Abstract— For applications where PPG signal AC component

needs to be measured without disturbances in its shape and

recorded digitally with high digitalization accuracy, the step-

by-step DC component eliminator is developed. This paper

describes step-by-step DC component eliminator, which is

utilized with analogue comparator and operational amplifier. It

allows to record PPG signal without disturbances in its shape in

24-hours PPG signal monitoring system. The experiments with

PPG signal have been carried out.

I. INTRODUCTION

hotoplethysmography (PPG) is a non-invasive optical technique for measuring changes in blood circulation. In

PPG the optical radiation from a light source is brought to illuminate the skin. The light, which is often red or infrared, is absorbed, reflected and diffusively scattered in the tissue and blood. Changes in intensity of received light, by photodetector, are related to blood flow and blood volume changes occurring in the underlying tissue [1].

Mainly there are two ways to measure PPG signal: reflection and transmission mode. In reflection mode, a photodetector is placed adjacent to the light source. Light from light emitting diode (LED) is directed toward the skin. Only a small fraction of the emitted light is received by the photodetector. The photodetector measures the reflected and scattered light intensity from skin surface. In transmission mode, the photodetector and the light source are placed opposite sides of the measured volume. The photodetector measures the transmitted light intensity.

In reflection mode the PPG signal shape depends on the depth that light penetrates in a tissue. It is function of wavelength and the optical geometry of the probe [2, 3]. Light penetration into the tissue increases primarily with increasing wavelength. It also increases with increasing distance between the light source and the photodetector. It is

Manuscript received April 16, 2007. This work was supported in part by

the Estonian Science Foundation under Grant G5888. K. Pilt, is with the Department of Biomedical Engineering, Tallinn

University of Technology, Tallinn, 19086 Estonia (phone: 372-56-918-736; e-mail: kristjan.pilt@gmail.com)

K. Meigas, is with the Department of Biomedical Engineering, Tallinn University of Technology, Tallinn, 19086 Estonia (e-mail: kalju@cb.ttu.ee)

J. Lass, is with the Department of Biomedical Engineering, Tallinn University of Technology, Tallinn, 19086 Estonia (e-mail: jaanus@cb.ttu.ee)

M.Rosmann, is with Tensiotrace OÜ, Majaka 26, 11412, Tallinn Estonia (e-mail: rosmann@trenet.ee)

J. Kaik, is with the Department of Biomedical Engineering, Tallinn University of Technology, Tallinn, 19086 Estonia (e-mail: jyri@cb.ttu.ee)

possible to reach different tissue layers and measure its PPG signal.

PPG signal consists of two main parts: DC and AC component. The AC component is synchronous with heart rate and depends on the pulsatile pressure [4] and pulsatile blood volume [5] changes. The DC component of the signal varies slowly and reflects variations in total blood volume of the examined tissue. PPG is used in a number of non-invasive instruments to monitor different parameters, such as pulse rate, respiratory rate, blood pressure, etc. In those devices, the AC component carries important information.

In PPG signal, the AC component amplitude can be more than ten times smaller than DC component amplitude. In 24-hour PPG signal monitoring devices, usually the signal is recorded digitally. Unfortunately, such a small AC signal is difficult to accurately measure directly with an analogue-to-digital converter (ADC). The DC component is needed to eliminate before PPG signal recording.

The aim of the study was to develop a device for 24-hour PPG signal monitoring to eliminate the DC component from the PPG signal and keep AC component shape unchanged from filter caused distortions before signal digital recording.

II. MATERIALS AND METHODS

The simplest way to remove DC component from PPG signal is to use analogue RC high-pass filter. The filter is suggested to set 10 times lower frequency from heart beat frequency, which reduces distortion caused by analogue filter non-linear phase. RC filter can also cause capacitor charging and discharging slopes, because of artifacts. The slope duration is inversely proportional to RC filter cut-off frequency. During artifact caused slopes, the PPG signal is disturbed and the signal under interest is lost.

For DC component elimination before recording the PPG signal the step-by-step DC component eliminator is used. It is implemented with comparator and operational amplifier (Fig. 1.). Due to the DC component and its drift the steps are created into the signal and the signal dynamic range is kept between certain values.

The DC component eliminator input signal is obtained from PPG sensor, which consists of infrared light emitting diode (IR LED) and photosensitive element. The signal is sent into analogue comparator, where it is compared with threshold voltages. The output of analogue comparator is compensation voltage, which is sent into operational amplifier. If input signal exceeds comparator threshold

Analogue Step-by-Step DC Component Eliminator for 24–Hour PPG

Signal Monitoring

Kristjan Pilt, Kalju Meigas, Jaanus Lass, Mart Rosmann, and Jüri Kaik

P

Proceedings of the 29th Annual InternationalConference of the IEEE EMBSCité Internationale, Lyon, FranceAugust 23-26, 2007.

ThP2C2.20

1-4244-0788-5/07/$20.00 ©2007 IEEE 1006

voltage, the constant compensation voltage is set respectively. To avoid the rapid comparator switching between the two states, the additional voltage is given into comparator input to provide hysteresis. Operational amplifier is set into inverting mode, where the input signal DC component is changed respectively to the compensation voltage.

The DC component eliminator circuit incorporates TL074 operational amplifiers, which are also used in analogue comparator and amplification block. The circuit supply voltage is +/-5V. In analogue comparator, the threshold voltage levels are set with the resistor array, the values of which are 1, 2, 3 and 4V. Because of hysteresis, the threshold voltages are 0.5, 1.5, 2.5, and 3.5V in switching the comparator on a lower state. Hysteresis is obtained by resistor, which is set as positive feedback from comparator output to comparator PPG signal input. The given compensation voltage is equal with every exceeded threshold voltage level. As a result, the output signal is kept between –1.0 and 0.5V.

DC component eliminator is part of the 24-hours PPG signal monitoring system [7], which block diagram is showed on Fig. 2. It consists of two parts. In the first part the signal after DC component elimination is passed through amplifier, low-pass filter, ADC and then being recorded into memory. In different applications, the signal can be saved into PC, flash card, etc. The second signal chain describes data processing of the recorded signal, which is done in PC.

Sometimes the AC component of PPG signal can be as small as one to four least significant bits on 10-bit system [6]. To increase the AC component ratio to the least significant bit value the PPG signal can be amplified. For example, with ADC, with the working range of 5V for analogue signals, the DC component eliminator output signal, which signal range is 1.5 volts, is possible to be amplified 3.3 times. Furthermore, the DC component eliminator output signal range can be more limited. It gives a possibility to increase amplifier gain and the ADC least significant bit value ratio to the AC component is improved. Still, the range cannot limit into too small value. It may cause

errors in further signal reconstruction.

To set reasonable DC component eliminator output signal range the experiments must be carried out. The change of comparator threshold voltages for every experiment can be too time consumable. For every experiment, the resistor arrays and hysteresis caused resistors need to be change. One of the possibilities is to use programmable integrated circuit (PIC) microcontroller instead of analogue comparator. In this case, the threshold voltages are easy to change and the PIC is only needed to reprogram.

Before the analogue to digital conversion is done for the signal recording, the PPG signal must be limited from high frequencies. The low-pass filter cut-off frequency must be high enough not to cause distortions in comparator made steps in signal. If the cut-off frequency is too low, the capacitor charging shaped slopes are appeared in the place of sharp steps. The signal reconstruction gives errors in the place, where comparator steps were made. The other possibility to eliminate high frequencies is to place the low-pass filter before DC component eliminator where are no limits in choosing cut-off frequency.

III. RESULTS

The experiment set up is showed on Fig 3. Three signals are recorded synchronously into PC memory using LabVIEW environment and National Instruments DAQCard 6034E. In DAQCard, the analogue-to-digital conversion is done with 1kHz sampling rate and 16-bit resolution.The first recorded signal (“A”) is PPG signal, which is lead through low-pass filter. The second recorded signal (“B”) is compensation signal, which is taken from DC component eliminator comparator output. The third recorded signal

Fig. 2. Block diagram of the 24-hour PPG signal monitoring system.

Fig. 1. Block diagram of the step-by-step DC component eliminator.

1007

(“C”) is output of signal chain, which includes DC component eliminator, amplifier with constant gain and low-pass filter. The amplifier is with gain 3 and low-pass filter cut-off frequency is set to 500Hz.

From previous studies, the signal reconstruction algorithm has been implemented in MATLAB. Signal reconstruction for third signal as well as first and second signal observation is done in MATLAB.

The PPG signal is measured with reflectance sensor. The signal is between 0 and 5 volt. The pressure on PPG sensor was changed to give drift to the DC component.

The recorded signals are given on Fig. 4. As the DC component eliminator operational amplifier is in inverting mode, the first signal (Fig. 4a) is inverted as well. Due to first signal inversion, it is in the range of –5 to 0 volt. The third signal (Fig. 4c) is in the range of –3.5 to 1.5 volt, which is caused by DC component eliminator and amplifier.

Second signal (Fig. 4b) is caused by PPG signal DC component and its drift. As the PPG signal DC component is in the range of –2V to –1V, the comparator output gives +2V compensation signal. Near 3rd second the PPG signal has drifted to the range of –3V to –2V and compensation signal is switched to 3V. Near 10th second the PPG signal has exceeded the hysteresis voltage and comparator is switching back to the range of –2 to –1V. Compensation signal returns to 2V.

Reconstructed signal is given on Fig. 4d. The steps are eliminated and signal has its shape with DC component. It is possible to eliminate the unwanted frequencies included DC component with different signal processing methods. One

Fig. 3. Block diagram for three signals A, B, C recording into memory and reconstructing signal C.

Fig. 4. a) Recorded PPG signal, which is lead through low-pass filter (“A”) b) Comparator compensation signal (“B”) c) Output of signal chain with DC eliminator (“C”) d) Reconstructed signal.

1008

possibility is to use finite impulse response filters, which have linear phase and no effect on signal shape.

Because of the amplification with gain 3, before ADC, it is visible from the Fig. 4d, that signal has higher AC component amplitude than it is on Fig. 4a. The third signal AC component is recorded with higher digitalization accuracy. In addition, it does not have any distortions, which are caused by analogue filter non-linear phase.

IV. CONCLUSION

The step-by-step DC component eliminator has been developed and implemented in hardware for PPG signal 24-hour monitoring. As result from the recorded signals it revealed, that in using step-by-step DC component eliminator the PPG signal has higher digitalization accuracy. Also after signal reconstruction, the AC component remains unchanged in the mean of its shape. For the future work there will be carried out experiments to set the threshold voltages with reasonable value for comparator.

REFERENCES [1] A. Kamal, J. Harness, G. Irving, and A. Mearns, “Skin

photoplethysmography – a review,” Comput. Methods Progr.

Biomed., 1989, 28, pp 257-269 [2] L. G. Lindberg, P. Å. Öberg, “Photoplethysmography. Part 2.

Influence of light source wavelength,” Med. Biol. Eng. Comput., 1991, 2, pp 48-54

[3] I. Fridolin, & L. G. Lindberg, “Optical non-invasive technique for vessel imaging: I. Experimental results,” Phys. Med. Biol., 2000, 45, pp 3765-3778

[4] S. C. Millasseau, F. G. Guigui, R. P. Kelly, et al., “Non-invasive assessment of the digital volume pulse. Comparison with the peripheral pressure pulse,” Hypertension, 2000, 36, pp 952-956

[5] A. Babchenko, E. Davidson, Y. Ginosar, et al., “Photoplethysmographic measurement of changes in total and pulsatile tissue blood volume, following sympathetic blockade,” Physiol. Meas., 2001, 22, pp 389-396

[6] D. Thompson, A. Wareing, D. Day, S. Warren, “Pulse oximeter improvement with and ADC-DAC feedback loop and a radial reflec-tance sensor,” IEEE Proc. EMBS Annual Int. Conference, New York

City, USA, 2006, pp 815-818 [7] K. Pilt, K. Meigas, J. Lass, M. Rosmann, “Signal processing methods

for PPG module to increase signal quality,” IFMBE Proc.

Mediterranean Conf. on Med. Biol. Eng. Comput., to be published

1009