http://informahealthcare.com/jmt ISSN: 0309-1902 (print), 1464-522X (electronic) J Med Eng Technol, 2015; 39(1): 79–85 ! 2015 Informa UK Ltd. DOI: 10.3109/03091902.2014.979953
REVIEW
Relating external compressing pressure to mean arterial pressure in non-invasive blood pressure measurements K. Y. Chin*1 and R. B. Panerai2,3
J Med Eng Technol Downloaded from informahealthcare.com by Nyu Medical Center on 05/28/15 For personal use only.
1
Department of Medical Physics, University Hospitals of Leicester NHS Trust, Leicester Royal Infirmary, Leicester LE1 5WW, UK, 2Department of Cardiovascular Sciences, University of Leicester, Leicester Royal Infirmary, Leicester LE1 5WW, UK, and 3Leicester NIHR Biomedical Research Unit in Cardiovascular Sciences, Glenfield Hospital, Leicester LE3 9QP, UK Abstract
Keywords
Arterial volume clamping uses external compression of an artery to provide continuous noninvasive measurement of arterial blood pressure. It has been assumed that mean arterial pressure (MAP) corresponds to the point where unloading leads to the maximum oscillation of the arterial wall as reflected by photoplethysmogram (PPG), an assumption that has been challenged. Five subjects were recruited for the study (three males, mean age (SD) ¼ 32 (15) years). The PPG waveform was analysed to identify the relationship between the external compressing pressure, PPG pulse amplitude and MAP. Two separate tests were carried out at compression step intervals of 10 mmHg and 2 mmHg, respectively. No significant differences were found between the two tests. The bias between the compressing pressure and the MAP was 4.7 ± 5.63 mmHg (p50.001) showing a normal distribution. Further research is needed to identify optimal algorithms for estimation of MAP using PPG associated with arterial compression.
Blood pressure, non-invasive, photoplethysmography, pulse amplitude, volume clamping technique
1. Introduction Continuous recording of arterial blood pressure (ABP) has been increasingly used for patient monitoring and cardiovascular research. Although intravascular recordings are still the gold standard technique for this purpose, its relatively high cost, associated to the risk of infection and other complications, has stimulated the search for alternative non-invasive methods [1,2]. Most commercial equipment for this purpose have been dominated by the use of arterial volume clamping, which provides continuous non-invasive estimates of ABP using photoplethysmography to detect changes in arterial volume, coupled with feedback controlled external compression of a peripheral artery, usually the digital artery in the middle or ring finger [3]. One central assumption of the arterial volume clamping approach though is the estimation of the mean arterial pressure (MAP). Most devices assume that MAP corresponds to the external compressing pressure when the photoplethysmogram (PPG) signal reaches its maximum oscillation [4,5]. However, this view has been challenged by several investigators and it remains controversial [6,7]. Of considerable importance, a similar assumption is adopted by other techniques or devices used for non-invasive measurement of ABP such as oscillometry, which can only provide intermittent measurements. In both cases though, accurate detection *Corresponding author. Email: [email protected]
History Received 13 August 2014 Revised 2 October 2014 Accepted 19 October 2014
of MAP is highly relevant as it determines the overall accuracy of the ABP values output by the device. In this study, we used arterial volume clamping to investigate the relationship between MAP and the external compressing pressure of the digital artery, testing two specific hypotheses: (i) that the MAP corresponds to the external compressing pressure leading to the maximum amplitude of the PPG signal; and (ii) that the relationship between MAP and external compression is independent of the compressing pressure step changes.
2. Methods 2.1. Volume clamping technique Measurement devices using the volume clamping technique have been reported in the literature for the finger [8], brachial [9] and superficial temporal [10] arteries. However, nearly all recordings in clinical and research protocols had been obtained from the finger artery due to the commercial availability of the FinapresÕ (Ohmeda Medical; Englewood, Colorado, USA) and its later models PortapresÕ and FinometerÕ (Finapres Medical Systems; Amsterdam, Netherlands) [11]. Other commercial devices currently available include CNAP500Õ (CNSystems; Graz, Austria) and NexfinÕ (BMEYE; Amsterdam, Netherlands). These devices also monitor the ABP in the finger. In the literature, the volume clamping technique is also known as the vascular unloading [3] or volume compensation technique [12].
J Med Eng Technol Downloaded from informahealthcare.com by Nyu Medical Center on 05/28/15 For personal use only.
80
K. Y. Chin & R. B. Panerai
The volume clamping technique incorporates the use of photoplethysmography to detect changes in arterial blood volume. Photoplethysmography uses an infrared emitting light source to illuminate the tissue within which the artery of interest is located. Depending on the measurement site, a photodetector is placed in either reflectance or transmittance configuration to measure the light transmitted through, scattered and/or reflected by the pulsating arterial blood. The recording of this photo-signal produces the PPG. When a range of external compressing pressures is applied to the artery, the PPG waveshape shows changes in its amplitude as well as in its temporal pattern. An example is given in Figures 1(a) and (b) where the external compressing pressure range encompasses the diastolic and systolic pressures. As the external compressing pressure rises, it causes a gradual reduction of arterial blood and a shift of extra-cellular fluid in the tissue. This leads to a gradual increase in the PPG baseline. An oscillation component due to the pulsating arterial blood superimposes on the PPG baseline. The amplitude of the oscillation component increases to a maximum and then decreases until it disappears at higher external compressing pressures.
J Med Eng Technol, 2015; 39(1): 79–85
A reference value is derived from the PPG pulse that could produce zero transmural pressure. At zero transmural pressure, the external compressing pressure is equal to the intraarterial pressure. The reference value, for example, could be from the mean value of the PPG with the largest pulse amplitude. Once a PPG reference value is available, a servofeedback control mechanism is incorporated to continuously modulate the external pressure to partially compress (hence ‘volume clamping’) the pulsating artery to maintain the real time PPG at this reference value. The external compressing pressure is recorded continuously, thus representing a measurement of the ABP. 2.2. Subjects and measurements Five subjects were recruited (three males, mean age (SD) of 32 (15) years). Subjects had no known history of cardiovascular or other pathologies that would affect the blood circulation in the fingers. Participants were all staff from the Medical Physics department. Consent was given by each participant prior to the arrangement for the ABP measurements. The study was approved by the local ethics committee. Finapres finger cuffs of appropriate size were fitted on the middle and ring fingers of the right hand. The fitting and positioning of the finger cuffs followed the manufacturer’s guidelines. The finger cuff on the ring finger was used to measure the real-time continuous ABP waveform with the Finapres (‘Ohmeda 2300’ model). The finger cuff on the middle finger was operated by an in-house built PPG recording device. Two ipsilateral instead of contralateral fingers were used in order to minimize the potential difference in regional regulation of arterial blood flow between the two hands. Figure 2 shows the block diagram of the in-house built PPG recording device. A sequence of regulated compressing pressure was applied as command output from a desktop computer to compress the artery in the middle finger. Simultaneously, the optoelectronic components in the finger cuff monitored the blood volume change in the artery. The output from the photodetector was converted to a PPG signal through transimpedance amplification. The PPG was recorded and displayed in real time by the desktop computer. A separate channel was used to record the continuous ABP waveform measured by the Finapres. Both the Finapres and the PPG waveforms were simultaneously sampled at 100 Hz (DT302, Data TranslationÕ ). The recording was managed with the Agilent VEEProÕ visual programming software. 2.3. Protocols
Figure 1. (a) External compressing pressure recording. (b) Simultaneous recording of PPG waveform. The PPG pulse with the largest pulsation amplitude is indicated as ‘1’. The corresponding compressing pressure value is also shown (indicated as ‘2’). (c) An example PPG waveform shows changes in its amplitude as well as its temporal pattern. ‘s’ refers to the lowest PPG value during the systolic phase, whereas ‘d’ refers to the highest PPG value during the diastolic phase.
Each subject attended two measurement sessions (Test #1 and Test #2). During each session, the right hand rested on a bench at heart level. Subjects were asked to relax their fingers and breathe normally while sitting on a chair with back support. Subjects were also asked not to engage in any conversation, laughter, holding their breath or shifting about in their seated position. In Test #1, the compressing pressure in the cuff (denoted as Pcom) was first increased by 10 mmHg every 5 s. The range of Pcom encompassed the physiological range of systolic and diastolic pressures. When Pcom had increased above systolic
DOI: 10.3109/03091902.2014.979953
Non-invasive blood pressure measurements
81
J Med Eng Technol Downloaded from informahealthcare.com by Nyu Medical Center on 05/28/15 For personal use only.
Figure 2. Block diagram of the in-house built PPG recording device with Finapres cuff. The device is developed to monitor the arterial blood volume change in the middle finger to generate a PPG signal. The PPG waveform can be recorded, stored and displayed.
pressure, the recording was complete. It was then decreased at the same rate until it was near 0 mmHg. This corresponded to a separate recording. The cycle of increasing and decreasing the Pcom was then repeated, for a total of four sets of recordings. Throughout this process, the real time PPG waveform was recorded simultaneously. In Test #2, the same set-up and protocol were repeated except that the Pcom steps were now changed at 2 mmHg every 5 s. Again, four sets of recordings were obtained in each subject. The rationale for this was to compare results between different pressure step intervals. The Finapres has a built-in physiological self-calibration facility known as the ‘Physiocal’ [13]. It was kept switched on throughout the ABP measurement. This is to prevent any potential drifts in the MAP measured with the Finapres [14]. 2.4. Data analysis When the ABP increases during the systolic phase, the PPG decreases. This forms an inverse relationship between the arterial blood volume light absorption and the PPG. Figure 1(c) shows an example that, at the peak of the systolic phase, the PPG decreases to a local minimum, indicated as ‘s’. Conversely, as the ABP decreases to the diastolic pressure, the PPG increases to a local maximum, indicated as ‘d’ in the figure. The PPG pulse amplitude is measured between each ‘s’ and the preceding ‘d’ and denoted as PPGa (in Volts). Depending on the heart rate, three-to-five pulses were recorded at each Pcom. The PPGa from each of these pulses during the same Pcom were then averaged (PPGa (SD)). Corresponding ABP pulses were recorded with the Finapres and the average (SD) of systolic, mean and diastolic pressures were also computed (i.e. SBPF , MAPF and DBPF , respectively) (Figure 3a). Each MAP was calculated as the timeintegrated average of each ABP pulse waveform. The array of PPGa was plotted against the sequence of Pcom. Figure 3(b) shows an example.
^
Figure 3. Example computation of PdiffA from the PPGacurve. (a) The Pcom (light square) was decreased at 2 mmHg steps every 5 s. The Finapres ABP is shown as: SBPF (diamond), MAPF (square) and DBPF (triangle). (b) The array of PPGacorresponding to decreasing Pcom. (c) Graph in (b) was re-plotted against Pdiff (equation 1). The PPGa array is ^
superimposed by a suitable polynomial order curve (PPGa). The PdiffA is indicated. In this example, PdiffA ¼ 6.2 mmHg. In all the graphs, the error bars are the respective ± 1 SD.
Next, the difference between MAPF and the external compressing pressure (Pcom) which produces the largest PPG pulse amplitude was computed. The generic pressure difference (Pdiff) equation was defined as Pdiff ¼ MAPF Pcom
ð1Þ
82
K. Y. Chin & R. B. Panerai
Pdiff was calculated for each Pcom. Following Figure 3(b), the PPGa array was re-plotted against the Pdiff (Figure 3c). First to sixth order polynomials were tested to fit the experiment data for the pressure range of ±40 mmHg around Pdiff ¼ 0. The optimal fitted curve was denoted ^
PPGa, a representative ^example is depicted in Figure 3(c). At the peak of the PPGacurve, the corresponding Pdiff was computed. This specific Pdiff was denoted as PdiffA (mmHg). PdiffA indicates the bias between the Pcom at the peak of the
J Med Eng Technol, 2015; 39(1): 79–85
Table 1. Individual results of PdiffA. A total of 40 values were obtained from five subjects, with eight per subject. Subject number 1
Step interval (mmHg)
Direction of Pcom sequence
Repeated measurement
PdiffA (mmHg)
2 (Test #2)
Pcom "
1st 2nd 1st 2nd 1st 2nd 1st 2nd 1st 2nd 1st 2nd 1st 2nd 1st 2nd 1st 2nd 1st 2nd 1st 2nd 1st 2nd 1st 2nd 1st 2nd 1st 2nd 1st 2nd 1st 2nd 1st 2nd 1st 2nd 1st 2nd
1.9 6.2 6.2 6.6 7.6 8.9 9.9 10.8 4.1 3.4 5.7 3.5 3.7 4.7 4.4 3.1 10.9 7.2 9.8 8.9 2.6 8.9 1.5 2.6 7.9 6.5 6.8 9.3 1.2 0.8 0.6 1.4 3.3 1.2 1.2 3.3 19.8 10.0 16.6 4.7
Pcom # 2
2 (Test #2)
Pcom #
^
PPGacurve and when Pcom is equal to MAPF (equation 1), i.e. at Pdiff ¼ 0.
3
2 (Test #2)
J Med Eng Technol Downloaded from informahealthcare.com by Nyu Medical Center on 05/28/15 For personal use only.
3. Results Good quality PPG recordings with stable Finapres ABP waveforms were obtained in all measurements. In each of the Tests #1 and #2, four sets of recordings were obtained, hence there were eight sets per subject, leading to a total of 40 PdiffA values. The curve fittings show that, above the fourth order, the root mean squared error (RMSE, mean ± SD) did not decrease significantly with higher order polynomials. Therefore, for further data analysis, each of the PPGa arrays was fitted with a fourth order polynomial curve. Table 1 presents all the 40 individual results of PdiffA from the five subjects. The range of PdiffA obtained was 19.8 to +6.6 mmHg. The Kolmogorov-Smirnov test showed that the sample distribution was Gaussian (p40.20). Seven out of the 40 PdiffA values (17.5%) were positive, while the remaining 82.5% were negative. The positive PdiffA are also shown in italics in Table 1. PdiffA was significantly different from 0 mmHg (p50.001). Consequently, the first null hypothesis was rejected. For the second hypothesis, the hierarchical statistical tests for PdiffA showed no statistical significance for the three
Pcom " Pcom #
2.5. Statistical analysis Provided that all measurements were satisfactory, eight PdiffA values would be available per subject. Normality was assessed with the Kolmogorov-Smirnov test and statistical tests were chosen appropriately. For Gaussian data, paired t-tests were used to assess if PdiffA was significantly different from zero. The second hypothesis tested states that the set-up and protocols in Tests #1 and #2 do not lead to significant differences in PdiffA. For this purpose, a hierarchical testing procedure was adopted to minimize the number of multiple comparisons. The hierarchical test began with the ‘repeated measurements’ factor. Measurement results which used the 2 mmHg step interval were tested first (i.e. higher resolution). This was followed by the ‘direction of the Pcom sequence’. Lastly, the influence of the ‘step interval of Pcom’ was tested. If the statistical result in each level showed no significant difference, the corresponding PdiffA values were pooled by averaging per subject before the next level of the hierarchy was tested. A value of p50.05 was taken as the level of statistical significance. Bonferroni correction was applied to account for multiple comparisons [15,16], with the -levels corrected to 0.05, 0.025 and 0.017 for the three factors, respectively.
Pcom "
4
2 (Test #2)
Pcom " Pcom #
5
2 (Test #2)
Pcom " Pcom #
1
10 (Test #1)
Pcom " Pcom #
2
10 (Test #1)
Pcom " Pcom #
3
10 (Test #1)
Pcom " Pcom #
4
10 (Test #1)
Pcom " Pcom #
5
10 (Test #1)
Pcom " Pcom #
Positive PdiffA values are shown in italics.
factors (i.e. ‘repeated measurements’, ‘direction of Pcom sequence’ and ‘step interval of Pcom’). When no statistical significance was detected from the paired t-tests at each level of the hierarchy, the PdiffA values were pooled to obtain an average value for that level. The PdiffA in Table 1 are plotted in Figure 4(a) for each subject. With the exception of subjects 1 and 3, all other subjects only have distributions with PdiffA50. The intrasubject mean (SD) of PdiffA are shown in Figure 4(b). In addition to this, it was found that there was no significant correlation between PdiffA and the subjects’ age. In summary, with the PdiffA results giving a normal distribution and no statistical significance detected for the three factors considered, the sample average (mean (SD)) for PdiffA was computed as 4.7 (5.63) mmHg (p50.001).
4. Discussion The original work performed in this study has important practical applications and ramifications that could lead to
DOI: 10.3109/03091902.2014.979953
Non-invasive blood pressure measurements
83
with measurements performed in the ascending aorta. Although it would be tempting to reduce the bias by subtracting a constant value from the Finapres estimate, this could be a poor engineering solution given the inter-subject differences we have observed in our study (Figure 4). Moreover, given the directional disparity of the MAP bias reported by different studies, it is still uncertain what the mean correction factor would be. What is more consistent in all studies in this area is the amplitude of the SD observed, usually in the range 5–7 mmHg [8,17]. From the corresponding 95% confidence limit, it is easy to understand the variability of individual mean biases.
J Med Eng Technol Downloaded from informahealthcare.com by Nyu Medical Center on 05/28/15 For personal use only.
4.2. Literature review
Figure 4. Intra-subject sample distribution for PdiffA. (a) Eight PdiffA values were obtained per subject. (b) The intra-subject mean of PdiffA were computed. The error bars are the respective ± 1 SD.
improvements in different types of equipment for noninvasive measurement of ABP. 4.1. Main findings This study demonstrated that the external compressing pressure that produced peak oscillation in the PPG did not coincide with the MAP estimated by the Finapres device, which is the most commonly used commercial device for continuous non-invasive measurement of ABP. The subsequent demonstration that the bias (PdiffA) was not influenced by three different factors (‘repeated measurements’, ‘direction of Pcom sequence’ and ‘step interval of Pcom’) suggest that the discrepancy between MAP values yielded by this device and the external compressing pressure producing maximum PPG oscillations could be assumed to also apply to other similar devices. The differences we found between the maximum PPG oscillation and the MAP values given by a widely used commercial device can be subjected to different interpretations. Assuming that the Finapres device was operating satisfactorily, which we have no evidence to the contrary, it is possible that it incorporates a proprietary algorithm which manufacturers believe provides higher accuracy for estimation of MAP than values corresponding to the maximum PPG oscillation. However, a large number of studies comparing the Finapres to intravascular measurements have shown systematic bias of MAP that are in very good agreement with the PdiffA we found of 4.7 (5.6) mmHg. As an example, reviewing a large number of studies, Imholz et al. [8] obtained an estimate of 5.6 (7.2) mmHg for the MAP bias from a large number of studies using laboratory measurements at rest. However, a positive bias was obtained when looking at other sub-groups of studies [8]. For studies performed more recently, using more reliable catheter-tip transducers, Sammons et al. [17] obtained a MAP bias of 3.2 (7.0) mmHg, in comparison
Table 2 lists 12 articles that have addressed the relationship between the pulsation amplitude and the MAP. Different measurement techniques were used to study this relationship including oscillometry, photoplethysmography or arterial volume clamping with PPG. The studies included the use of excised animal arteries (Table 2, numbered 1, 4, 5, 9 and 11), in vivo (numbered 2, 3, 9 and 11) and/or in vitro set-ups (numbered 1, 4, 5 and 9), theoretical modelling and/or experiments (numbered 4, 6, 8 and 10–12) and subjects with different pathologies (numbered 2, 3 and 9). Furthermore, subjects included the human, dog or rat. The main arteries studied were mainly the carotid, brachial, finger or radial; others included the rat’s tail or arteries in the calf and ankle. Unfortunately, the diversity of techniques, subjects and preparations do not allow direct comparisons with our study. Ramsey [18] and Yamakoshi et al. [5] reported that, at the largest pulsation amplitude in oscillometry, the cuff pressure coincided with the intra-arterial MAP. Penaz [4] concluded the same, except that it was based on the volume clamping technique. Conversely, a much larger number of articles (Table 2 numbered 1, 3, 4, 6 and 8–12) reported that the largest pulsation amplitude did not coincide with the MAP. Nevertheless, the reported biases were considered relatively small by these authors. Table 2 contains only a small number of articles based on the volume clamping technique. The large majority of articles were based on oscillometry (Table 2 numbered 1–4, 6, 8 and 10–12) which operates under different principles from arterial volume clamping. In oscillometry the pulse wave is due to the pressure oscillation in the compressing chamber, e.g. the inflatable arm cuff as the ABP is transmitted through the tissue to the cuff. In the volume clamping technique, on the other hand, the pulse wave is due to the amount of illuminated light being scattered or reflected by the pulsating blood volume in the artery to the photodetector, which is then recorded as the PPG. Despite these differences though, MAP would be expected to occur at the point where external compression would induce the largest oscillation in the unloaded vessel. Finally, in the relatively fewer cases where estimates of bias were reported (Table 2), the SD and range are in agreement with the corresponding values we found, although the direction of the bias, as either positive or negative, is fairly variable between studies.
84
K. Y. Chin & R. B. Panerai
J Med Eng Technol, 2015; 39(1): 79–85
J Med Eng Technol Downloaded from informahealthcare.com by Nyu Medical Center on 05/28/15 For personal use only.
Table 2. Relationship between the pulsation amplitude and MAP. This relationship was evaluated by authors in the articles listed below.
#
Article
Measurement technique
1. 2. 3. 4. 5. 6. 7. 8. 9.
Posey et al. [20] Ramsey III [18] Yelderman and Ream [7] Mauck et al. [21] Yamakoshi et al. [5] Forster and Turney [22] Penaz [4] Drzewiecki et al. [23] Wesseling et al. [13]
Oscillometric Oscillometric Oscillometric Oscillometric Photoplethysmography Oscillometric Volume clamping Oscillometric Volume clamping
10. 11. 12. 13.
Ursino and Cristalli [24] Baker et al. [25] Raamat et al. [6] This study
Oscillometric Oscillometric Oscillometric Volume clamping
Theoretical model
3
Set-up
Subjects
Artery
In vitro In vivo In vivo In vitro In vitro
Dog Human Human Dog Dog
Carotid Bicep/calf/ankle Brachial Carotid Carotid Brachial Finger Brachial Finger/tail Finger Brachial Radial Finger Finger
3 Not applicable (review article) 3 In vitro In vivo 3 3 In vivo 3 In vivo
Human Human/rat Human Dog Human
Bias mean (SD) n/a 0.2 (4.21) mmHg 1.4 (6.22) mmHg n/a ± 3.0 mmHg n/a n/a 2.0 mmHg n/a n/a n/a n/a range 8 to 19 mmHg 4.7 (5.63) mmHg
A tick (3) means that a theoretical model was evaluated in the article. The last row refers to the experimental setup in this article. n/a ¼ bias values not reported.
4.3. Limitations of the study Under ideal conditions, more accurate measurements of ABP could have been performed with an intravascular catheter-tip transducer. Unfortunately this procedure was not deemed ethical at this stage of the study. Alternatively, MAP could have been estimated by measuring systolic and diastolic BP with a sphygmomanometer in the brachial artery. However, due to the uncertain relationship between MAP and systolic/ diastolic pressures and the fact that these values would be measured at different times and would be variable with each heartbeat, discouraged us to add this measurement to our protocol. There are several uncertainties in interpreting the experimental results when using the Finapres to measure the MAP. The MAPF in equation (1) was computed from the Finapres ABP waveforms. It is reported in a comparative study between two Finapreses [19] that there are inherent differences even between monitoring devices of the same specification and model number. It is possible that the PdiffA should always be zero (i.e. MAP ¼ Pcom, see equation 1), but the Finapres did not consistently produce accurate MAP values. In addition to the inherent differences between Finapres devices mentioned above, the Finapres does not allow independent callibration of its pressure transducer and PPG sensing mechanism. Moreover, as shown by the differences between Figures 3(b) and 3(c), it is likely that MAP, as is the case of most physiological measurements, was changing over time, thus explaining a changeable relationship between Pcom and Pdiff. For these reasons, the accuracy of the MAP based on the Finapres measurement needs to be interpreted with caution. Our study focused on the digital artery. Results cannot be automatically transferred to other arteries, such as the radial, brachial or temporal, without further investigation. The same uncertainty applies to ABP measurements performed with different device types or techniques other than the Finapres. The experiment was carried out on healthy adults at rest and cannot be extended to other age groups, physiological conditions (e.g. standing or exercise) or to subjects with specific conditions, e.g. atherosclerosis, hypertension, diabetes, obesity or autonomic dysfunction.
The exact causes for the inter- and intra-subject differences in PdiffA are not clear. These could be attributed to local regulation of blood pressure, vasomotor tone (i.e. vasoconstriction or dilation), emotional state, environment (e.g. ambient temperature and lighting) and/or other factors. 4.4. Suggestions for future work The relationship between the external compressing pressure at the largest PPG pulse amplitude and the MAP may be more complex than previously anticipated. Additional parameter(s) may be required to develop a more robust algorithm for determining the PPG reference value. More extensive evaluation of the temporal patterns and the frequency content of the PPG waveform are among the parameters showing potentials to help improve estimation of MAP. Mathematical modelling that takes into account the physical properties of the interaction of light with tissue, as well as the mechanical properties of arteries, may also help to identify the key parameters influencing the relationship between the PPG and MAP.
5. Conclusion This study demonstrated that the external compressing pressure that produces the largest PPG pulse amplitude does not coincide with the MAP (i.e. PdiffA is not zero), but follows a normal distribution. Also, the PdiffA shows both intra- and inter-subject variability and hence cannot be used as a fixed general correction factor. Further evaluation of the PPG waveform is suggested, which includes the temporal patterns and the frequency content of the PPG waveform; this may provide more answers to improve the determination of the PPG reference value in the arterial volume clamping technique.
Acknowledgements This study was approved by the Leicestershire ethics committee (reference number: 4680).
Declaration of interest The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.
DOI: 10.3109/03091902.2014.979953
J Med Eng Technol Downloaded from informahealthcare.com by Nyu Medical Center on 05/28/15 For personal use only.
References 1. Carroll, G.C., 1988, Blood pressure monitoring. Critical Care Clinics, 4, 411–434. 2. Murray, I.P., 1981, Complications of invasive monitoring. Medical Instrumentation, 15, 85–89. 3. Penaz, J., 1973, Photoelectric measurement of blood pressure, volume and flow in the finger. Digest of the 10th International Conference on Medical and Biological Engineering (Dresden), pp. 104. 4. Penaz, J., 1992, Criteria for set point estimation in the volume clamp method of blood pressure measurement. Physiological Research, 41, 5–10. 5. Yamakoshi, K., Shimazu, H., Shibata, M., and Kamiya, A., 1982, New oscillometric method for indirect measurement of systolic and mean arterial pressure in the human finger. Part 1: Model experiment. Medical & Biological Engineering & Computing, 20, 307–313. 6. Raamat, R., Talts, J., Jagomagi, K., and Lansimies, E. 1999, Mathematical modelling of non-invasive oscillometric finger mean blood pressure measurement by maximum oscillation criterion. Medical & Biological Engineering & Computing, 37, 784–788. 7. Yelderman, M., and Ream, A.K., 1979, Indirect measurement of mean blood pressure in the anesthetized patient. Anesthesiology, 50, 253–256. 8. Imholz, B.P., Wieling, W., van Montfrans, G.A., and Wesseling, K.H., 1998, Fifteen years’ experience with finger arterial pressure monitoring: Assessment of the technology. Cardiovascular Research, 38, 605–616. 9. Laurent, C., Jonsson, B., Vegfors, M., and Lindberg, L.G., 2005, Non-invasive measurement of systolic blood pressure on the arm utilising photoplethysmography: development of the methodology. Medical & Biological Engineering & Computing, 43, 131–135. 10. Chin, K.Y., and Panerai, R.B., 2013, A new noninvasive device for continuous arterial blood pressure monitoring in the superficial temporal artery. Physiological Measurement, 34, 407–421. 11. Langewouters, G.J., Settels, J.J., Roelandt, R., and Wesseling, K.H., 1998, Why use Finapres or Portapres rather than intra-arterial or intermittent non-invasive techniques of blood pressure measurement? Journal of Medical Engineering & Technology, 22, 37–43. 12. Tanaka, S., and Yamakoshi, K., 1996, Ambulatory instrument for monitoring indirect beat-to-beat blood pressure in superficial temporal artery using volume-compensation method. Medical & Biological Engineering & Computing, 34, 441–447.
Non-invasive blood pressure measurements
85
13. Wesseling, K.H., de Wit, B., van der Hoeven, G.M.A., van Goudoever, J., Settels, J.J., 1995, Physiocal, calibrating finger vascular physiology for Finapres. Homeostasis, 36, 67–82. 14. Panerai, R.B., Sammons, E.L., Smith, S.M., Rathbone, W.E., Bentley, S., Potter, J.F., and Samani, N.J., 2007, Transient drifts between Finapres and continuous intra-aortic measurements of blood pressure. Blood Pressure Monitor, 12, 369–376. 15. Bland, M., 1995, Significance tests. An introduction to medical statistics, 2nd edn. (New York: Oxford University Press). pp. 133–151. 16. Campbell, M.J., and Machin, D., 1995, Medical Statistics: A common sense approach, 2nd edn. (Chichester, West Sussex: John Wiley & Sons). 17. Sammons, E.L., Samani, N.J., Smith, S.M., Rathbone, W.E., Bentley, S., Potter, J.F., and Panerai, R.B., 2007, Influence of noninvasive peripheral arterial blood pressure measurements on assessment of dynamic cerebral autoregulation. Journal of Applied Physiology, 103, 369–375. 18. Ramsey III, M., 1979, Noninvasive automatic determination of mean arterial pressure. Medical & Biological Engineering & Computing, 17, 11–18. 19. Chin, K.Y., and Panerai, R.B., 2012, Comparative study of Finapres devices. Blood Pressure Monitoring, 17, 171–178. 20. Posey, J.A., Geddes, L.A., Williams, H., and Moore, A.G., 1969, The meaning of the point of maximum oscillations in cuff pressure in the indirect measurement of blood pressure: Part 1. Cardiovascular Research Center Bulletin, 8, 15–25. 21. Mauck, G.W., Smith, C.R., Geddes, L.A., and Bourland, J.D., 1980, The meaning of the point of maximum oscillations in cuff pressure in the indirect measurement of blood pressure–part ii. Journal of Biomechanical Engineering, 102, 28–33. 22. Forster, F.K., and Turney, D., 1986, Oscillometric determination of diastolic, mean and systolic blood pressure–a numerical model. Journal of Biomechanical Engineering, 108, 359–364. 23. Drzewiecki, G., Hood, R., and Apple, H., 1994, Theory of the oscillometric maximum and the systolic and diastolic detection ratios. Annals of Biomedical Engineering, 22, 88–96. 24. Ursino, M., and Cristalli, C., 1996, A mathematical study of some biomechanical factors affecting the oscillometric blood pressure measurement. IEEE Transactions on Biomedical Engineering, 43, 761–778. 25. Baker, P.D., Westenskow, D.R., and Kuck, K., 1997, Theoretical analysis of non-invasive oscillometric maximum amplitude algorithm for estimating mean blood pressure. Medical & Biological Engineering & Computing, 35, 271–278.
REVIEW
Relating external compressing pressure to mean arterial pressure in non-invasive blood pressure measurements K. Y. Chin*1 and R. B. Panerai2,3
J Med Eng Technol Downloaded from informahealthcare.com by Nyu Medical Center on 05/28/15 For personal use only.
1
Department of Medical Physics, University Hospitals of Leicester NHS Trust, Leicester Royal Infirmary, Leicester LE1 5WW, UK, 2Department of Cardiovascular Sciences, University of Leicester, Leicester Royal Infirmary, Leicester LE1 5WW, UK, and 3Leicester NIHR Biomedical Research Unit in Cardiovascular Sciences, Glenfield Hospital, Leicester LE3 9QP, UK Abstract
Keywords
Arterial volume clamping uses external compression of an artery to provide continuous noninvasive measurement of arterial blood pressure. It has been assumed that mean arterial pressure (MAP) corresponds to the point where unloading leads to the maximum oscillation of the arterial wall as reflected by photoplethysmogram (PPG), an assumption that has been challenged. Five subjects were recruited for the study (three males, mean age (SD) ¼ 32 (15) years). The PPG waveform was analysed to identify the relationship between the external compressing pressure, PPG pulse amplitude and MAP. Two separate tests were carried out at compression step intervals of 10 mmHg and 2 mmHg, respectively. No significant differences were found between the two tests. The bias between the compressing pressure and the MAP was 4.7 ± 5.63 mmHg (p50.001) showing a normal distribution. Further research is needed to identify optimal algorithms for estimation of MAP using PPG associated with arterial compression.
Blood pressure, non-invasive, photoplethysmography, pulse amplitude, volume clamping technique
1. Introduction Continuous recording of arterial blood pressure (ABP) has been increasingly used for patient monitoring and cardiovascular research. Although intravascular recordings are still the gold standard technique for this purpose, its relatively high cost, associated to the risk of infection and other complications, has stimulated the search for alternative non-invasive methods [1,2]. Most commercial equipment for this purpose have been dominated by the use of arterial volume clamping, which provides continuous non-invasive estimates of ABP using photoplethysmography to detect changes in arterial volume, coupled with feedback controlled external compression of a peripheral artery, usually the digital artery in the middle or ring finger [3]. One central assumption of the arterial volume clamping approach though is the estimation of the mean arterial pressure (MAP). Most devices assume that MAP corresponds to the external compressing pressure when the photoplethysmogram (PPG) signal reaches its maximum oscillation [4,5]. However, this view has been challenged by several investigators and it remains controversial [6,7]. Of considerable importance, a similar assumption is adopted by other techniques or devices used for non-invasive measurement of ABP such as oscillometry, which can only provide intermittent measurements. In both cases though, accurate detection *Corresponding author. Email: [email protected]
History Received 13 August 2014 Revised 2 October 2014 Accepted 19 October 2014
of MAP is highly relevant as it determines the overall accuracy of the ABP values output by the device. In this study, we used arterial volume clamping to investigate the relationship between MAP and the external compressing pressure of the digital artery, testing two specific hypotheses: (i) that the MAP corresponds to the external compressing pressure leading to the maximum amplitude of the PPG signal; and (ii) that the relationship between MAP and external compression is independent of the compressing pressure step changes.
2. Methods 2.1. Volume clamping technique Measurement devices using the volume clamping technique have been reported in the literature for the finger [8], brachial [9] and superficial temporal [10] arteries. However, nearly all recordings in clinical and research protocols had been obtained from the finger artery due to the commercial availability of the FinapresÕ (Ohmeda Medical; Englewood, Colorado, USA) and its later models PortapresÕ and FinometerÕ (Finapres Medical Systems; Amsterdam, Netherlands) [11]. Other commercial devices currently available include CNAP500Õ (CNSystems; Graz, Austria) and NexfinÕ (BMEYE; Amsterdam, Netherlands). These devices also monitor the ABP in the finger. In the literature, the volume clamping technique is also known as the vascular unloading [3] or volume compensation technique [12].
J Med Eng Technol Downloaded from informahealthcare.com by Nyu Medical Center on 05/28/15 For personal use only.
80
K. Y. Chin & R. B. Panerai
The volume clamping technique incorporates the use of photoplethysmography to detect changes in arterial blood volume. Photoplethysmography uses an infrared emitting light source to illuminate the tissue within which the artery of interest is located. Depending on the measurement site, a photodetector is placed in either reflectance or transmittance configuration to measure the light transmitted through, scattered and/or reflected by the pulsating arterial blood. The recording of this photo-signal produces the PPG. When a range of external compressing pressures is applied to the artery, the PPG waveshape shows changes in its amplitude as well as in its temporal pattern. An example is given in Figures 1(a) and (b) where the external compressing pressure range encompasses the diastolic and systolic pressures. As the external compressing pressure rises, it causes a gradual reduction of arterial blood and a shift of extra-cellular fluid in the tissue. This leads to a gradual increase in the PPG baseline. An oscillation component due to the pulsating arterial blood superimposes on the PPG baseline. The amplitude of the oscillation component increases to a maximum and then decreases until it disappears at higher external compressing pressures.
J Med Eng Technol, 2015; 39(1): 79–85
A reference value is derived from the PPG pulse that could produce zero transmural pressure. At zero transmural pressure, the external compressing pressure is equal to the intraarterial pressure. The reference value, for example, could be from the mean value of the PPG with the largest pulse amplitude. Once a PPG reference value is available, a servofeedback control mechanism is incorporated to continuously modulate the external pressure to partially compress (hence ‘volume clamping’) the pulsating artery to maintain the real time PPG at this reference value. The external compressing pressure is recorded continuously, thus representing a measurement of the ABP. 2.2. Subjects and measurements Five subjects were recruited (three males, mean age (SD) of 32 (15) years). Subjects had no known history of cardiovascular or other pathologies that would affect the blood circulation in the fingers. Participants were all staff from the Medical Physics department. Consent was given by each participant prior to the arrangement for the ABP measurements. The study was approved by the local ethics committee. Finapres finger cuffs of appropriate size were fitted on the middle and ring fingers of the right hand. The fitting and positioning of the finger cuffs followed the manufacturer’s guidelines. The finger cuff on the ring finger was used to measure the real-time continuous ABP waveform with the Finapres (‘Ohmeda 2300’ model). The finger cuff on the middle finger was operated by an in-house built PPG recording device. Two ipsilateral instead of contralateral fingers were used in order to minimize the potential difference in regional regulation of arterial blood flow between the two hands. Figure 2 shows the block diagram of the in-house built PPG recording device. A sequence of regulated compressing pressure was applied as command output from a desktop computer to compress the artery in the middle finger. Simultaneously, the optoelectronic components in the finger cuff monitored the blood volume change in the artery. The output from the photodetector was converted to a PPG signal through transimpedance amplification. The PPG was recorded and displayed in real time by the desktop computer. A separate channel was used to record the continuous ABP waveform measured by the Finapres. Both the Finapres and the PPG waveforms were simultaneously sampled at 100 Hz (DT302, Data TranslationÕ ). The recording was managed with the Agilent VEEProÕ visual programming software. 2.3. Protocols
Figure 1. (a) External compressing pressure recording. (b) Simultaneous recording of PPG waveform. The PPG pulse with the largest pulsation amplitude is indicated as ‘1’. The corresponding compressing pressure value is also shown (indicated as ‘2’). (c) An example PPG waveform shows changes in its amplitude as well as its temporal pattern. ‘s’ refers to the lowest PPG value during the systolic phase, whereas ‘d’ refers to the highest PPG value during the diastolic phase.
Each subject attended two measurement sessions (Test #1 and Test #2). During each session, the right hand rested on a bench at heart level. Subjects were asked to relax their fingers and breathe normally while sitting on a chair with back support. Subjects were also asked not to engage in any conversation, laughter, holding their breath or shifting about in their seated position. In Test #1, the compressing pressure in the cuff (denoted as Pcom) was first increased by 10 mmHg every 5 s. The range of Pcom encompassed the physiological range of systolic and diastolic pressures. When Pcom had increased above systolic
DOI: 10.3109/03091902.2014.979953
Non-invasive blood pressure measurements
81
J Med Eng Technol Downloaded from informahealthcare.com by Nyu Medical Center on 05/28/15 For personal use only.
Figure 2. Block diagram of the in-house built PPG recording device with Finapres cuff. The device is developed to monitor the arterial blood volume change in the middle finger to generate a PPG signal. The PPG waveform can be recorded, stored and displayed.
pressure, the recording was complete. It was then decreased at the same rate until it was near 0 mmHg. This corresponded to a separate recording. The cycle of increasing and decreasing the Pcom was then repeated, for a total of four sets of recordings. Throughout this process, the real time PPG waveform was recorded simultaneously. In Test #2, the same set-up and protocol were repeated except that the Pcom steps were now changed at 2 mmHg every 5 s. Again, four sets of recordings were obtained in each subject. The rationale for this was to compare results between different pressure step intervals. The Finapres has a built-in physiological self-calibration facility known as the ‘Physiocal’ [13]. It was kept switched on throughout the ABP measurement. This is to prevent any potential drifts in the MAP measured with the Finapres [14]. 2.4. Data analysis When the ABP increases during the systolic phase, the PPG decreases. This forms an inverse relationship between the arterial blood volume light absorption and the PPG. Figure 1(c) shows an example that, at the peak of the systolic phase, the PPG decreases to a local minimum, indicated as ‘s’. Conversely, as the ABP decreases to the diastolic pressure, the PPG increases to a local maximum, indicated as ‘d’ in the figure. The PPG pulse amplitude is measured between each ‘s’ and the preceding ‘d’ and denoted as PPGa (in Volts). Depending on the heart rate, three-to-five pulses were recorded at each Pcom. The PPGa from each of these pulses during the same Pcom were then averaged (PPGa (SD)). Corresponding ABP pulses were recorded with the Finapres and the average (SD) of systolic, mean and diastolic pressures were also computed (i.e. SBPF , MAPF and DBPF , respectively) (Figure 3a). Each MAP was calculated as the timeintegrated average of each ABP pulse waveform. The array of PPGa was plotted against the sequence of Pcom. Figure 3(b) shows an example.
^
Figure 3. Example computation of PdiffA from the PPGacurve. (a) The Pcom (light square) was decreased at 2 mmHg steps every 5 s. The Finapres ABP is shown as: SBPF (diamond), MAPF (square) and DBPF (triangle). (b) The array of PPGacorresponding to decreasing Pcom. (c) Graph in (b) was re-plotted against Pdiff (equation 1). The PPGa array is ^
superimposed by a suitable polynomial order curve (PPGa). The PdiffA is indicated. In this example, PdiffA ¼ 6.2 mmHg. In all the graphs, the error bars are the respective ± 1 SD.
Next, the difference between MAPF and the external compressing pressure (Pcom) which produces the largest PPG pulse amplitude was computed. The generic pressure difference (Pdiff) equation was defined as Pdiff ¼ MAPF Pcom
ð1Þ
82
K. Y. Chin & R. B. Panerai
Pdiff was calculated for each Pcom. Following Figure 3(b), the PPGa array was re-plotted against the Pdiff (Figure 3c). First to sixth order polynomials were tested to fit the experiment data for the pressure range of ±40 mmHg around Pdiff ¼ 0. The optimal fitted curve was denoted ^
PPGa, a representative ^example is depicted in Figure 3(c). At the peak of the PPGacurve, the corresponding Pdiff was computed. This specific Pdiff was denoted as PdiffA (mmHg). PdiffA indicates the bias between the Pcom at the peak of the
J Med Eng Technol, 2015; 39(1): 79–85
Table 1. Individual results of PdiffA. A total of 40 values were obtained from five subjects, with eight per subject. Subject number 1
Step interval (mmHg)
Direction of Pcom sequence
Repeated measurement
PdiffA (mmHg)
2 (Test #2)
Pcom "
1st 2nd 1st 2nd 1st 2nd 1st 2nd 1st 2nd 1st 2nd 1st 2nd 1st 2nd 1st 2nd 1st 2nd 1st 2nd 1st 2nd 1st 2nd 1st 2nd 1st 2nd 1st 2nd 1st 2nd 1st 2nd 1st 2nd 1st 2nd
1.9 6.2 6.2 6.6 7.6 8.9 9.9 10.8 4.1 3.4 5.7 3.5 3.7 4.7 4.4 3.1 10.9 7.2 9.8 8.9 2.6 8.9 1.5 2.6 7.9 6.5 6.8 9.3 1.2 0.8 0.6 1.4 3.3 1.2 1.2 3.3 19.8 10.0 16.6 4.7
Pcom # 2
2 (Test #2)
Pcom #
^
PPGacurve and when Pcom is equal to MAPF (equation 1), i.e. at Pdiff ¼ 0.
3
2 (Test #2)
J Med Eng Technol Downloaded from informahealthcare.com by Nyu Medical Center on 05/28/15 For personal use only.
3. Results Good quality PPG recordings with stable Finapres ABP waveforms were obtained in all measurements. In each of the Tests #1 and #2, four sets of recordings were obtained, hence there were eight sets per subject, leading to a total of 40 PdiffA values. The curve fittings show that, above the fourth order, the root mean squared error (RMSE, mean ± SD) did not decrease significantly with higher order polynomials. Therefore, for further data analysis, each of the PPGa arrays was fitted with a fourth order polynomial curve. Table 1 presents all the 40 individual results of PdiffA from the five subjects. The range of PdiffA obtained was 19.8 to +6.6 mmHg. The Kolmogorov-Smirnov test showed that the sample distribution was Gaussian (p40.20). Seven out of the 40 PdiffA values (17.5%) were positive, while the remaining 82.5% were negative. The positive PdiffA are also shown in italics in Table 1. PdiffA was significantly different from 0 mmHg (p50.001). Consequently, the first null hypothesis was rejected. For the second hypothesis, the hierarchical statistical tests for PdiffA showed no statistical significance for the three
Pcom " Pcom #
2.5. Statistical analysis Provided that all measurements were satisfactory, eight PdiffA values would be available per subject. Normality was assessed with the Kolmogorov-Smirnov test and statistical tests were chosen appropriately. For Gaussian data, paired t-tests were used to assess if PdiffA was significantly different from zero. The second hypothesis tested states that the set-up and protocols in Tests #1 and #2 do not lead to significant differences in PdiffA. For this purpose, a hierarchical testing procedure was adopted to minimize the number of multiple comparisons. The hierarchical test began with the ‘repeated measurements’ factor. Measurement results which used the 2 mmHg step interval were tested first (i.e. higher resolution). This was followed by the ‘direction of the Pcom sequence’. Lastly, the influence of the ‘step interval of Pcom’ was tested. If the statistical result in each level showed no significant difference, the corresponding PdiffA values were pooled by averaging per subject before the next level of the hierarchy was tested. A value of p50.05 was taken as the level of statistical significance. Bonferroni correction was applied to account for multiple comparisons [15,16], with the -levels corrected to 0.05, 0.025 and 0.017 for the three factors, respectively.
Pcom "
4
2 (Test #2)
Pcom " Pcom #
5
2 (Test #2)
Pcom " Pcom #
1
10 (Test #1)
Pcom " Pcom #
2
10 (Test #1)
Pcom " Pcom #
3
10 (Test #1)
Pcom " Pcom #
4
10 (Test #1)
Pcom " Pcom #
5
10 (Test #1)
Pcom " Pcom #
Positive PdiffA values are shown in italics.
factors (i.e. ‘repeated measurements’, ‘direction of Pcom sequence’ and ‘step interval of Pcom’). When no statistical significance was detected from the paired t-tests at each level of the hierarchy, the PdiffA values were pooled to obtain an average value for that level. The PdiffA in Table 1 are plotted in Figure 4(a) for each subject. With the exception of subjects 1 and 3, all other subjects only have distributions with PdiffA50. The intrasubject mean (SD) of PdiffA are shown in Figure 4(b). In addition to this, it was found that there was no significant correlation between PdiffA and the subjects’ age. In summary, with the PdiffA results giving a normal distribution and no statistical significance detected for the three factors considered, the sample average (mean (SD)) for PdiffA was computed as 4.7 (5.63) mmHg (p50.001).
4. Discussion The original work performed in this study has important practical applications and ramifications that could lead to
DOI: 10.3109/03091902.2014.979953
Non-invasive blood pressure measurements
83
with measurements performed in the ascending aorta. Although it would be tempting to reduce the bias by subtracting a constant value from the Finapres estimate, this could be a poor engineering solution given the inter-subject differences we have observed in our study (Figure 4). Moreover, given the directional disparity of the MAP bias reported by different studies, it is still uncertain what the mean correction factor would be. What is more consistent in all studies in this area is the amplitude of the SD observed, usually in the range 5–7 mmHg [8,17]. From the corresponding 95% confidence limit, it is easy to understand the variability of individual mean biases.
J Med Eng Technol Downloaded from informahealthcare.com by Nyu Medical Center on 05/28/15 For personal use only.
4.2. Literature review
Figure 4. Intra-subject sample distribution for PdiffA. (a) Eight PdiffA values were obtained per subject. (b) The intra-subject mean of PdiffA were computed. The error bars are the respective ± 1 SD.
improvements in different types of equipment for noninvasive measurement of ABP. 4.1. Main findings This study demonstrated that the external compressing pressure that produced peak oscillation in the PPG did not coincide with the MAP estimated by the Finapres device, which is the most commonly used commercial device for continuous non-invasive measurement of ABP. The subsequent demonstration that the bias (PdiffA) was not influenced by three different factors (‘repeated measurements’, ‘direction of Pcom sequence’ and ‘step interval of Pcom’) suggest that the discrepancy between MAP values yielded by this device and the external compressing pressure producing maximum PPG oscillations could be assumed to also apply to other similar devices. The differences we found between the maximum PPG oscillation and the MAP values given by a widely used commercial device can be subjected to different interpretations. Assuming that the Finapres device was operating satisfactorily, which we have no evidence to the contrary, it is possible that it incorporates a proprietary algorithm which manufacturers believe provides higher accuracy for estimation of MAP than values corresponding to the maximum PPG oscillation. However, a large number of studies comparing the Finapres to intravascular measurements have shown systematic bias of MAP that are in very good agreement with the PdiffA we found of 4.7 (5.6) mmHg. As an example, reviewing a large number of studies, Imholz et al. [8] obtained an estimate of 5.6 (7.2) mmHg for the MAP bias from a large number of studies using laboratory measurements at rest. However, a positive bias was obtained when looking at other sub-groups of studies [8]. For studies performed more recently, using more reliable catheter-tip transducers, Sammons et al. [17] obtained a MAP bias of 3.2 (7.0) mmHg, in comparison
Table 2 lists 12 articles that have addressed the relationship between the pulsation amplitude and the MAP. Different measurement techniques were used to study this relationship including oscillometry, photoplethysmography or arterial volume clamping with PPG. The studies included the use of excised animal arteries (Table 2, numbered 1, 4, 5, 9 and 11), in vivo (numbered 2, 3, 9 and 11) and/or in vitro set-ups (numbered 1, 4, 5 and 9), theoretical modelling and/or experiments (numbered 4, 6, 8 and 10–12) and subjects with different pathologies (numbered 2, 3 and 9). Furthermore, subjects included the human, dog or rat. The main arteries studied were mainly the carotid, brachial, finger or radial; others included the rat’s tail or arteries in the calf and ankle. Unfortunately, the diversity of techniques, subjects and preparations do not allow direct comparisons with our study. Ramsey [18] and Yamakoshi et al. [5] reported that, at the largest pulsation amplitude in oscillometry, the cuff pressure coincided with the intra-arterial MAP. Penaz [4] concluded the same, except that it was based on the volume clamping technique. Conversely, a much larger number of articles (Table 2 numbered 1, 3, 4, 6 and 8–12) reported that the largest pulsation amplitude did not coincide with the MAP. Nevertheless, the reported biases were considered relatively small by these authors. Table 2 contains only a small number of articles based on the volume clamping technique. The large majority of articles were based on oscillometry (Table 2 numbered 1–4, 6, 8 and 10–12) which operates under different principles from arterial volume clamping. In oscillometry the pulse wave is due to the pressure oscillation in the compressing chamber, e.g. the inflatable arm cuff as the ABP is transmitted through the tissue to the cuff. In the volume clamping technique, on the other hand, the pulse wave is due to the amount of illuminated light being scattered or reflected by the pulsating blood volume in the artery to the photodetector, which is then recorded as the PPG. Despite these differences though, MAP would be expected to occur at the point where external compression would induce the largest oscillation in the unloaded vessel. Finally, in the relatively fewer cases where estimates of bias were reported (Table 2), the SD and range are in agreement with the corresponding values we found, although the direction of the bias, as either positive or negative, is fairly variable between studies.
84
K. Y. Chin & R. B. Panerai
J Med Eng Technol, 2015; 39(1): 79–85
J Med Eng Technol Downloaded from informahealthcare.com by Nyu Medical Center on 05/28/15 For personal use only.
Table 2. Relationship between the pulsation amplitude and MAP. This relationship was evaluated by authors in the articles listed below.
#
Article
Measurement technique
1. 2. 3. 4. 5. 6. 7. 8. 9.
Posey et al. [20] Ramsey III [18] Yelderman and Ream [7] Mauck et al. [21] Yamakoshi et al. [5] Forster and Turney [22] Penaz [4] Drzewiecki et al. [23] Wesseling et al. [13]
Oscillometric Oscillometric Oscillometric Oscillometric Photoplethysmography Oscillometric Volume clamping Oscillometric Volume clamping
10. 11. 12. 13.
Ursino and Cristalli [24] Baker et al. [25] Raamat et al. [6] This study
Oscillometric Oscillometric Oscillometric Volume clamping
Theoretical model
3
Set-up
Subjects
Artery
In vitro In vivo In vivo In vitro In vitro
Dog Human Human Dog Dog
Carotid Bicep/calf/ankle Brachial Carotid Carotid Brachial Finger Brachial Finger/tail Finger Brachial Radial Finger Finger
3 Not applicable (review article) 3 In vitro In vivo 3 3 In vivo 3 In vivo
Human Human/rat Human Dog Human
Bias mean (SD) n/a 0.2 (4.21) mmHg 1.4 (6.22) mmHg n/a ± 3.0 mmHg n/a n/a 2.0 mmHg n/a n/a n/a n/a range 8 to 19 mmHg 4.7 (5.63) mmHg
A tick (3) means that a theoretical model was evaluated in the article. The last row refers to the experimental setup in this article. n/a ¼ bias values not reported.
4.3. Limitations of the study Under ideal conditions, more accurate measurements of ABP could have been performed with an intravascular catheter-tip transducer. Unfortunately this procedure was not deemed ethical at this stage of the study. Alternatively, MAP could have been estimated by measuring systolic and diastolic BP with a sphygmomanometer in the brachial artery. However, due to the uncertain relationship between MAP and systolic/ diastolic pressures and the fact that these values would be measured at different times and would be variable with each heartbeat, discouraged us to add this measurement to our protocol. There are several uncertainties in interpreting the experimental results when using the Finapres to measure the MAP. The MAPF in equation (1) was computed from the Finapres ABP waveforms. It is reported in a comparative study between two Finapreses [19] that there are inherent differences even between monitoring devices of the same specification and model number. It is possible that the PdiffA should always be zero (i.e. MAP ¼ Pcom, see equation 1), but the Finapres did not consistently produce accurate MAP values. In addition to the inherent differences between Finapres devices mentioned above, the Finapres does not allow independent callibration of its pressure transducer and PPG sensing mechanism. Moreover, as shown by the differences between Figures 3(b) and 3(c), it is likely that MAP, as is the case of most physiological measurements, was changing over time, thus explaining a changeable relationship between Pcom and Pdiff. For these reasons, the accuracy of the MAP based on the Finapres measurement needs to be interpreted with caution. Our study focused on the digital artery. Results cannot be automatically transferred to other arteries, such as the radial, brachial or temporal, without further investigation. The same uncertainty applies to ABP measurements performed with different device types or techniques other than the Finapres. The experiment was carried out on healthy adults at rest and cannot be extended to other age groups, physiological conditions (e.g. standing or exercise) or to subjects with specific conditions, e.g. atherosclerosis, hypertension, diabetes, obesity or autonomic dysfunction.
The exact causes for the inter- and intra-subject differences in PdiffA are not clear. These could be attributed to local regulation of blood pressure, vasomotor tone (i.e. vasoconstriction or dilation), emotional state, environment (e.g. ambient temperature and lighting) and/or other factors. 4.4. Suggestions for future work The relationship between the external compressing pressure at the largest PPG pulse amplitude and the MAP may be more complex than previously anticipated. Additional parameter(s) may be required to develop a more robust algorithm for determining the PPG reference value. More extensive evaluation of the temporal patterns and the frequency content of the PPG waveform are among the parameters showing potentials to help improve estimation of MAP. Mathematical modelling that takes into account the physical properties of the interaction of light with tissue, as well as the mechanical properties of arteries, may also help to identify the key parameters influencing the relationship between the PPG and MAP.
5. Conclusion This study demonstrated that the external compressing pressure that produces the largest PPG pulse amplitude does not coincide with the MAP (i.e. PdiffA is not zero), but follows a normal distribution. Also, the PdiffA shows both intra- and inter-subject variability and hence cannot be used as a fixed general correction factor. Further evaluation of the PPG waveform is suggested, which includes the temporal patterns and the frequency content of the PPG waveform; this may provide more answers to improve the determination of the PPG reference value in the arterial volume clamping technique.
Acknowledgements This study was approved by the Leicestershire ethics committee (reference number: 4680).
Declaration of interest The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.
DOI: 10.3109/03091902.2014.979953
J Med Eng Technol Downloaded from informahealthcare.com by Nyu Medical Center on 05/28/15 For personal use only.
References 1. Carroll, G.C., 1988, Blood pressure monitoring. Critical Care Clinics, 4, 411–434. 2. Murray, I.P., 1981, Complications of invasive monitoring. Medical Instrumentation, 15, 85–89. 3. Penaz, J., 1973, Photoelectric measurement of blood pressure, volume and flow in the finger. Digest of the 10th International Conference on Medical and Biological Engineering (Dresden), pp. 104. 4. Penaz, J., 1992, Criteria for set point estimation in the volume clamp method of blood pressure measurement. Physiological Research, 41, 5–10. 5. Yamakoshi, K., Shimazu, H., Shibata, M., and Kamiya, A., 1982, New oscillometric method for indirect measurement of systolic and mean arterial pressure in the human finger. Part 1: Model experiment. Medical & Biological Engineering & Computing, 20, 307–313. 6. Raamat, R., Talts, J., Jagomagi, K., and Lansimies, E. 1999, Mathematical modelling of non-invasive oscillometric finger mean blood pressure measurement by maximum oscillation criterion. Medical & Biological Engineering & Computing, 37, 784–788. 7. Yelderman, M., and Ream, A.K., 1979, Indirect measurement of mean blood pressure in the anesthetized patient. Anesthesiology, 50, 253–256. 8. Imholz, B.P., Wieling, W., van Montfrans, G.A., and Wesseling, K.H., 1998, Fifteen years’ experience with finger arterial pressure monitoring: Assessment of the technology. Cardiovascular Research, 38, 605–616. 9. Laurent, C., Jonsson, B., Vegfors, M., and Lindberg, L.G., 2005, Non-invasive measurement of systolic blood pressure on the arm utilising photoplethysmography: development of the methodology. Medical & Biological Engineering & Computing, 43, 131–135. 10. Chin, K.Y., and Panerai, R.B., 2013, A new noninvasive device for continuous arterial blood pressure monitoring in the superficial temporal artery. Physiological Measurement, 34, 407–421. 11. Langewouters, G.J., Settels, J.J., Roelandt, R., and Wesseling, K.H., 1998, Why use Finapres or Portapres rather than intra-arterial or intermittent non-invasive techniques of blood pressure measurement? Journal of Medical Engineering & Technology, 22, 37–43. 12. Tanaka, S., and Yamakoshi, K., 1996, Ambulatory instrument for monitoring indirect beat-to-beat blood pressure in superficial temporal artery using volume-compensation method. Medical & Biological Engineering & Computing, 34, 441–447.
Non-invasive blood pressure measurements
85
13. Wesseling, K.H., de Wit, B., van der Hoeven, G.M.A., van Goudoever, J., Settels, J.J., 1995, Physiocal, calibrating finger vascular physiology for Finapres. Homeostasis, 36, 67–82. 14. Panerai, R.B., Sammons, E.L., Smith, S.M., Rathbone, W.E., Bentley, S., Potter, J.F., and Samani, N.J., 2007, Transient drifts between Finapres and continuous intra-aortic measurements of blood pressure. Blood Pressure Monitor, 12, 369–376. 15. Bland, M., 1995, Significance tests. An introduction to medical statistics, 2nd edn. (New York: Oxford University Press). pp. 133–151. 16. Campbell, M.J., and Machin, D., 1995, Medical Statistics: A common sense approach, 2nd edn. (Chichester, West Sussex: John Wiley & Sons). 17. Sammons, E.L., Samani, N.J., Smith, S.M., Rathbone, W.E., Bentley, S., Potter, J.F., and Panerai, R.B., 2007, Influence of noninvasive peripheral arterial blood pressure measurements on assessment of dynamic cerebral autoregulation. Journal of Applied Physiology, 103, 369–375. 18. Ramsey III, M., 1979, Noninvasive automatic determination of mean arterial pressure. Medical & Biological Engineering & Computing, 17, 11–18. 19. Chin, K.Y., and Panerai, R.B., 2012, Comparative study of Finapres devices. Blood Pressure Monitoring, 17, 171–178. 20. Posey, J.A., Geddes, L.A., Williams, H., and Moore, A.G., 1969, The meaning of the point of maximum oscillations in cuff pressure in the indirect measurement of blood pressure: Part 1. Cardiovascular Research Center Bulletin, 8, 15–25. 21. Mauck, G.W., Smith, C.R., Geddes, L.A., and Bourland, J.D., 1980, The meaning of the point of maximum oscillations in cuff pressure in the indirect measurement of blood pressure–part ii. Journal of Biomechanical Engineering, 102, 28–33. 22. Forster, F.K., and Turney, D., 1986, Oscillometric determination of diastolic, mean and systolic blood pressure–a numerical model. Journal of Biomechanical Engineering, 108, 359–364. 23. Drzewiecki, G., Hood, R., and Apple, H., 1994, Theory of the oscillometric maximum and the systolic and diastolic detection ratios. Annals of Biomedical Engineering, 22, 88–96. 24. Ursino, M., and Cristalli, C., 1996, A mathematical study of some biomechanical factors affecting the oscillometric blood pressure measurement. IEEE Transactions on Biomedical Engineering, 43, 761–778. 25. Baker, P.D., Westenskow, D.R., and Kuck, K., 1997, Theoretical analysis of non-invasive oscillometric maximum amplitude algorithm for estimating mean blood pressure. Medical & Biological Engineering & Computing, 35, 271–278.
