**6.2 Calculation method of blood pressure from the FBG sensor signal**

When blood pressure was calculated from FBG sensor signal, PLS regression analysis, which is a widely known multivariate analysis among others, was used. The PLS regression analysis can construct calibration curves from the explanatory and objective variables. At this time, it is a feature to construct a calibration curve on the premise that an explanatory variable and an objective variable contain errors. The explanatory variable is the FBG sensor signal waveform, and the objective variable is the blood pressure measured simultaneously by the electronic sphygmomanometer. The FBG sensor signal is processed through the following steps [19].


The step 1 of the signal processing a range for covering a signal with a pulse rate of 30–300 times/min. In step 4 of the signal processing, considering that the measurement time of the electronic sphygmomanometer is ~30 s, the average is calculated for the pulse wave signals measured within that time. The step 5 of signal processing is to cancel the vertical axis fluctuations caused by pressure while installing the FBG sensor in humans. The step 6 of signal processing is to cancel the pulse rate fluctuations caused by respiratory sinus arrhythmia.

The FBG sensor signal processed through the aforementioned signal processing steps is used as an explanatory variable, the blood pressure value of the electronic sphygmomanometer measured simultaneously is used as a target variable, and a calibration curve is constructed by PLS regression analysis. The newly measured FBG sensor signal is substituted into this calibration curve to calculate the blood pressure.

#### **6.3 Experimental result of calculating the blood pressure**

This experiment was performed on three subjects. A schematic of the experimental blood pressure measurement is shown in **Figure 9** [19]. The posture of the subject was supine, and the FBG sensor was installed at the pulsation point of the radial artery of the right wrist. The reference blood pressure value (objective

*Fiber Optic Sensing - Principle, Measurement and Applications*

subjects. The measurement accuracies observed for the three subjects were 0.4, 0.6, and 0.4 per min, which were considered to be reasonably good. The high measurement accuracy was observed even for different respiration rates of the same subject and for different subjects. The change in pulse rate (change in PPI interval) due to respiratory dynamic arrhythmia was very small; however, since the sampling rate of the FBG sensor was 10 kHz, it is considered that the calculated respiratory rate was accurate. This measurement method can calculate a respiratory rate in the range of 6–10 bpm; therefore, it is suitable for measurement of slow breathing (~12 bpm or less). It is evident from the above results that the high accuracy of measurement of

**6. Calculating of blood pressure from the waveform of FBG sensor signal**

In this section, the blood pressure is calculated from the waveform of the FBG sensor signal. As shown in Section 3, the FBG sensor signal is measured representing the pressure of the blood flow that causes a change in the diameter of the blood vessel. Pulsation is a distortion that causes an arterial distortion on the skin surface. Therefore, information on blood pressure is considered to be present in the FBG

A signal measured with a general photoelectric pulse wave sensor is a volume pulse wave signal indicating the volume of blood. A signal obtained by second derivative of the volume pulse wave signal is an acceleration plethysmogram. The basic shape of acceleration plethysmogram includes five peaks [18]. The A-wave to the E-wave are called initial systolic positive wave, initial contraction negative wave, mid-systole re-elevation wave, post-contraction descent wave, and expansion initial positive wave, respectively. Therefore, an acceleration pulse wave contains information on systole and diastole of the heart. The first derivative signal of the

respiratory rate is attributable to the high sampling rate.

**6.1 Waveform of the FBG sensor signal**

sensor signal from which a distortion is measured.

**50**

**Figure 8.**

*Result of calculating respiratory rate.*

#### **Figure 9.** *Experimental image of blood pressure measurement.*

variable) was measured simultaneously with the electronic sphygmomanometer installed at the left upper arm. Systolic and diastolic blood pressures were measured with an electronic sphygmomanometer. In the calculation of the systolic blood pressure, the signal-processed FBG sensor signal waveform was used as an explanatory variable, and the systolic blood pressure measured simultaneously with the electronic blood pressure monitor was used as the objective variable. Similarly, in the calculation of the diastolic blood pressure, the same FBG sensor signal waveform and the diastolic blood pressure measured simultaneously with electronic sphygmomanometer were used. The measurement time was 30 s, while the number of measurements was 75 times. Whereas 50 data points were used for construction of calibration curve, the remaining 25 data points were assigned to the calibration curve and used as verification data for blood pressure calculation. The target measurement accuracy was ±5 mmHg.

**Table 2** shows the calibration curve construction data sets of systolic and diastolic blood pressures in each subject [19]. A calibration curve for calculating systolic blood pressure or diastolic blood pressure was constructed using the data sets of each subject. **Table 3** shows verification data sets for calculation of systolic and diastolic blood pressures for each subject. The verification data set is substituted into the constructed calibration curve, and systolic and diastolic blood


**53**

**Figure 10.**

*Vital Sign Measurement Using FBG Sensor for New Wearable Sensor Development*

**(mmHg)**

**Min (mmHg)**

**Ave (mmHg)**

**Subject Number Max** 

pressures were calculated. For example, a calibration curve was constructed with 50 data points of the systolic blood pressure of the subject A; the 25 data points of the systolic blood pressure of verification data of the subject A were substituted into the calibration curve, and the systolic blood pressure of subject A was calculated. Similarly, a calibration curve was constructed with 50 data points of the diastolic blood pressure of the subject B in **Table 2**; the 25 data points of validation data of the diastolic blood pressure of the subject B in **Table 3** were substituted into the calibration curve, and the diastolic blood pressure of subject B was calculated. **Figure 10** shows a scatter plot of reference blood pressure and calculated blood pressure during systole and diastole of each subject. **Table 4** shows the results of blood pressure calculation, whereby it was observed that the calculation accuracy of systolic and diastolic blood pressures were ±5 mmHg, and it was calculated with the same blood pressure value as that of a commercially available blood pressure monitor. In the case of the systolic blood pressure, the average value of the verification data for blood pressure calculation was 110.9 mmHg, while the average value

*Scatter plot of reference blood pressure and calculated blood pressure during systolic and diastolic. (A) Calculated* 

*result in systolic blood pressure. (B) Calculated result in diastolic blood pressure.*

A 25 131 100 110.1 B 25 138 112 122.2 C 25 110 93 100.4 Average systolic blood pressure in data set (mmHg) 110.9

A 25 77 46 60.3 B 25 76 58 68.0 C 25 65 39 53.1 Average diastolic blood pressure in data set (mmHg) 60.5

*DOI: http://dx.doi.org/10.5772/intechopen.84186*

(a) Systolic blood pressure data set

(b) Diastolic blood pressure data set

*Verification data set in each subjects.*

**Table 3.**

**Table 2.** *Calibration curve construction data sets.*


*Vital Sign Measurement Using FBG Sensor for New Wearable Sensor Development DOI: http://dx.doi.org/10.5772/intechopen.84186*

#### **Table 3.**

*Fiber Optic Sensing - Principle, Measurement and Applications*

measurement accuracy was ±5 mmHg.

(a) Systolic blood pressure data set

(b) Diastolic blood pressure data set

*Calibration curve construction data sets.*

*Experimental image of blood pressure measurement.*

**Figure 9.**

variable) was measured simultaneously with the electronic sphygmomanometer installed at the left upper arm. Systolic and diastolic blood pressures were measured with an electronic sphygmomanometer. In the calculation of the systolic blood pressure, the signal-processed FBG sensor signal waveform was used as an explanatory variable, and the systolic blood pressure measured simultaneously with the electronic blood pressure monitor was used as the objective variable. Similarly, in the calculation of the diastolic blood pressure, the same FBG sensor signal waveform and the diastolic blood pressure measured simultaneously with electronic sphygmomanometer were used. The measurement time was 30 s, while the number of measurements was 75 times. Whereas 50 data points were used for construction of calibration curve, the remaining 25 data points were assigned to the calibration curve and used as verification data for blood pressure calculation. The target

**Table 2** shows the calibration curve construction data sets of systolic and diastolic blood pressures in each subject [19]. A calibration curve for calculating systolic blood pressure or diastolic blood pressure was constructed using the data sets of each subject. **Table 3** shows verification data sets for calculation of systolic and diastolic blood pressures for each subject. The verification data set is substituted into the constructed calibration curve, and systolic and diastolic blood

A 50 125 100 111.3 B 50 136 113 123.1 C 50 111 93 100.9

A 50 79 46 62.7 B 50 80 56 68.2 C 50 60 46 53.7

**(mmHg)**

**Min (mmHg)**

**Ave (mmHg)**

**Subject Number Max** 

**52**

**Table 2.**

*Verification data set in each subjects.*

pressures were calculated. For example, a calibration curve was constructed with 50 data points of the systolic blood pressure of the subject A; the 25 data points of the systolic blood pressure of verification data of the subject A were substituted into the calibration curve, and the systolic blood pressure of subject A was calculated. Similarly, a calibration curve was constructed with 50 data points of the diastolic blood pressure of the subject B in **Table 2**; the 25 data points of validation data of the diastolic blood pressure of the subject B in **Table 3** were substituted into the calibration curve, and the diastolic blood pressure of subject B was calculated.

**Figure 10** shows a scatter plot of reference blood pressure and calculated blood pressure during systole and diastole of each subject. **Table 4** shows the results of blood pressure calculation, whereby it was observed that the calculation accuracy of systolic and diastolic blood pressures were ±5 mmHg, and it was calculated with the same blood pressure value as that of a commercially available blood pressure monitor. In the case of the systolic blood pressure, the average value of the verification data for blood pressure calculation was 110.9 mmHg, while the average value

#### **Figure 10.**

*Scatter plot of reference blood pressure and calculated blood pressure during systolic and diastolic. (A) Calculated result in systolic blood pressure. (B) Calculated result in diastolic blood pressure.*


#### **Table 4.**

*Results of calculating blood pressure.*

of the calculation accuracy was 2.9 mmHg (~2.6%). Similarly, in the diastolic blood pressure, the average value of the verification data for blood pressure calculation was 60.5 mmHg, while the average value of the calculation accuracy was 2.8 mmHg (~4.7%). These results indicate that the calculation accuracy of diastolic blood pressure is lower than that of systolic blood pressure. This is due to step 6 of signal processing, whereby the canceling of the pulse rate fluctuation was performed in order to calculate the FBG sensor signal by the PLS regression analysis. The so-called latter half portion of a single beat of the FBG sensor signal is truncated. There is a peak of expansion initial positive wave in this part, which represents the diastole of the heart. Therefore, it is considered that the deletion of the diastolic information from the FBG sensor signal caused a decline in the calculation accuracy. In other words, in this calculation method, since the negative characteristics of the signal processing is reflected in the result, the blood pressure is calculated from the movement of the heart included in the FBG sensor signal. However, considering that the results of all blood pressure calculations were ±5 mmHg, it is considered that the blood pressure was calculated with high calculation accuracy. Therefore, it is established that the blood pressure can be calculated with high calculation accuracy by constructing the calibration curve by PLS regression analysis of the waveform of the FBG sensor signal. It is evident from this result that it is possible to calculate blood pressure from the same FBG sensor signal in addition to pulse rate, respiratory rate, and stress load.
