`Elsevier
`
`257
`
`CPB 00983
`
`Section II. Systems and programs
`
`Skin photoplethysmography — a review
`
`A.A.R. Kama] 1"“, J.B. Harness 2, G. Irving 3 and AI. Mearns 1'4
`
`‘1 Postgraduate School of Chemical Engineering, 3 Postgraduate School of Pharmacology,
`I Postgraduate school of Control Engineering,
`University of Bradford, Bradford, UK, and 4 Bradford Royal Infirmary, Bradford, UK.
`
`The photoplethysmograph has been used for over 50 years but there are still misconceptions in how and what is the
`information obtained. A photoplethysmograph signal from any site on the skin can be separated into an oscillating
`(a.c.) and a steady-state (d.c.) component, their amplitudes dependent upon the stnicture and flow in the vascular bed.
`Many simple applications are available: pulse counters, using the a.c. component, skin colour and haemoglobin
`saturation meters, using the dc. component. The dc. component of the photoplethysmograph signal is a function of the
`blood flux beneath the device. A good emitter for use in a photoplethysmograph of skin blood flow is one in the
`frequency range 600—700 nm and the best signal for ac. analysis is obtained from the finger pulp. The frequency range
`of the electronic circuitry should be from 0.01 to 15 Hz, then all the information in the signal can be extracted about
`the autonomic nervous system control of the cardiovascular system, particularly between 0.01 and 2 Hz. Comparative
`studies may be drawn between similar skin sites on a subject or between subjects if the afferent inputs to the brain stem
`are controlled or driven at a known frequency. These afferents, inputs, will modulate the efferents, outputs, which
`generate variations in the a.c. component of the detected photoplethysmograph signal.
`
`Photoplethysmography; Skin; Blood flow; Signal analysis
`
`reliable, reproducible,
`should be safe, sensitive,
`simple to use and inexpensive.
`In skin thermometry the surface temperature is
`used as an indicator of changes in skin blood flow.
`t is predominantly sensitive to the component of
`
`/5,m,u,,, ”mum
`“MW/tens
`
`<ggzz’m’gegops
`
`_ _\Dermm plexus
`
`
`
`Fibrous flssue
`Subdermal News
`
`
`
`
`
`1. Measurement of skin blood flow
`
`To date, a precise quantitative measurement of
`skin blood flow has proved to be impossible. The
`skin has a complex structure (see Fig. 1), and
`many factors affect the skin blood flow. Methods
`used to date for skin capillary blood flow mea-
`surement include skin thermometry [35], thermal
`clearance [36],
`laser Doppler plethysmography
`[34,58], radioactive isotope clearance [55,56], elec-
`trical
`impedance methods [18,53] and photo-
`plethysmography [25,42], each having their own
`advantages and disadvantages. When choosing a
`method to measure skin blood flow it should be
`remembered that the ideal non-invasive technique
`
`
`Correspondence: Dr. J.B. Harness, Department of Chemical
`Engineering, University of Bradford, Bradford BD7 1DP, UK.
`* Now at University of Bahrain, Gulf Polytechnique.
`
`”Idem":
`
`[
`
`Derms
`
`igggg'm“
`
`
`
`F‘” ‘9’“
`a/Subtutuneous
`
`\arterySubcutaneous
`
`yam
`Fig. l. Diagrammatic representation of skin and the cutaneous
`vascular system.
`
`0169-2607/89/$03.50 © 1989 Elsevier Science Publishers B.V. (Biomedical Division)
`
`Page 1
`
`VALENCELL EXHIBIT 2013
`
`IPR2017-00321
`
`Page 1
`
`VALENCELL EXHIBIT 2013
`IPR2017-00321
`
`
`
`258
`
`total skin blood flow that is employed for thermo-
`regulation, with a reduced sensitivity to changes in
`nutritional capillary flow. It has the practical dis-
`advantage of requiring an environment in which
`ambient temperature and humidity are kept con-
`stant. Thermal clearance or thermal conductance
`[9] methods assess the rate of removal of heat
`from a heated area at the centre of a probe by the
`skin’s nutrient blood flow in the dermis. The
`thermal clearance transducer measures the tem-
`perature difference between a heated copper disc
`at the centre of the probe and an unheated, con-
`centric copper annulus at its periphery. The main
`practical disadvantages are the relatively long lag
`period before a constant reading is obtained and
`the application of heat to the skin modifies local
`skin conditions.
`
`The laser Doppler technique depends on the
`Doppler shift of coherent
`(laser)
`light
`‘back
`scattered’,
`from moving red blood cells. This
`frequency shift is due to the velocity of blood cells
`
`(particles) within the tissue and, therefore, is re-
`lated to tissue blood flow. To measure changes in
`nutritional blood supply the geometry of
`the
`capillary loops needs to be known which is an
`impossibility because of its random variations.
`Laser light.
`in the red region of the spectrum,
`penetrates further than incoherent
`light as the
`beam is narrower making the intensity per unit
`area greater rendering the laser Doppler method
`able to collect data from deeper lying blood ves-
`sels [48].
`The radioactive isotope clearance technique in-
`volves the measurement of the clearance rate of
`
`the
`intradermally injected radiopharmaceuticals,
`most widely used tracer being ”3 Xe. This method
`required access to a 7 (gamma) camera or surface
`counting technique with exposure to a small
`amount of radiation and involves percutaneous
`injection. The trauma may be unacceptable to
`diseased skin and repeated measurement in the
`same patient is limited by the radiation dose.
`
`l
`
`" III
`
`"into
`
`
`llllmummmgunllnmlmIIt
`
`l
`
`l
`20
`
`l
`30
`
`
`
`lllllll‘lllllllllllullIllllllllllIlll|||Illlllll||||l‘ll|||||Ill
` 708090m00l20|30|40|5
`
`
`
`so SO 60
`
`53-22
`
`TlllllllllllllllllllIITIIIl|lllllllIllllllllllllllllllllmllll
`
`red LED photoplethysmograph. green LED photoplethysmograph.
`Fig. 2. Photograph of plethysmographs. From left to right:
`piezoelectric plethysmograph. and bulb photoplethysmograph.
`
`Page 2
`
`
`
`259
`
`[26] were the first to use the term ‘Photoplethys—
`mograph’
`and suggested that
`the
`resultant
`‘Plethysmogram’ represented volumetric changes
`in the blood vessels of the skin. Hertzman [25]
`published his paper on the subject of photo-
`plethysmography demonstrating the effect of cold
`and exercise on blood volume changes in the limb.
`He also established the validity of the method for
`estimating both skin blood flow and blood volume
`changes, where a steady blood flow appears as a
`constant volume.
`
`2. Skin photoplethysmography
`
`Photoplethysmography satisfies most of the condi-
`tions for a non-invasive technique to estimate skin
`blood flow and is ideally suited to situations which
`require measurement to be made over long peri-
`ods. It is a technique that provides a signal pro-
`portional to changes in skin blood volume but
`does not produce a quantitative measure [15]. The
`dc. component represents total red cell volume
`below the sensor plus some reflected components
`from within the skin (see below). The a.c. compo-
`nent is produced by the fluctuations in the blood
`volume below the sensor. The volume changes
`recorded sequentially reflect the variations in flow.
`Thus the a.c. component is a measure of changing
`flow. lts attraction is that it is the least invasive
`
`
`
`method and atraumatic, as well as being inexpen-
`sive. Fairs et a1. [21] have recently demonstrated
`that photoplethysmography and Doppler flowme-
`try correlate well with venous occlusion plethys-
`mography and stated that ‘optical methods of skin
`blood flow measurement approach the ideai situa-
`tion of non-invasion’.
`Fig. 2 shows a photograph of photoplethysmo-
`graphs constructed from two light sources and a
`detector. The signal produced by the photo-
`plethysmograph depends upon the location and
`the properties of the subject’s skin at that site,
`including the skin structure,
`the blood oxygen
`saturation, blood flow rate and the skin tempera-
`tures.
`The importance of the spectral reflectance and
`transmittance of skin has long been recognised
`and many workers have used such measures to
`
`The basic technique applied in electrical imped-
`ance methods requires a high frequency, constant
`amplitude, current to pass through the segment of
`interest. The measured voltage appears as a result
`of the segment impedance. The electrode area and
`spacing,
`together with the choice of operating
`frequency determine the form of signal produced.
`The impedance techniques provide an indication
`of blood volume in deeper vessels. The main diffi-
`culty of this technique is to provide good contact
`between the skin and the electrodes.
`
`Plastic piezoelectric microphones are now com-
`mercially available and they have many applica-
`tions [23], they can be used as plethysmographs
`[45]; one is shown in Fig. 2. As piezoelectric
`plethysmographs they are used to measure the
`movement of the skin. A major disadvantage is
`that to obtain a reasonable signal their size has to
`be about 70 mm by 10 mm so the signal produced
`gives an indication of the skin movement over a
`relatively large area. The signal contains only an
`oscillating (a.c.) component because of the rapid
`decay of the acquired charge through its internal
`resistance and the impedance of the connected
`circuits. The lower frequencies are attenuated due
`to the rapid loss of the charge. No steady-state
`(d.c.) component is generated. The piezoelectric
`plethysmograph produces a signal which is the
`differential of
`the photoplethysmograph signal
`
`[45].
`Plethysmography applies to the measurement
`or estimation of the fullness or volume of an
`object. Volume and strain gauge techniques re-
`main true to this definition in the assessment of
`blood flow. Volumetric measurements of limbs
`were accurately taken well before the advent of
`electronic transducers and polygraphs [4,37,41].
`Plethysmographs have been productively applied
`in peripheral circulatory studies [25,27,621 also to
`estimate arterial blood flow [22,33] and variations
`in vasomotor tone [28—30]. A recent review of
`volume plethysmography is given by Porter and
`Swain [50].
`Photoelectric plethysmography, or photo-
`plethysmography, was introduced almost simulta-
`neously in 1938 by Hertzman [25] in the United
`States and Matthes and Hauss [42] in Germany
`but Hertzman [24] and Hertzman and Spielman
`
`Page 3
`
`Page 3
`
`
`
`260
`
`determine the state of the skin circulation. Daw-
`son et a1. [14], in an in vivo study of light absorp-
`tion and scattering in skin, described a theoretical
`basis for the application of the logarithm of the
`inverse of reflectance (LIR). An instrument was
`constructed specifically for the measurement of
`LIR spectra in the assessment of skin colour. The
`spectrum of light reflected from the skin is related
`to the light absorbing and scattering structures
`within the skin. The clinical use of the instrument
`
`was also reported and the calculation of indices
`which may be used to quantify erythema and
`pigmentation of skin. From these observations
`Dawson et a1. [14] concluded that the LIR spectra
`can be resolved into three principal components
`which correspond to two layers, one containing
`melanin and the other the subpapillary venous
`plexus, and a residual term comprising the contri-
`butions of
`fibrous protein, collagen and fat.
`Melanin absorbs strongly over the visible spec-
`trum, its absorption increasing towards the ultra—
`violet while whole blood has a relatively small
`absorption at wavelengths greater than 620 nm, as
`illustrated in Fig. 3.
`Haemoglobin absorbs strongly in the yellow
`band of the visible spectrum. Oxygenated and
`de-oxygenated blood have similar absorption coef-
`ficients at wavelengths greater than 805 nm [11],
`but below 805 nm their absorption coefficients
`
`greatly diverge with those of oxygenated greater
`than the de—oxygenated blood [61]. The differen-
`tial haemoglobin saturation level
`is significant
`when considering total blood volume. Recently,
`devices based on the photoplethysmograph have
`been available for use in anaesthesia which pro-
`duce a pulse rate and percentage oxygen satura-
`tion of haemoglobin (pulse oximetry).
`
`3. Photoplethysmograph operation
`
`The construction of a photoplethysmograph re-
`quires a light source and a detector. Their relative
`positions may vary in individual designs. On cer-
`tain areas of the body, such as the ear lobe and
`fingers, it is possible to have the light source and
`the detector opposite each other on the skin
`surface, an arrangement known as transmission
`mode, but this limits the application possibilities.
`A more acceptable arrangement, known as reflec~
`tion mode [46], is such that the light source and
`the photodetector are placed side by side, and this
`arrangement can be applied to any part of the
`body. The principle of operation of the photo-
`plethysmograph is based on the assumption that
`light is attenuated when it is shone on to the skin
`and the attenuation shows variation depending on
`the volume of blood entering the tissue under
`observation. This attenuation is due to reflection,
`
`log{inverse
`
`03
`
`0'2
`
`01
`
`0
`450
`
`
`
`scattering and absorption of light. Such absorp-
`tion of light depends not only upon the skin
`pigments, melanin and other substances which
`absorb specific spectral bands, see Fig. 3, but also
`upon a scattering action due to the structural
`inhomogeneities in the skin. Absorption of light
`may depend on haematocrit and on the orienta-
`tion of red cells within the vessels, an explanation
`suggested by Hocherman and Polti [32].
`The wavelength of the source used is of signifi-
`cant importance. Light sources that operate in the
`near red (600-700 nm) region of the spectrum are
`most effective because haemoglobin is the major
`protein present which changes with time and
`scatters the light in this region. In this area of the
`spectrum the absorption of
`skin pigments,
`oxyhaemoglobin, haemoglobin and bilimbin with
`the exception of melanin fall
`to their minima.
`
`ref/Edam)
`
`500
`
`550
`
`600
`
`650
`
`700
`
`Wavelength nm
`
`Fig. 3. Absorption curves for melanin and blood in vitro. (A)
`Melanin; (B) whole blood.
`
`Page 4
`
`Page 4
`
`
`
`261
`
`Roberts [52] in a study of fundamental aspects
`of the optical properties of blood in motion, out-
`lined a review of mathematical analyses of light
`scattering and diffusion processes taking place in
`blood and proposed that light diffusion through
`blood, from a source outside a blood vessel, can
`
`diffuse preferentially in the direction of motion of
`blood which is added to the static diffusion pro-
`cess. Roberts [52] demonstrated that a reflectance
`photoplethysmograph, of reasonable size (not in-
`corporating fibre optics or miniature transducers),
`constructed as a detector placed between two un-
`collimated light sources would be relatively insen-
`sitive to small positional changes in measuring
`blood flow either over a diffuse vascular bed or
`
`over a single artery.
`
`4. LED photoplethysmograph and amplifier
`
`The problem with using tungsten filament bulbs is
`the excessive heat produced. The frequency con-
`tent of the light emitted from tungsten fillament
`bulbs is in the infrared area of the spectrum. The
`ratio of heat to light output is large, therefore the
`heating of the skin can be a limiting factor. We
`have found that usually the bulb fails when the
`filament partially shorts out producing extra light
`and heat. This will disturb the calibration of the
`photoplethysmograph and may cause burning of
`the skin. Normally a reflection type photo-
`plethysmograph is used which incorporates a de-
`tector mounted in between two light sources or a
`light source alongside the detector. We have re-
`
`vessels and hence of the activity of the sym-
`pathetic nervous system [10].
`
`3
`1
`
`5
`
`072V
`Red
`Duran
`Yellow
`
`0V
`Green
`
`”1
`15K
`
`Surfare
`
`
`
`Infill
`
`0—-
`
`l
`
`Another factor which makes this part of the spec-
`trum, 600—700 nm, favourable is that the skin not
`only has minimum absorption but also the dif-
`ference in the optical density of the skin due to
`erythema is negligible [3]. With pulsating flow,
`there is a variation in the light received at the
`detector on the skin surface [20], but the pulsatile
`component is largely independent of wavelength
`in the range 660-805 nm. Challoner [12] con-
`firmed this by comparing the outputs of photo-
`plethysmographs operating at 650 nm and 805
`nm, and found that their pulsatile outputs were
`substantially the same. Although the pulsatile
`component at heart rate does not require a critical
`narrow spectral range,
`the values of the lower
`frequency components are better detected using a
`light emitter in the range 600—700 nm. Much of
`the skin has a poor pulsatile component but the
`precapillary activity is present throughout.
`The signal detected by the photoplethysmo-
`graph consists of a steady component (d.c.), which
`is related to the relative vascularisation of the
`tissue, and a pulsatile component (a.c.), which is
`related to changing blood pulse volume [24]. The
`resultant signal is a measure of expansion of skin
`vessels (predominantly arterial) which is a summa-
`tion effect of the arterial pulse and the opposing
`elastic properties of the vessel wall. The amplitude
`of the volume pulsation with each heart beat is
`closely correlated with the flow. Both amplitude of
`pulsation and blood flow serve as an index of the
`state of contraction of the walls of the blood
`
`
`
`
`
`5
`
`c
`
`at
`hmsmg
`
`HPF
`
`I
`I
`
`I
`l
`
`Eufler
`
`I
`l
`
`Amplifier
`
`6
`
`
`Blue
`
`- 12 v
`“3
`4 0——
`Green
`C
`
`of the photoplethysmow
`Fig. 4. (a) Diagrammatic representation of the photoplethysmograph electrical circuit. (b) Circuit diagram
`graph amplifier. Typical values are Rl,6 100 k; R2 500 k; R3,7 l M; R45 15 k; Cl 10 MP; C2 0.022 pF; C3 100 pF; Al.2 324E].
`
`Page 5
`
`Page 5
`
`
`
`‘0
`
`Modified
`
`Commercial
`
`
`
`0- 01
`
`0-1
`
`1-0
`
`10 '0
`
`Frequency Hz ——->
`
`Fig. 5. Bode diagram of plethysmograph amplifier. The com-
`mercial curve is obtained from a PP120 Pulse Plethysmograph
`Transducer manufactured by BioScience. Sheerness, Kent, U.K.
`
`placed the tungsten bulbs with light emitting di—
`odes (LEDs) working in the red region with a
`narrow wave band and peak output at 640 nm.
`Above 620 nm absorption by whole blood is rela-
`tively small, Fig. 3 [14]. The LEDs last for a much
`longer time and produce less heat compared to the
`tungsten bulbs. A typical example is shown in Fig.
`2. Similar photoplethysmographs using LEDs are
`now commercially available.
`When an LED photoplethysmograph is con»
`nected to an appropriately designed amplifier cir-
`cuit as shown in Fig. 41), it provides an acceptable
`pass band (Fig. 5). The low frequency cutoff is a
`compromise between the lowest frequency which
`can be measured and patient acceptability,
`the
`lower the cutoff the longer the amplifier settling
`time. With the response shown in Fig. 5 the
`amplifier design is simple and has a low frequency
`cutoff at about 0.01 Hz, with a settling time of 1
`min, enabling nearly all the required information
`to be measured. By adjusting the resistor R2 (Fig.
`4b), d.c. biasing is achieved and is set to be nearly
`zero, this reduces the time taken by the circuit to
`settle down. The total signal is obtained from the
`points marked (i) and (ii) while the ac. compo-
`nent is obtained from the points marked (iii) and
`(iv) after filtering and amplification. Care must be
`taken to ensure that the operating point of the
`detector is on the linear part of its characteristics
`so that harmonics are not generated that will add
`to the harmonic components of the blood flow
`signal.
`
`
`
`After calibration of the energy output from the
`LEDs by adjusting the current such that the detec-
`tor receives the correct amount of energy for it to
`work within the linear part of its characteristics,
`the photoplethysmograph will produce a satisfac--
`tory signal. The signal is even larger, for a given
`power input,
`than that produced by the bulb
`photoplethysmograph. Typical power inputs are
`about 30 mW and 5 mW. (Note that these powers
`depend upon the recent history of the detector [47]
`if a cadmium sulphide detector is used.) The oper—
`ating temperature of both the LED and bulb
`photoplethysmographs, contained in a tempera-
`ture-controlled environment, was measured by
`placing a thermocouple on each device and moni-
`toring their temperatures. This showed that the
`LED produces much less heat output than the
`bulb, making heat injury to the skin less probable.
`This is a minor problem but local heating effects
`cause vasodilatations and modify the signal.
`Both photoplethysmographs were applied to the
`same site on the arm and the output signal was
`measured over a period of time, within a con-
`trolled environment. There was no change in the
`LED photopiethysmograph signal while the bulb
`photoplethysmograph produced an increase in
`output amplitude after a short period of time. This
`increase was due to heat produced by the bulb
`causing vasodilatation in the vascular bed.
`Measurements have been performed to test the
`standardization and performance; and to evaluate
`the reproducibility and comparability of the pho-
`toplethysmographs [63]. To show that
`the per-
`ceived differences in the spectra from contralateral
`anatomical sites were due to physiological effects
`as opposed to transducer and instrumentation
`artefacts the following experiments were per-
`formed. Spectra were obtained from the right and
`left finger pulps of four healthy male subjects at
`rest;
`the measurements were repeated with the
`transducers transposed. The derived spectral pairs
`were differenced on a point-to-point basis and a
`one-way analysis of variance was performed on
`the data sets to establish whether or not the mean
`
`was significantly different from zero. This was
`performed individually for the low (below 0.5 Hz)
`and the high (0.5—2 Hz) spectral regions and no
`significant differences were detected (P < 0.01).
`
`Page 6
`
`Page 6
`
`
`
`
`
`blood flows away from the heart the number of
`harmonics in the blood pressure signal increases.
`The photoplethysmograph detects this signal after
`it has been filtered by the precapillary sphincter
`which has the effect of attenuating the higher
`harmonics. Taylor [59] deduced from a model of
`the vascular
`system consisting of
`randomly
`branching tubes that under conditions of propa-
`gation equivalent
`to those in a living body a
`visco-elastic vessel wall attenuates the amplitude
`of an oscillation by more than 40% after travelling
`one wavelength. The distance from the heart to
`the feet is about a quarter of the wavelength of the
`fundamental component of the pressure pulse [43].
`The fundamental is decreased and the 2nd, 3rd
`
`units
`
`Frequenty Hz
`
`Fig. 6. Total frequency content of photoplethysmograph signal
`(sample time 41 5, AF 24.41 mHz. AT 40 ms). Note that all
`the spectra described in this paper have been obtained using a
`Hewlett Packard 5420A Digital Spectrum Analyser coupled to
`a Hewlett Packard 70058 X —Y Plotter and some of the rapid
`descending curves seem to have a negative slope. This illustra-
`tion is due to the method of producing the graphs.
`
`at the finger capillary bed, Fig. 6, it is difficult to
`detect at and above the third harmonic. As the
`
`and 4th harmonics are amplified in transmission.
`The higher harmonics are filtered out by the pre-
`capillary sphincter.
`As the heart rate increases the sidebands de-
`
`crease; above a heart rate of 1.5 Hz (90 bpm)
`there is no heart rate variability (HRV) [2]. Sayers
`[54] indicates how HRV can be measured; a good
`review of the present status is given by Kitney and
`Rompleman [38]. HRV is due to the interaction of
`
`263
`
`
`
`Amplitudearbitrary
`
`This procedure was repeated at weekly intervals
`for 4 weeks and similar results were obtained.
`
`Reproducibility and comparability are such that
`real differences detected under control conditions
`
`are true estimates of physiological functions.
`
`5. Photoplethysmograph signal
`
`The pulsatile component of the photoplethysmo-
`graph signal is due to the pumping action of the
`heart. Measurement of this signal can give consid-
`erable information about the changes in the circu-
`lation of a specific vascular bed [12]. Simultaneous
`measurement of the a.c. and the dc. components
`provides more information on the vascular changes
`of the skin than could be obtained by either
`component separately [17].
`Weinmann [61] concluded that the pulse shape
`and amplitude can vary with the relative position
`between the detector and the vessel under study.
`This may produce unreliable information about
`the same skin site. Weinmann increased his
`
`experimental errors by comparing the data from
`two different devices, one using the translumina-
`tion principle with the detector placed on the
`other side of the finger from the illuminator, the
`other the back-scattered principle with the detec-
`tor alongside the illuminator. These criticisms,
`however, do not apply to the cutaneous measure-
`ments.
`
`the a.c. components of the
`Fig. 6 shows all
`photoplethysmograph signal for a resting supine
`male of 20 years taken from the index finger pulp
`in the frequency domain from 0.01 Hz to 6 Hz, i.e.
`with the dc. content
`removed. There are low
`
`frequency components, below 0.5 Hz, and the
`heart rate component between 0.5 and 2 Hz de-
`pending on the individual heart rate. The normal
`subject has a heart rate between 0.9 and 1.17 Hz
`(55—70 beats per minute (bpm)) so the harmonics
`will be above 2 Hz and can be seen to have the
`
`same sidebands as the fundamental. In Fig. 6 the
`heart rate is 1.14 Hz (68 bpm).
`Attenuation of pressure and flow waves occurs
`as a function of distance travelled through the
`arterial system [43]. In the aorta the harmonic
`components spread up to the 20th harmonic while
`
`Page 7
`
`Page 7
`
`
`
`264
`
`
`
`
`
`AmplitudearbitraryUfllf’S
`
`Fig. 7, is at 1.14 Hz and can be considered as a
`carrier frequency, with its sidebands being the
`other peaks to each side of it. The sidebands occur
`at the same frequency differences from the carrier
`frequency as the frequencies which are present
`below 0.5 Hz. Mearns et al. [44] used the tech-
`nique of forcing frequency to change the frequency
`of some of the components below 0.5 Hz and the
`corresponding side bands above 0.5 Hz changed.
`Side bands occur when there is modulation of the
`
`carrier frequency, this modulation could be either
`amplitude modulation (AM), frequency modula-
`tion (FM) or a combination of both. In the photo-
`plethysmograph signal both AM and FM are pre-
`sent but can vary in proportion. The variation in
`the resistance to blood flow due to the changes in
`the intrathoracic pressure is the main cause of the
`amplitude modulation of the ventricular volume.
`This causes the AM of the ventricular ejection into
`the aorta. The heart rate is frequency modulated
`by the effect of the vagal efferents upon the sinus
`node. Therefore the train of pressure waves travel-
`ling throughout
`the vascular
`tree carries
`the
`breathing rate as AM and FM.
`The smooth muscle of the vascular tree beats
`
`co-ordinately at about 0.1 Hz and this modulates
`the travelling pressure flow pulse train. This
`Traub-Hering-Mayer (THM) frequency [49] is the
`product of a brain stem outflow into the auto-
`nomic nervous system; via the sympathetic out-
`flow the smooth muscle of the vascular tree is
`
`
`
`co—ordinated at 0.1 Hz; via the parasympathetic
`outflow (vagus nerve) the sinus node is frequency
`modulated. The information content of the signal
`arriving at
`the precapillary sphincter contains
`breathing rate and THM frequencies in both am-
`plitude- and frequency-modulated form. The re-
`flection coefficient at the precapillary sphincter is
`around 70% [43],
`the rest of the energy travels
`through the precapillary sphincter. This explains
`the information content about the pulse rate in
`Fig. 7. A small amount of the energy below 0.5 Hz
`is provided in this manner by the filtering effect of
`the precapillary sphincter and the arteries. How-
`ever,
`the low frequency components are more
`powerful
`than suggested by these mechanisms,
`especially in areas of the skin where the pulsatile
`components are weak or absent. This lower
`
`Frequent'y Hz
`
`Fig. 7. Frequency spectrum of photoplethysmograph signal —
`subject at rest (sample time 128 5. AF 7.8 ml-lz, AT 125 ms).
`
`the parasympathetic nervous system (vagal con-
`trol) with the heart rate in normal sinus rhythm.
`Fig. 7 illustrates the ac. frequency components
`of the photoplethysmograph signal between 0.01
`and 2 Hz, and is derived from the same signal as
`Fig. 6, but. because of the reduced frequency range
`Fig. 7 has a better resolution; again the dc. com-
`ponent has been removed.
`The information content above 2 Hz is due to
`
`the harmonics and has minimal clinical value.
`
`Therefore the important information in the signal
`can be extracted by observing the signal up to 2
`Hz. If the heart rate is below‘l Hz (60 bpm) the
`harmonics of t