New Application of an Instantaneous Frequency Parameter for Assessing Far Infrared Fabric E ﬀ ects in Aged Subjects

: A microcirculation microscope has recently been introduced to reveal ﬁnger blood ﬂow changes by visualization, before and after using far-infrared fabric. Digital volume pulses (DVPs) from the dominant index ﬁngertip of healthy young subjects (Group 1, n = 66) and healthy upper middle-aged subjects (Group 2, n = 33) were acquired through a photoplethysmographic electrical device (PED). By using the one intrinsic mode function (i.e., IMF5), an instantaneous frequency di ﬀ erence ( ∆ f Emax ) was revealed through the second part of the Hilbert–Huang transformation. Parameters from DVPs in the time domain, i.e., the sti ﬀ ness index, crest time, crest time ratio, and ﬁnger perfusion index, were also obtained for comparison. The results showed signiﬁcant di ﬀ erences in FPI and ∆ f Emax between the two groups ( p = 0.002 and p = 0.043, respectively). A signiﬁcant ∆ f Emax was also noted for the two groups under the e ﬀ ects of far-infrared radiation (FIR) (Group 1: p = 0.046; Group 2: p = 0.002). In conclusion, this study aimed to validate a self-developed and economical device, with a good extensibility, which can be operated in a domestic setting, and to demonstrate that the PED performed quantitative indexes on ﬁnger blood ﬂow comparable to those investigated through a microcirculation microscope. electrical device (PED) system. This system consists of ( A ) one photoplethysmography sensor, ( B ) ﬁltration and ampliﬁcation circuits, and a USB-6008 DAQ, and ( C ) a notebook computer for real data analysis. It can have great potential in both research and clinical applications. J.-J.C., H.-T.W.; administration: H.-C.W., resources:


Introduction
Far-infrared radiation (FIR), which comprises electromagnetic waves of 3-100 µm [1], produces physiological actions not only because of its high power in human tissue but also because of its ability to elicit both heat-related [2] and non-heat-related [3] biological effects. It has been shown that FIR causes vessel vasodilatation, thereby improving human tissue perfusion [2,4,5] and skin micro perfusion in • a notebook computer for real data analysis.
Although 30-min DVPs on the index finger of the dominant hand were collected, with the PED as described previously in [22,23] for offline computation, the two 1-min digitized DVP signals were processed through the USB-6008 DAQ for the calculation of real-time parameters using the LabVIEW G programming language in the current study. Immediate information on all of the parameters in Section 2.2.1, therefore, can be provided for all testing subjects ( Figure 2).

Study Population and Grouping
Between September 2018 and August 2019, 108 volunteers were originally enrolled for this study. The study proceeded in accordance with the Declaration of Helsinki, and the protocol was approved by the Ethics Committee of Ningxia Medical University Hospital (Yinchuan City, Ningxia Province, PRC) (No. 2018-229). All subjects gave their informed consent for inclusion before they participated in the study. Of the 108 subjects, nine were excluded due to unstable or incomplete waveform data acquisition. The remaining 99 subjects were then divided into two groups, including 66 healthy young subjects (Group 1, age range: 18-40) and 33 healthy upper middle-aged subjects (Group 2, age range: 41-70) [18]. All of the subjects had no personal or family history of cardiovascular diseases.

Study Protocol
All measurements were performed during two periods in the morning and afternoon (i.e., 08:30-10:30 and 14: 30-16.30). All subjects were asked to refrain from caffeine-containing beverages for at least 8 h before each testing. In addition, to minimize potential errors in the infrared sensor readings arising from involuntary vibrations of the subjects, all participants were allowed to rest in a supine position for 3 min in a quiet room with a temperature maintained at 26 ± 1 • C. The finger blood flow assessment-related parameters (i.e., SI, CT, CTR, FPI, and ∆f Emax ) were measured and calculated by our new photoplethysmographic electrical device (PED) for finger blood flow assessment under the radiation effect of far-infrared fabric. According to previous studies [14,19,20], it has been proposed that the stiffness index can be calculated from the body height divided by the transit time (T DVP ) (i.e., the difference time between systolic pulse peak and the following diastolic pulse peak in seconds) ( Figure 2

Photoplethysmographic Electrical
Hardware block diagram of the PED system. After filtration and amplification, waveform signals from the photoplethysmography (PPG) sensor went through a USB-6008 DAQ, which stored the digitized digital volume pulse signals in a computer using the LabVIEW G programming language for data analysis, evaluating the applications of PPG contour analysis in the time domain and the Hilbert-Huang transformation (HHT) domain. CT: crest time (i.e., time from the foot point to the peak of a waveform); CTR: crest time ratio; FPI: finger perfusion index; SI: stiffness index; ΔfEmax: instantaneous frequency difference (i.e., fEmax: the instantaneous frequency of maximal energy, before the use of far-infrared fabric (SY2); fEmaxSY2: the instantaneous frequency of maximal energy with the SY2 attached).

Statistical Analysis
All average values are expressed as mean ± SD. One sample Kolmogorov-Smirnov test was adopted for testing the normality of the distribution, while the Statistical Package for the Social Sciences (SPSS, version 14.0 for Windows, SPSS Inc. Chicago, II) was used for verifying the homoscedasticity of the variables. The significance of difference in anthropometric and computational parameters (i.e., CT, CTR, SI, FPI, and ΔfEmax) between the two groups was proven using an independent sample t-test. The correlation between ΔfEmax and risk factors for the two different groups was determined using Pearson's correlation test. A probability value (p) of less than 0.05 was considered statistically significant.

Characteristics of the Testing Subjects
Comparison of the demographic and anthropometric parameters of the testing subjects showed no remarkable difference in body height and body weight between the two groups (both p > 0.05). On the other hand, significantly lower levels of age as well as a reduced body mass index were noted in Group 1, compared with Group 2 (both p < 0.05) ( Table 1).  Hardware block diagram of the PED system. After filtration and amplification, waveform signals from the photoplethysmography (PPG) sensor went through a USB-6008 DAQ, which stored the digitized digital volume pulse signals in a computer using the LabVIEW G programming language for data analysis, evaluating the applications of PPG contour analysis in the time domain and the Hilbert-Huang transformation (HHT) domain. CT: crest time (i.e., time from the foot point to the peak of a waveform); CTR: crest time ratio; FPI: finger perfusion index; SI: stiffness index; ∆f Emax : instantaneous frequency difference (i.e., f Emax : the instantaneous frequency of maximal energy, before the use of far-infrared fabric (SY2); f EmaxSY2 : the instantaneous frequency of maximal energy with the SY2 attached).
In the present study, two SI values of each subject were calculated from two 1-min DVP waveforms, which were recorded in a PED for finger blood flow comparison under the effect of far-infrared fabric ( Figure 2).

•
Crest Time (CT) and Crest Time Ratio (CTR) The crest time is the time interval from the foot point of the DVP waveform to the first peak in seconds. The crest time ratio can be calculated from the CT divided by the cycle time (i.e., duration from the foot point of one wave to another in seconds) ( Figure 2).
Based on previous findings [14,15], CT and CTR indices showed significant positive correlations with SI. In the present study, we attempted to find the CT and CTR indices, irrespective of whether or not they reflect the effect of far-infrared fabric ( Figure 2).

•
Finger Perfusion Index (FPI) In accordance with the finding of previous studies [16,21] that the integral of blood flow under an arterial pulse waveform reflects blood perfusion within a defined time period, the area under the finger arterial waveform contour is a rational estimation of the perfusion volume of the finger blood flow. The areas under the arterial waveforms within 1 min of the baseline recording were summated (Area Baseline ). The areas under the waveforms during SY2-attachment represented 1 min of recordings (Area SY2 ). The change in blood flow every 1 min after SY2 attachment, compared with the baseline, was then equal to (Area SY2 − Area Baseline ). Therefore, values were obtained for each testing subject. The figure perfusion index (FPI) was defined as the percentage change of blood flow every 1 min before/after SY2 attachment [16,21], defined as follows.
On the other hand, the significance of the Hilbert-Huang spectrum, which is the other component of HHT that investigates the relationship among energy, time, and frequency, has not been explored and verified in terms of the effect of far-infrared fabric. By recruiting diabetic patients and healthy subjects, a previous study aimed to investigate the significance of instantaneous frequency in vascular health. The values of the instantaneous frequency of maximal energy (f Emax ) were higher in diabetic patients than those in non-diabetic volunteers [17].
In general, 3000 points of sampled DVP signals (i.e., y(t) in Equations (4) and (5)) comprise noise-free and noise components: in which s(t) and n(t) are the true signal and white noise, respectively. As a result, the signal y(t), after ensemble empirical mode decomposition (EEMD), can be expressed as in which IMFn(t) and r n (t) are intrinsic mode functions and the residue function, respectively. By using the one intrinsic mode function (IMF5) and marginal spectral density after Hilbert-Huang spectrum analysis, the instantaneous frequency of maximal energy (f Emax ) was then obtained (Figure 2) after real time computation. The value of the instantaneous frequency of maximal energy within 3 s of the baseline recording was summated (f Emax ), whereas f EmaxSY2 represented the instantaneous frequency of maximal energy with SY2 attachment. Hence, the instantaneous frequency difference (∆f Emax ) was defined as the difference change of the instantaneous frequency of maximal energy after SY2 was attached: ∆f

Procedures of Examinations
Each subject received two different stages of measurements: Session 1 (baseline stage) and Session 2 (far-infrared fabric attached). Before data acquisition, one PPG detector was attached to the index fingertip of the dominant hand. The digital volume pulse (DVP) waveforms were recorded for a 1-min duration in Session 1. Subsequently, a far-infrared fabric (i.e., Leadtek-SY2®, medical equipment, class 1, No. 004873, Taiwan Food and Drug Administration) was attached (i.e., SY2 attachment) on the dominant wrist of the testing subject for Session 2. Then, after 2 min of rest [11], the digital volume pulse (DVP) waveforms were also recorded for the second 1-min duration. In the seven-minute measurement, two 1-min DVP waveforms were recorded for the computation of parameters (i.e., SI, CT, CTR, FPI, and ∆f Emax ) in the PED for finger blood flow assessment under the effect of far-infrared fabric. The values of all parameters were averaged from all the beats in a selected time interval (i.e., a 1-min DVP signal or a 3-s DVP signal).

Hardware of PED
The hardware of the PED system consists of four major components, namely, the PPG sensor, analog circuits for filters and analog amplification, a data acquisition module (i.e., USB-6008 DAQ, National Instruments, Austin, TX), and a notebook computer for real data analysis ( Figure 1). The system was used as a diagnostic tool by Ningxia Medical University Hospital. With a PED, many useful experiments and applications can also be conducted in a home medical setting.
The function of each component is described as follows: • a PPG sensor: one pair of infrared transmitter and receiver with a 940 nm wavelength; • analog filters: a 2nd order band pass filter, with cut-off frequencies of 0.48-10 Hz; • an analog amplification circuit: digital volume pulses (DVPs) with 1-10 mV; • a USB-6008 DAQ: a sampling frequency of 1000 Hz and 12-bit ADC with USB (DVPs stored in a computer for later computation); • a notebook computer for real data analysis.
Although 30-min DVPs on the index finger of the dominant hand were collected, with the PED as described previously in [22,23] for offline computation, the two 1-min digitized DVP signals were processed through the USB-6008 DAQ for the calculation of real-time parameters using the LabVIEW G programming language in the current study. Immediate information on all of the parameters in Section 2.2.1, therefore, can be provided for all testing subjects ( Figure 2).

Statistical Analysis
All average values are expressed as mean ± SD. One sample Kolmogorov-Smirnov test was adopted for testing the normality of the distribution, while the Statistical Package for the Social Sciences (SPSS, version 14.0 for Windows, SPSS Inc. Chicago, II) was used for verifying the homoscedasticity of the variables. The significance of difference in anthropometric and computational parameters (i.e., CT, CTR, SI, FPI, and ∆f Emax ) between the two groups was proven using an independent sample t-test. The correlation between ∆f Emax and risk factors for the two different groups was determined using Pearson's correlation test. A probability value (p) of less than 0.05 was considered statistically significant.

Characteristics of the Testing Subjects
Comparison of the demographic and anthropometric parameters of the testing subjects showed no remarkable difference in body height and body weight between the two groups (both p > 0.05). On the other hand, significantly lower levels of age as well as a reduced body mass index were noted in Group 1, compared with Group 2 (both p < 0.05) ( Table 1). There are very similar DVP waveforms before vs. after SY2 was attached for the case of one young man. This is the reason why SI, CT, and CTR were not changed very much. In addition, the finger perfusion became smaller (FPI = −9.82%) ( Figure 3a). As for one 48-year-old woman in Group 2, the DVP waves increased after SY2 was attached, with a notable increase in finger perfusion (i.e., FPI = 70.89%), whereas SI, CT, and CTR were still not changed very much (Figure 3b). As shown in Figure 3b, both FPI and ∆f Emax successfully discriminated between the two states, in which SY2 was attached or not for the 48-year-old woman. The parameters (i.e., SI, CT, CTR, FPI, and ∆f Emax ) were calculated in 10 s on the PED and were then provided for all testing subjects.

Failure of CT, CTR, SI, and FPI Change during SY2 Attachment for a Group 1 Subject
There are very similar DVP waveforms before vs. after SY2 was attached for the case of one young man. This is the reason why SI, CT, and CTR were not changed very much. In addition, the finger perfusion became smaller (FPI = −9.82%) ( Figure 3a). As for one 48-year-old woman in Group 2, the DVP waves increased after SY2 was attached, with a notable increase in finger perfusion (i.e., FPI = 70.89%), whereas SI, CT, and CTR were still not changed very much (Figure 3b). As shown in Figure 3b, both FPI and ΔfEmax successfully discriminated between the two states, in which SY2 was attached or not for the 48-year-old woman. The parameters (i.e., SI, CT, CTR, FPI, and ΔfEmax) were calculated in 10 s on the PED and were then provided for all testing subjects. (a)

Impact of Far-Infrared Fabric for Five Parameters in the Same Group
In terms of the impact of far-infrared fabric (i.e., Leadtek-SY2) on the five parameters in Group 1, no significant difference was noted in CT, CTR, FP, and SI (p = 0.721, 0.787, 0.845, and 0.493, respectively), whereas fEmax (p = 0.046) was higher before SY2 was attached than after SY2 was attached ( Table 2). As for Group 2, there were still no significant differences noted in CT, CTR, FP, and SI (p = 0.652, 0.986, 0.219, and 0.297, respectively), whereas fEmax (p = 0.002) was higher before SY2 was attached than after SY2 was attached ( Table 2).

Impact of Far-Infrared Fabric for Five Parameters in the Same Group
In terms of the impact of far-infrared fabric (i.e., Leadtek-SY2) on the five parameters in Group 1, no significant difference was noted in CT, CTR, FP, and SI (p = 0.721, 0.787, 0.845, and 0.493, respectively), whereas f Emax (p = 0.046) was higher before SY2 was attached than after SY2 was attached ( Table 2). As for Group 2, there were still no significant differences noted in CT, CTR, FP, and SI (p = 0.652, 0.986, Electronics 2020, 9, 138 9 of 14 0.219, and 0.297, respectively), whereas f Emax (p = 0.002) was higher before SY2 was attached than after SY2 was attached (Table 2).

Comparison of FPI and ∆f Emax for the Two Groups
A significant difference was noted in FPI and ∆f Emax between the two groups (p = 0.002 and p = 0.043, respectively), with both values of FPI and ∆f Emax being lower in Group 1 than in Group 2 ( Figure 4).

Comparison of FPI and ΔfEmax for the Two Groups
A significant difference was noted in FPI and ΔfEmax between the two groups (p = 0.002 and p = 0.043, respectively), with both values of FPI and ΔfEmax being lower in Group 1 than in Group 2 ( Figure 4).

Correlations between ΔfEmax and Demographic and Anthropometric Parameters
Despite the lack of significant correlation between ΔfEmax and demographic and anthropometric parameters (i.e., age, body height, body weight, and BMI) for the subjects in Group 1, ΔfEmax was found to be positively associated with body weight (p = 0.021) and body mass index (BMI) (p = 0.005) ( Table 3).

Correlations between ∆f Emax and Demographic and Anthropometric Parameters
Despite the lack of significant correlation between ∆f Emax and demographic and anthropometric parameters (i.e., age, body height, body weight, and BMI) for the subjects in Group 1, ∆f Emax was found to be positively associated with body weight (p = 0.021) and body mass index (BMI) (p = 0.005) ( Table 3). Table 3. Pearson correlations between the instantaneous frequency difference ∆f Emax and demographic and anthropometric parameters in young and upper middle-aged subjects.

Multivariate Regression Analysis for ∆f Emax
The demographic and anthropometric parameters of the upper middle-aged subjects (i.e., Group 2) found to be significantly associated with ∆f Emax in this study using Pearson's correlation test were body weight and body mass index (BMI), for which multivariate analysis was performed, as follows (p = 0.02): ∆f Emax = −0.902 + 0.002 × Body weight + 0.043 × BMI + error (7)

Discussion
The clinical benefits of combined acupuncture-FIR heat lamps in enhancing peripheral perfusion and parasympathetic activity have been demonstrated [7]. Compared with FIR heat lamps, FIR-emitting ceramics, and fabrics with a low power of FIR can also be effective for finger blood perfusion. A microcirculation microscope has recently been introduced to reveal the finger blood flow changes by visualization, before and after the far-infrared fabric was attached to the wrist. There are some very interesting studies that focused on accurate measurements of finger blood flow velocity using a microcirculation image analysis system [10], magnetic resonance imaging [24], ultrasound [25], and a heat source chip and a temperature sensor [26]. However, those measuring instruments could be expensive, have complete technical independence, or be less portable. This study aimed to investigate the usefulness of a photoplethysmographic electrical device (PED) design, distinguishing healthy young subjects from elder subjects using quantitative data output. One PPG sensor, analog filtering and amplifying circuits, a USB-6008 DAQ open-ended program, and a notebook computer were needed for the PED in the present study, which it was able to provide quantitative data for proof of the effect of far-infrared fabric at a low cost.
As Figures 1 and 2 show, a PED system was proposed, with the application of parameters (i.e., SI [14,17,19,20], CT and CTR [14,17], FPI [16,21], and ∆f Emax [17]) for finger blood flow (relative to microcirculation) assessment under the effect of far-infrared fabric. In general, microcirculation is known to be affected by cardiovascular risk factors, including aging, hypertension, and diabetes [27,28]. Therefore, both values of FPI and ∆f Emax in the present study were lower in Group 1 than in Group 2 ( Figure 3). This verified the first hypothesis of this study that a PED can perform a quantitative index (i.e., FPI and ∆f Emax ) on microvascular blood flow change due to the attachment of far-infrared fabric. Moreover, there was no significant difference noted in CT, CTR, FP, and SI (all p > 0.05) for the two groups, whereas f Emax (p < 0.05) was larger before SY2 was attached than after SY2 was attached ( Table 2). In Table 2 with f Emax and in Figure 4 with FPI, significant differences are shown, but not in other parameters. It is shown that IMF5 best represents DVP signals, through which SI can be calculated. It was hypothesized in a previous study [17] that physiological information hidden in the inseparable signal of a noise-free IMF as reflected in the energy-frequency spectrum can be revealed through Hilbert-Huang spectrum analysis. The sensitivity and simplicity of the PED, shown in the present study, therefore have a significant implication for the development of the next generation of non-invasive portable finger microcirculation monitoring devices. In addition, people with high BMI may contribute to microvascular disease and affect clinical functional outcomes in the older population [29]. For healthy humans, body weight is a major determinant of the resting rate of muscle sympathetic nerve dysfunction [30]. This is consistent with the same findings that body weight and body mass index were found to be the two most important determining factors of finger microcirculation assessment in the present study (Table 3).
PPG is a non-invasive, low cost, and simple optical measurement technique applied at the fingertip to measure DVP signals for physiological parameters. Scientific interest has continued to look beyond the pulse oximetry and more into the new potential applications of this technology in clinical settings [31]. In the first generation of self-developed devices, utilizing 8-channel PPG and ECG, a previous study [16] showed significant elevations in peripheral blood flow and autonomic nervous function after acupuncture at zusanli st36 acupoint. However, the device had limitations. First, the process of data acquisition was time-consuming and took more than 30 min. Second, the offline computation for EEMD and short-time multiscale entropy index was not able to give immediate information to testing subjects. Furthermore, the first generation of proposed devices in [14] and [22] collected PPG signals for a very long period of time (i.e., 30 min) on the index finger, all for offline analysis for the computation of parameters. Although f Emax was computed using 5 min of sampled data in the first generation of self-developed devices in [17], offline computations also required the use of another package (i.e., Matlab) on a PC. On the other hand, the values of all parameters were averaged from DVPs in a selected time interval (i.e., a 1-min DVP signal for the comparison parameters or a 3-s DVP signal for instantaneous frequency difference), all for real-time computation in the new generation of self-developed devices (i.e., PED) in the current study.
This study has its limitations. First, although the current technique of PPG is a very popular non-invasive method of waveform contour analysis in assessing finger blood perfusion, the data obtained are always affected by various environmental and physiological factors. The conventional pulse wave velocity (PWV) was not measured in this study, because one pair of infrared transmitter and receiver was used. Secondly, it is not a large-scale investigation because of the limited number of subjects in each group. Thirdly, the DVP waveforms may also be affected by finger vibrations, which may lead to a distortion of the ensemble averaged signals for the subsequent CT, CTR, and SI determinations. Finally, only young and upper middle-aged healthy subjects were enrolled in this study, so the impact of disease on treatment outcomes was not evaluated.
Comparisons of the advantages and disadvantages of non-invasive devices for the evaluation of finger blood flow between the use of a microcirculation microscope and the method of the present study are summarized in Table 4.

Conclusions
The results of the current study not only propose a versatile photoplethysmographic electrical device for finger blood perfusion assessment under the effect of a far-infrared fabric, but also suggest the possibility of the clinical use of instantaneous frequency for FIR treatment. Moreover, with this self-developed device, many useful parameters and applications can be conducted by researchers. That is to say, the new generation of self-developed devices (i.e., PED) in the current study could be adopted in assessing the impact of acupuncture on peripheral blood flow and autonomic function for elder and overweight subjects. In addition, the PED could be used in the field of hyperbaric oxygen therapy for type 2 diabetes patients in glucose homeostasis improvement and diabetic foot wound care.