysis on first time-series data including biological information acquired from video data. The first time-series data is, for example, 15-second time-series data obtained by sampling 450 points of a signal from 30 fps video data. As the first time-series data, time-series data with another frame rate or another duration in seconds may be used.
[0042] Next, in step S12 (acquisition step), the acquisition unit 12 acquires frequency information from the result of the frequency analysis in step S11. The frequency information includes a first peak frequency having the largest peak value of signal strength in a measurement frequency band as a specific frequency band and the signal strength thereof, and a second peak frequency having the second largest peak value of signal strength and the signal strength thereof.
[0043] Next, in step S13 (generation step), the filter generation unit 13 generates a band-pass filter based on the frequency information acquired in step S12. Details of the generation of the band-pass filter will be described later.
[0044] Next, in step S14 (output step), the waveform generation unit 14 applies the band-pass filter generated in step S13 to the first time-series data to generate a waveform as second time-series data, and outputs the waveform to the measurement unit 15.
[0045] Next, in step S15 (measurement step), the measurement unit 15 calculates biological information using the waveform acquired from the waveform generation unit 14. Then, this process ends.
[0046] FIG. 4 is a time chart of the processing described with reference to FIG. 3. First, in period [i], frequency analysis is performed on the acquired first time-series data (step S11). Next, based on the result of the frequency analysis, a band-pass filter to be applied to the first time-series data in period [i] is generated (steps S12, S13). Then, biological information is measured using a waveform (second time-series data) obtained by applying the band-pass filter to the first time-series data in period [i] (steps S14, S15).
[0047] The start time of period [i+1] for performing the next frequency analysis is a predetermined time (1 second or another number of seconds) shorter than period [i] after the start time of period [i]. In period [i+1], frequency analysis is performed on the acquired first time-series data (step S11). Next, a band-pass filter to be applied to the first time-series data of period [i+1] is generated based on the result of the frequency analysis (steps S12, S13). Then, biological information is measured using a waveform (second time-series data) obtained by applying the band-pass filter to the first time-series data of period [i+1] (steps S14, S15).
[0048] Further, the start time of period [i+2] for performing the next frequency analysis is after the predetermined time that is shorter than period [i+1] from the start time of period [i+1]. In period [i+2], frequency analysis is performed on the acquired first time-series data (step S11). Next, a band-pass filter to be applied to the first time-series data of period [i+2] is generated based on the result of the frequency analysis (steps S12, S13). Then, biological information is measured using a waveform (second time-series data) obtained by applying the band-pass filter to the first time-series data of period [i+2] (steps S14, S15).
[0049] As described above, in the present embodiment, frequency analysis is performed on the first time-series data for each period shifted by the predetermined time, and a band-pass filter is generated based on the result of the frequency analysis for each period. Then, the first time-series data is filtered using the band-pass filter for each period. Thereby, biological information can be measured in real time.
[0050] The flowchart of FIG. 5 shows the filter generation processing executed by the filter generation unit 13 in step S13 of FIG. 3.
[0051] In step S101, the filter generation unit 13 determines whether or not the ratio between the first peak frequency f1 and the second peak frequency f2 included in the measurement frequency band satisfies a predetermined condition represented by the following formula (1).
[0052] f1 < 1.0 Hz 1.0 < f2 / f1 (1) Satisfying 1.0 < f2 / f1 increases the likelihood that the second peak frequency f2 is the frequency of the subject's biological information component. Satisfying f1 < 1.0 Hz decreases the likelihood that the first peak frequency f1 is the frequency of the subject's motion component. In other words, it increases the likelihood that the second peak frequency f2 is not the subject's biological information component but the frequency of a noise component.
[0053] Furthermore, if the condition 1.0 < f2 / f1 is not satisfied, the first peak frequency f1 will be greater than the second peak frequency f2. In other words, the first peak frequency f1 is more likely to be the frequency of the subject's biological information component, and the second peak frequency f2 is more likely to be the frequency of the noise component.
[0054] Furthermore, the following conditions in equation (1) are preferable.
[0055] f1 < 1.0 Hz 1.2 < f2 / f1 (1a) If the filter generation unit 13 determines that the conditions of equation (1) are satisfied, in step S102 it generates a bandpass filter whose passband is a single frequency band that includes the first peak frequency f1 and the second peak frequency f2. By generating a bandpass filter that includes the second peak frequency f2, which is highly likely to be the frequency of the subject's biological information components, in its passband, it is possible to generate a waveform (second time-series data) that includes the subject's biological information components.
[0056] On the other hand, the filter generation unit 13, which determines that the conditions of equation (1) are not satisfied, generates a bandpass filter in step S103 that has a passband that includes the first peak frequency f1 but does not include the second peak frequency f2. By generating a bandpass filter that does not include the second peak frequency f2, which is highly likely to be the frequency of the motion component of the subject, in the passband, it is possible to generate a waveform in which the motion component of the subject is suppressed.
[0057] According to the Example 1 described above, it is possible to generate a waveform that contains the subject's biological information components with a high probability while suppressing the motion component of the subject. Therefore, biological information can be measured accurately in real time.
[0058] Next, we will describe Example 2. The flowchart in Figure 6 shows the filter generation process performed by the filter generation unit 13 in step S13 of Figure 3 in this embodiment. The configuration of the measuring device in this embodiment is the same as in Example 1.
[0059] In step S201, the filter generation unit 13 determines whether the ratio of the first peak frequency f1 and the second peak frequency f2 included in the measurement frequency band satisfies the predetermined conditions shown in the following equation (2).
[0060] 0.6 < f2 / f1 < 1.4 (2) When the condition of equation (2) is satisfied, the first peak frequency f1 and the second peak frequency f2 are close to each other, so it can be determined that these first and second peak frequencies f1 and f2 are likely to be the frequencies of the subject's pulse rate fluctuations. In other words, it is highly likely that the first peak frequency f1 and the second peak frequency f2 are the frequencies of the subject's biological information components.
[0061] On the other hand, if the conditions of equation (2) are satisfied, the first peak frequency f1 and the second peak frequency f2 are far apart from each other, making it highly likely that the second peak frequency f2 is the frequency of the motion component of the subject.
[0062] Furthermore, it is more preferable to set the conditions in equation (2) as follows.
[0063] 0.9 < f2 / f1 < 1.1 (2a) If the filter generation unit 13 determines that the condition of equation (2) is satisfied, it performs the same processing as in step S102 of Example 1 in step S202.
[0064] On the other hand, the filter generation unit 13, which has determined that the conditions of equation (2) are not satisfied, determines in step S203 whether the ratio of the first peak frequency f1 to the second peak frequency f2 satisfies the predetermined conditions shown in the following equation (3).
[0065] f1 < 1.0 Hz 1.4 ≤ f2 / f1 (3) If 1.4 ≤ f2 / f1 is satisfied, it is highly likely that the second peak frequency f2 is the frequency of the subject's biological information component. If f1 < 1.0 Hz is satisfied, it is less likely that the first peak frequency f1 is the frequency of the subject's motion component.
[0066] Furthermore, if 1.4 ≤ f2 / f1 is not satisfied, the first peak frequency f1 will be greater than the second peak frequency f2. In other words, the first peak frequency f1 is more likely to be the frequency of the subject's biological information component, and the second peak frequency f2 is more likely to be the frequency of the subject's motion component.
[0067] Furthermore, the following conditions in equation (3) are preferable.
[0068] f1 < 0.9 Hz 1.4 ≤ f2 / f1 (3a) If the filter generation unit 13 determines that the conditions of equation (3) are satisfied, it performs the same processing as in step S202 in step S204.
[0069] On the other hand, the filter generation unit 13, which determines that the conditions of equation (3) are not satisfied, performs the same processing in step S205 as in step S103 of Example 1.
[0070] According to the Example 2 described above, it is possible to generate a waveform that includes the subject's biological information components, taking into account pulse rate fluctuations, while suppressing the motion component of the subject. Therefore, biological information can be measured accurately in real time.
[0071] Next, we will describe Example 3. The flowchart in Figure 7 shows the filter generation process performed by the filter generation unit 13 in step S13 of Figure 3 in this embodiment. The configuration of the measuring device in this embodiment is the same as in Example 1.
[0072] In step S301, the filter generation unit 13 determines whether the ratio of the first peak frequency f1 and the second peak frequency f2 included in the measurement frequency band satisfies equation (2) of Example 2. If the filter generation unit 13 determines that the conditions of equation (2) are satisfied, it performs the process in step S302; if it determines that the conditions are not satisfied, it performs the process in step S303.
[0073] In step S302, the filter generation unit 13 generates a bandpass filter whose passband is a single frequency band including the frequency of the first peak f1 and the second peak frequency f2. As described in Example 2, if the conditions of equation (2) are satisfied, it is highly likely that the first peak frequency f1 and the second peak frequency f2 are the frequencies of the biological information components of the subject.
[0074] In this case, the filter generation unit 13 sets the center frequency fc of the bandpass filter's passband to fc = (f1 + f2) / 2. Furthermore, the upper and lower limits of the bandpass filter's passband are set to a frequency of +0.1 Hz above the higher of the first peak frequency f1 and the second peak frequency f2, and to a frequency of -0.1 Hz above the lower of the two peak frequencies f2, respectively. Specifically, if f1 > f2, the upper limit of the passband is set to f1 + 0.1 Hz and the lower limit to f2 - 0.1 Hz. If f1 < f2, the upper limit of the passband is set to f2 + 0.1 Hz and the lower limit to f1 - 0.1 Hz. However, ±0.1 Hz is merely an example; other frequencies such as ±0.05 Hz or ±0.2 Hz may also be used.
[0075] In step S303, the filter generation unit 13 determines whether the ratio of the first peak frequency f1 to the second peak frequency f2 satisfies the condition of equation (3) in Example 2. If the filter generation unit 13 determines that the condition of equation (3) is satisfied, it performs the process in step S304; if it determines that the condition is not satisfied, it performs the process in step S305.
[0076] In step S304, the filter generation unit 13 generates a bandpass filter whose passband is a single frequency band including the frequency of the first peak f1 and the second peak frequency f2. As explained in Example 2, if the conditions of equation (3) are satisfied, there is a high probability that the second peak frequency f2 is the frequency of the biological information component of the subject.
[0077] In this case, the filter generation unit 13 sets the center frequency fc of the bandpass filter's passband to satisfy the condition f1 < fc < (f1 + f2) / 2. This is because, when detecting peak values (maximum or minimum values) from the generated waveform, if the waveform contains high-frequency components, it may incorrectly detect peak values. Therefore, it is preferable to set the center frequency to a frequency lower than (f1 + f2) / 2.
[0078] In step S305, the filter generation unit 13 performs the same processing as in step S103 of Example 1.
[0079] According to the embodiment 3 described above, by appropriately setting the center frequency of the bandpass filter's passband, it is possible to generate a waveform that includes the subject's biological information components, taking into account pulse rate fluctuations, while suppressing the subject's motion components. Therefore, biological information can be measured accurately in real time.
[0080] Next, we will describe Example 4. The flowchart in Figure 8 shows the filter generation process performed by the filter generation unit 13 in step S13 of Figure 3 in this embodiment. The configuration of the measuring device in this embodiment is the same as in Example 1.
[0081] In step S401, the filter generation unit 13 determines whether the ratio of the first peak frequency f1 and the second peak frequency f2 included in the measurement frequency band satisfies equation (2) of Example 2. If the filter generation unit 13 determines that the conditions of equation (2) are satisfied, it performs the process in step S402; if it determines that the conditions are not satisfied, it performs the process in step S403.
[0082] In step S402, the filter generation unit 13 performs the same processing as in step S302 of Example 3.
[0083] In step S403, the filter generation unit 13 determines whether the ratio of the first peak frequency f1 to the second peak frequency f2 satisfies the condition of equation (3) in Example 2. If the filter generation unit 13 determines that the condition of equation (3) is satisfied, it performs the process in step S405; if it determines that the condition is not satisfied, it performs the process in step S404.
[0084] In step S404, the filter generation unit 13 performs the same processing as in step S103 of Example 1.
[0085] In step S405, the filter generation unit 13 determines whether the signal strength V1 at the first peak frequency f1 and the signal strength V2 at the second peak frequency f2 satisfy the predetermined conditions shown in the following equation (4).
[0086] 0.7 < V2 / V1 (4) If the condition in equation (4) is satisfied, it is highly likely that the second peak frequency f2 is the frequency of the subject's biological information component. If the condition in equation (4) is not satisfied, the signal intensity V1 will be significantly higher than the signal intensity V2. In other words, it is highly likely that the first peak frequency f1 is the frequency of the subject's biological information component.
[0087] Furthermore, the following conditions in equation (4) are preferable.
[0088] 0.8 < V2 / V1 (4a) If the filter generation unit 13 determines that equation (4) is satisfied, it performs the same processing as in step S304 of Example 3 in step S406. If the filter generation unit 13 determines that equation (4) is not satisfied, it performs the same processing as in step S103 of Example 1 in step S407.
[0089] According to the embodiment 4 described above, it is possible to determine with greater accuracy whether the second peak frequency is likely to be the frequency of the subject's motion component, and to generate a waveform that includes the subject's biological information component while considering pulse rate fluctuations, while suppressing the subject's motion component. Therefore, biological information can be measured accurately in real time.
[0090] Next, we will describe Example 5. The flowchart in Figure 9 shows the filter generation process performed by the filter generation unit 13 in step S13 of Figure 3 in this example. The configuration of the measuring device in this example is the same as in Example 1.
[0091] In step S501, the filter generation unit 13 determines whether the difference between the first peak frequency f1 and the second peak frequency f2 included in the measurement frequency band satisfies the predetermined conditions shown in the following equation (5). In equation (5), f3 is a reference frequency corresponding to the reference pulse rate as reference information among the biological information. The reference pulse rate may be, for example, the pulse rate obtained by an actual measurement of the subject immediately before the filter generation process, the pulse rate obtained by an actual measurement of the subject in a resting state in the past, or a general resting pulse rate based on the subject's age and sex.
[0092] f1 < 1.0 Hz 0 < (f2 - f1) / f3 (5) If 0 < (f2 - f1) / f3 is satisfied, the second peak frequency f2 is likely to be the frequency of the subject's biological information component. If f1 < 1.0 Hz is not satisfied, the first peak frequency f1 is unlikely to be the frequency of the subject's motion component. In other words, the second peak frequency f2 is unlikely to be the frequency of the subject's biological information component.
[0093] Furthermore, if the condition 0 < (f2 - f1) / f3 is not satisfied, the first peak frequency f1 becomes larger than the second peak frequency f2. In other words, the first peak frequency f1 is more likely to be the frequency of the subject's biological information component, and the second peak frequency f2 is more likely to be the frequency of the subject's motion component.
[0094] Furthermore, it is more preferable to set the conditions in equation (5) as follows.
[0095] f1 < 1.0 Hz 0.2 < (f2 - f1) / f3 (5a) If the filter generation unit 13 determines that the conditions of equation (5) are satisfied, it performs the same processing as in step S102 of Example 1 in step S502. If the filter generation unit 13 determines that the conditions of equation (5) are not satisfied, it performs the same processing as in step S103 of Example 1 in step S503.
[0096] According to the Example 5 described above, by utilizing the frequency of the reference pulse rate, it is possible to generate a waveform that includes the subject's biological information components while suppressing the motion component of the subject. Therefore, biological information can be measured accurately in real time.
[0097] Next, we will describe Example 6. The flowchart in Figure 10 shows the filter generation process performed by the filter generation unit 13 in step S13 of Figure 3 in this embodiment. The configuration of the measuring device in this embodiment is the same as in Example 1.
[0098] In step S601, the filter generation unit 13 determines whether the difference between the frequency f1 of the first peak and the second peak frequency f2 included in the measurement frequency band satisfies the predetermined conditions shown in the following equation (6). f3 is the reference frequency described in Example 5.
[0099] -0.2 < (f2 - f1) / f3 < 0.2 (6) If the condition of equation (6) is satisfied, the first peak frequency f1 and the second peak frequency f2 are close to each other, and it can be determined that there is a high probability that these first and second peak frequencies f1 and f2 are the frequencies of the subject's pulse rate fluctuations. In other words, there is a high probability that the first peak frequency f1 and the second peak frequency f2 are the frequencies of the subject's biological information components. If the condition of equation (6) is not satisfied, the first peak frequency f1 and the second peak frequency f2 are far apart from each other, and there is a high probability that the second peak frequency f2 is the frequency of the subject's motion component.
[0100] Furthermore, the following conditions in equation (6) are preferable.
[0101] -0.1 < (f2 - f1) / f3 < 0.1 (6a) If the filter generation unit 13 determines that the conditions of equation (6) are satisfied, in step S602 it performs the same processing as in step S302 of Example 3. If the filter generation unit 13 determines that the conditions of equation (6) are not satisfied, in step S603 it determines whether the difference between the first peak frequency f1 and the second peak frequency f2 satisfies the predetermined conditions shown in the following equation (7).
[0102] f1 < 1.0 Hz 0.2 ≤ (f2 - f1) / f3 (7) If 0.2 ≤ (f2 - f1) / f3 is satisfied, the second peak frequency f2 is likely to be the frequency of the subject's biological information component. If f1 < 1.0 Hz is not satisfied, the first peak frequency f1 is unlikely to be the frequency of the subject's motion component.
[0103] If the condition 0.2 ≤ (f2 - f1) / f3 is not satisfied, the first peak frequency f1 will be greater than the second peak frequency f2. In other words, the first peak frequency f1 is more likely to be the frequency of the subject's biological information component, and the second peak frequency f2 is more likely to be the frequency of the subject's motion component.
[0104] Furthermore, the following conditions in equation (7) are preferable.
[0105] f1 < 0.8 Hz 0.1 ≤ (f2 - f1) / f3 (7a) If the filter generation unit 13 determines that the conditions of equation (7) are satisfied, in step S604 it performs the same processing as in step S304 of Example 3. If the filter generation unit 13 determines that the conditions of equation (7) are not satisfied, in step S605 it performs the same processing as in step S103 of Example 1.
[0106] According to the embodiment 6 described above, it is possible to generate a waveform that includes the subject's biological information components, taking into account pulse rate fluctuations, while suppressing the motion component of the subject. Therefore, biological information can be measured accurately in real time.
[0107] (Other Embodiments) The present disclosure can also be realized by supplying a program that implements one or more of the functions of the embodiments described above to a system or device via a network or storage medium, and by having one or more processors in the computer of that system or device read and execute the program. It can also be realized by a circuit (e.g., ASIC) that implements one or more functions.
[0108] The embodiments described above are merely representative examples, and various modifications and changes can be made to each embodiment when implementing this disclosure.
Claims
1. A measurement method for acquiring biological information of a subject using video data obtained by imaging the subject, comprising: a step of performing frequency analysis on first time-series data generated from the video data; a step of acquiring two different peak frequencies that are included in a specific frequency band and whose signal intensity is at its peak value, based on the results of the frequency analysis; a generation step of generating a bandpass filter with a passband set based on the two peak frequencies; and a step of generating second time-series data containing the biological information by applying the bandpass filter to the first time-series data.
2. The measurement method according to claim 1, characterized in that the two peak frequencies are the frequency with the largest peak value and the second largest peak value in the specific frequency band.
3. The measurement method according to claim 1 or 2, characterized in that, in the generation step, one frequency band including the two peak frequencies is set as the passband.
4. The measurement method according to claim 3, characterized in that, in the generation step, when the ratio or difference of the two peak frequencies satisfies a predetermined condition, one frequency band including the two peak frequencies is set as the passband.
5. The measurement method according to claim 4, characterized in that, in the generation step, when the ratio or difference does not satisfy the predetermined conditions, one frequency band is set as the passband, which includes the peak frequency with the larger peak value among the two peak frequencies and does not include the peak frequency with the smaller peak value.
6. The measurement method according to claim 4, characterized in that, when the peak value of the two peak frequencies is denoted as f1 and the peak value of the other as f2, the predetermined conditions are f1 < 1.0 Hz and 1.0 < f2 / f1.
7. The measurement method according to claim 4, characterized in that when the peak with the larger peak value of the two peak frequencies is denoted as f1, the peak with the smaller peak value as f2, and the frequency corresponding to the reference information among the biological information as f3, the predetermined conditions are f1 < 1.0 Hz and 0 < (f2 - f1) / f3.
8. The measurement method according to claim 4, characterized in that when the peak with the larger peak value of the two peak frequencies is denoted as f1 and the peak with the smaller peak value as f2, the predetermined condition is 0.6 < f2 / f1 < 1.
4.
9. The measurement method according to claim 4, characterized in that, when the peak frequency with the larger peak value among the two peak frequencies is denoted as f1, the peak frequency with the smaller peak value as f2, and the frequency corresponding to the reference information among the biological information as f3, the predetermined condition is -0.2 < (f2 - f1) / f3 < 0.
2.
10. The measurement method according to 8 or 9, characterized in that when the center frequency of the passband is fc, the condition fc = (f1 + f2) / 2 is satisfied.
11. The measurement method according to claim 4, characterized in that when the peak with the larger peak value of the two peak frequencies is denoted as f1 and the peak with the smaller peak value as f2, the predetermined conditions are f1 < 1.0 Hz and 1.4 ≤ f2 / f1.
12. The measurement method according to claim 4, characterized in that, when the peak value of the two peak frequencies with the larger peak value is V1 and the peak value of the peak frequency with the smaller peak value is V2, the predetermined condition is 0.7 < V2 / V1.
13. The measurement method according to claim 4, characterized in that when the peak with the larger peak value of the two peak frequencies is denoted as f1, the peak with the smaller peak value as f2, and the frequency corresponding to the reference information among the biological information as f3, the predetermined conditions are f1 < 1.0 Hz and 0.2 ≤ (f2 - f1) / f3.
14. The measurement method according to any one of claims 6, 7, and 11 to 13, characterized in that when the center frequency of the passband is fc, the condition f1 < fc < (f1 + f2) / 2 is satisfied.
15. The measurement method according to any one of claims 1 to 14, further comprising the step of obtaining the biological information from the second time-series data.
16. The measurement method according to any one of claims 1 to 15, characterized in that the biological information includes at least one of pulse rate, respiratory rate, and oxygen saturation.
17. A program characterized by causing a computer to perform a process according to the measurement method described in claims 1 to 16.
18. A measuring device for acquiring biological information of a subject using video data obtained by imaging the subject, comprising: an analysis means for performing frequency analysis on first time-series data generated from the video data and acquiring two distinct peak frequencies that are included in a specific frequency band and whose signal intensity is at its peak value, based on the results of the frequency analysis; a filter generation means for generating a bandpass filter with a passband set based on the two peak frequencies; and a data generation means for applying the bandpass filter to the first time-series data to generate second time-series data including the biological information.
19. An imaging device comprising an image sensor for imaging the subject and the measuring device described in claim 18.
20. A system characterized by comprising an imaging device for imaging the subject and the measuring device described in claim 18.
