Cardio-Vascular Response Following Exercise Final Report for EECE502

By Mohamed Ali Eid Yusr Sabra Mirna Abou Mjahed

American University of Beirut May 23, 2005

Cardio-Vascular Response Following Exercise Progress Report for EECE501

By Mohamed Ali Eid Yusr Sabra Myrna Abou Mjahed

American University of Beirut January 12, 2005

Table of Contents Abstract........................................................................................................ iv List of Figures .............................................................................................. v List of Tables ............................................................................................... vii 1.0 Introduction ............................................................................................ 1 1.1 Background and Overview .............................................................. 1 1.2 Project Objectives ........................................................................... 1 1.3 Report Organization ........................................................................ 2 2.0 Literature Review ................................................................................... 4 2.1 The Heart: General Anatomy and Cardiac Cycle ............................ 4 2.2 Heart Signals .................................................................................. 5 2.2.1 Electrical Signal .................................................................... 5 2.2.2 Sound Signal......................................................................... 7 2.2.3 Pressure Signal..................................................................... 8 2.2.4 Impedance Cardiograph........................................................ 8 3.0 Project Approach: Design and Analysis ................................................. 12 3.1 Signal Choice.................................................................................. 12 3.2 ICG Circuit ...................................................................................... 14 3.2.1 Voltage Controlled Current Source. ....................................... 15 3.2.2 Implemented Design. ............................................................. 16 3.3 Computer Analysis .......................................................................... 21 3.3.1 LabView Modules................................................................... 22 3.3.2 Visual Basic Application. ........................................................ 24 4.0 Implementation ...................................................................................... 25 4.1 Implementation of the ICG Circuit ................................................... 25 4.2 Computer Analysis .......................................................................... 26 4.2.1 Implementation of the LabView Modules. .............................. 26 4.2.2 Implementation of the Visual Basic Application..................... 33 5.0 Results and Analysis.............................................................................. 36 5.1 Test Results .................................................................................... 36 5.2 Analysis........................................................................................... 39 6.0 Other Issues........................................................................................... 41 7.0 Conclusion. ............................................................................................ 42 References............................................................................................. 43 Appendix ................................................................................................ 45

iii

Abstract Cardio-Vascular Response Following Exercise By: Mohamed Ali-Eid, Yusr Sabra and Mirna Abou Mjahed The correlation between the rate the strength of a heart beat returns to normal following exercise and cardiovascular health has never been documented. This report presents the process of developing a device capable of picking up an impedance signal from the heart and transmitting it into a personal computer for analysis. The impedance signal is proportional to the stroke volume (i.e. strength of the heart beat) which allows the device to be used for research on the correlation between cardiac health and the rate of decrease of the stroke volume.

iv

List of Figures Figure 1 Basic anatomy of the human heart ................................................ 4 Figure 2 A typical ECG recording of a single cardiac cycle.......................... 6 Figure 3 A typical sound recording of a single cardiac cycle........................ 7 Figure 4 Typical impedance cardiography signal ......................................... 9 Figure 5 Diagrammatic representation of impedance cardiography............. 10 Figure 6 Stages of the design process......................................................... 12 Figure 7 The basic concept governing impedance cardiography................. 14 Figure 8 Voltage Controlled Current Converter............................................ 15 Figure 9 The 4-Electrode arrangement ........................................................ 17 Figure 20 Schematic of the developed circuit for current injection and voltage pickup .......................................................................................................... 17 Figure 11 Input resistance and CMRR vs. Frequency curves...................... 19 Figure 13 XR-2206 oscillator ...................................................................... 20 Figure 14 Schematic showing the implemented circuit of our project. ......... 21 Figure 15 Block diagram showing procedure following signal detection. ..... 22 Figure 16 Amplitude modulation resulting from the followed procedure....... 23 Figure 17 Waveform to demonstrate the application operation.................... 24 Figure 18 Implemented circuit...................................................................... 25 Figure 19 Block diagram showing the transmission of the signal from the PCB to the pc ....................................................................................................... 27 Figure 110 The SCB-68 used to transmit the signal from the PCB onto the PC ..................................................................................................................... 27 Figure 20 DAQ-assistant VI module in LabView. It is set to sample at a rate of 100kHz......................................................................................................... 27 Figure 21 AM Demodulation module (Envelope Detection) on LabView...... 28 Figure 22 LabView module used to verify Envelope Detection is functioning effectively..................................................................................................... 29 Figure 23 Carrier signal ............................................................................... 29 Figure 24 Inputted signal ............................................................................. 30 Figure 25 Modulated signal.......................................................................... 30 Figure 26 Demodulated signal = Inputted signal.......................................... 30 Figure 27 Low pass filter on LabView .......................................................... 31 Figure 28 Configuration of the low pass filter............................................... 31 Figure 29 Differentiation of the demodulated and filtered signal .................. 32 Figure 30 Extraction of peak points onto a text file. ..................................... 32 Figure 31 LabView application for signal processing. .................................. 33 Figure 32 The developed user-interface of our application .......................... 33 Figure 33 Excel output for the Average Rate of Decrease = 0.021.............. 35 Figure 34 VB output for the Average Rate of Decrease = 0.021.................. 35 Figure 35 Electrode placement ................................................................... 36

v

List of Tables Table 1 Results produced upon testing the implemented circuit with a variable resistor ........................................................................................................ 26 Table 2 Date file to be tested on the VB application and Excel worksheet . 34 Table 3 Results of testing project on Mohamed. .......................................... 37 Table 4 Results of testing project on Mirna.................................................. 38

vi

1.0 Introduction 1.1 Background and Overview The Nebraska Medical Center at Clarkson and University Hospital reports that almost one of 2.5 deaths result from cardiovascular disease. Cardiovascular disease presents a major health and economic burden throughout the world and especially on developing countries. According to the Australian Institute of Health and Welfare, by year 2020 heart disease will have grown to become the leading health problem for the world. The ECG, stethoscope, X-ray, ultrasound and stress tests are all examples of diagnostic techniques used to examine heart functionality. Most of these devices yield data produced over and in a defined period of time and space (medical center, on a treadmill, in a physician’s clinic, etc…). We postulate that effectiveness of diagnosis could be improved by results confined neither in time nor space. In other words, we propose the development of a device capable of picking up an appropriate signal from the heart and picking it up and storing it on any personal computer for later analysis by the physician. It is hoped that the proposed diagnostic method might detect abnormalities in the cardiovascular system earlier than conventional methods. 1.2 Project Objectives The problem was suggested by our supervisor Dr. Nassir Sabah who specified that our device should be capable of: 1- Detecting a signal that is proportional to the rhythm and the strength of the heart beat. 2- Transmitting the signal to a personal computer for analysis.

1

3- Calculating the rate of decrease of the impedance cardiograph signal to normal following exercise. In the Interim Report we presented in Fall2005-2006 we had included the storage of the signal as one of the project’s objectives. We chose to modify this section for feasibility reasons. Research shows that the correlation between the rate and the strength of a heart beat as they return to normal following exercise and cardiovascular health have not been adequately documented. Thus we emphasize that the signal detected by our device should carry information about the strength of the heart beat as well as the heart rate. By strength of heart beat we are referring to the stroke volume of the heart defined as the amount of blood pumped by the heart into the body. The stroke volume is a well identified variable parameter correlated to the healthiness of the heart. As a result, the completion of our project will present a device that has the potential of investigating a new method for the detection of abnormalities in the cardiovascular system. 1.3 Report Organization Literature reviewed throughout the fall semester is presented in Chapter 2 of the report. It is followed by Chapter 3 which considers our project approach and describes the various stages of our design. Chapter 4 discusses the implementations of our design. Chapter 5 includes the results of runs we carried out on the implemented design. Chapter 6 describes the health, economic and safety considerations of our design. Our conclusions are finally presented in Chapter 7.

2

In summary, the report is organized as follows: Chapter 1.0 Introduction Chapter 2.0 Literature Review Chapter 3.0 Project Approach: Design and Analysis Chapter 4.0 Implementations Chapter 5.0 Results and Analysis Chapter 6.0 Health, Safety and Economic Considerations Chapter 7.0 Conclusion

3

2.0 Literature Review The purpose of this literature review chapter is to develop a solid background in the topic of our project. Such a background will allow us to submit a design whose every stage can be substantiated. We rely on consultations with Dr. Sabah, the AUB Library services and Internet resources in our search. 2.1 The Heart: General Anatomy and Cardiac Cycle This section includes a general overview of the functionality of the normal heart whose basic anatomy is illustrated in Figure 1.

Figure 1 Basic anatomy of the human heart. Retrieved and modified from http://www.cvphysiology.com

As shown in the figure above, the human heart is composed of four chambers (2 atria and 2 ventricles). Each atrium is separated from its corresponding ventricle by a valve. Valves are also present between the ventricles and their corresponding arteries. As a result, the human heart can be thought of as two separate pumping systems operating within a single organ. The right pump sends CO2 rich blood to the lungs whereas the left pump sends O2 rich blood to the body [10’]. The cardiac cycle, movement of blood through the heart, is divided into two parts: diastole and systole. During the diastole, blood from the body empties into the right atrium whereas blood from the lungs empties into the left atrium. 4

The pressure developed in the atria due to their filling causes the AV valves1 to open; blood moves to fill 80% of the ventricles. The following contraction of the atria will allow the rest of the blood to move into the ventricles. During systole, the ventricles contract and the rise in pressure in these chambers forces the AV valves to close and pulmonary and aortic valves to open delivering blood to the lungs and body, respectively. Once the blood leaves the heart and the pressure drops in the relaxing ventricles, the pulmonary and aortic valves close [10]. There are a number of signals that can be picked up from heart during its cardiac cycle. The following section presents an overview of each of these signals, one of which we will choose for our project. 2.2 Heart Signals This section of the report is further divided into 4 sections, each dealing with a different signal that can be picked up during the cardiac cycle. 2.2.1 Electrical Signal The most common and well-understood signal arising from the heart during the cardiac cycle is the electrical signal recorded as the electrocardiogram (ECG). The electrical signal arises from the polarization and depolarization of the cardiac muscle membrane-the basic concepts behind muscle contraction. A typical electrical signal wave recorded by an ECG is presented below along side an explanation of its various sections.

1

Valves separating the ventricles and atria.

5

Figure 2 A typical ECG recording of a single cardiac cycle. Retrieved and modified from http://www.guidant.com/

The P-wave occurs at the contraction of the atria, at the beginning of systole and end diastole. The QRS- complex occurs at the contraction of the ventricles, i.e. during systole. The T-wave occurs at the relaxation of the ventricle, i.e. during end systole [8]. The ECG signal if measured over a period of time will be capable of producing information about heart rate. Dr. Sabah pointed out that we should check whether or not the magnitude of the QRS complex varies in proportion to the heart beat strength. The Circulation Journal of the American Heart Association published an article by Simoons M.D and HugenHoltz M.D entitled ‘ECG Changes During and After Exercise’. The purpose of the authors was to analyze the magnitude and direction of time-normalized P, QRS and ST sections of the ECG during and after multistage exercise. The authors found that the magnitude and spatial orientation of the maximum QRS vector remains substantially constant during and after exercise [17]. The findings of the article presented above suggest that the electrical signal ofthe heart is not proportional to the strength of the heart beat. Although the heart rate can be measured from the ECG, the strength of the heart beat can not.

6

2.2.2 Sound Signal After the electrical signal we considered the sound signal of the heart. Listening to the heart is a diagnostic method used long before the development of the current stethoscope. The heart produces different sounds throughout its cycle and a trained physician is capable of identifying faults in the heart by listening to the sounds it produces. A typical sound signal is presented below along side an explanation of its various sections.

Figure 3 A typical sound recording of a single cardiac cycle. Retrieved and modified from http://www.ed4nurses.com

S1 represents the closing of the AV valves after the blood moves from atria to the ventricles. S2 represents the closing of the aortic and pulmonic valves after the blood leaves the ventricles. S1 is lower pitched and has a longer duration than S2 [18]. There are two types of stethoscopes present on the market today: acoustic stethoscopes and electronic stethoscopes [7]. An electronic stethoscope appeared to be the ideal solution for our design. Upon literature review and consultation with Dr. Sabah, however we found that electronic stethoscopes have their own draw-backs which can be summed up by the fact that they are very sensitive to impact, manipulation and ambient noise [7]. 7

At that point, Dr. Sabah suggested we try to look into other signals that can be picked up by heart. 2.2.3 Pressure Signal We came up with idea of using a pressure transducer placed on the chest above the heart. Changes in the size of the heart during the cardiac cycle would be perceived as changes in pressure by this transducer, converting the signal into an electrical one. After referring to several resources, however, we found that the literature available on pressure transducers, especially when used in the cardiovascular field, to be extremely limited. 2.2.4 Impedance Cardiography The final signal we considered was suggested by Dr. Sabah. In short, electrodes around the neck and around the waist cause a current of low magnitude and high frequency to flow through the major vessels connected to the heart. The resulting changes in impedance provide a rough estimate of beat-by-beat changes in cardiac output [12]. Impedance of the thorax can be considered to consist of two types of impedances: 1- time-invariant impedance due to tissues in the thorax and 2time-varying impedance due to time variations associated with the cardiac cycle [12]. Because of the complex structure of the thorax, the origin of the impedance signal has been extensively studied [15]. In 1952, M.D Bonjer attributed the impedance change to the changes in the size of both the heart and blood vessels [2]. A later study done by Patterson proposed that there could be up to four sources for the ICG (impedance cardiograph) signal: 1- ventricular

8

blood volume and velocity, 2- aortic blood volume, 3- Lung resistivity and 4Blood resistivity [21].

This study demonstrated that the use of band

electrodes for current injection and impedance measurements produces an impedance cardiograph signal that is a representation of all four mentioned factors, none dominating the rest. It further demonstrated that the use of spot electrodes placed near the walls of the heart generates an impedance signal of which 80% is contributed from the ventricular contraction. The figure below illustrates a typical signal picked up by impedance cardiography.

Figure 4 Typical impedance cardiography signal. Retrieved from [22].

As seen in Figure 4, the change in impedance Z is differentiated into dZ/dt; the signal used in the analysis. This signal takes into consideration only timevarying impedances and ignores the constant impedances of time-invariant tissues of the thorax. The ECG signal is normally recorded along side an impedance cardiograph for the identification of specific points on the dz/dt signals used for the calculation of the value of the stroke volume.

9

The variations present in Figure 4 can be explained as follows considering the use of spot electrodes placed near the anterior walls of the heart in order to amplify the contribution of ventricular contraction. As mentioned previously in 2.1, the contraction of ventricles is followed by the movement of blood from the ventricles into the aorta and the pulmonary artery. Since blood is a highly conductive tissue, its leaving the heart chambers leads to the increase shown in the recorded signal [21]. Since the signal depends on blood in the ventricles, Z decreases as the ventricles fill up and increases as they empty. Impedance cardiography has been employed in the medical field to calculate stroke volume and cardiac output using Kubicek’s model. Significant correlations (0.63-0.97) between the stroke volume measured with impedance cardiography and invasive clinical techniques exist [21]. Nevertheless, Kubicek’s model has been challenged in a number of papers and is not widely accepted in the medical community as a reliable method to calculating the stroke volume [16]. However, though the absolute value of the stroke volume calculated by Kubicek’s model is controversial, the relative changes of the stroke volume are valid [3]. In other words, the ICG is proving to be an invalid signal for the direct measurement of stroke volume but, using spot electrodes local events such as ventricular volume changes can be traced [21].

Figure 5 Diagrammatic representation of impedance cardiography. Retrieved from [11].

10

Changes in skin temperature and hydration have been suggested as a common cause to variations in ICG measurements. As mentioned earlier, the type of electrodes (band or spot) used is significant for determining the accuracy of the measured signal. Changes in skin temperature and hydration however have been found to have no significant effect on the impedance measurements when four electrodes are used – 2 for current injection and 2 for signal acquisition. The use of only two electrodes however does cause such changes [4].

11

3.0 Project Approach: Design and Analysis Based on the idea developed in the introduction, we represent the stages of our design in the figure below.

Figure 6 Stages of the design process

In the sections that follow, we describe the specifications of each stage in Figure 6. 3.1 Choice of Signal Chapter 2 included a literature review of all the signals that can be recorded from the heart. Among the signals discussed (electrical, sound, pressure and impedance) the impedance signal was selected for recording. We justify our choice by revisiting each signal. Electrical Signal: It was presented in Section 2.2 that the ECG is not proportional to the strength of the heart beat (i.e. to the stroke volume). This renders the signal not useful for our purpose. Sound Signal: Though the sound signal is proportional to the strength of the heart beat, the picking up of this signal presents a set of difficulties we may be 12

capable of avoiding by choosing another signal. Among these preventable difficulties is the sensitivity of the microphone required to pick up the signal as mentioned in Section 2.3. Though such sensitivity is needed to detect the sound signal of the heart, it also causes it to be highly susceptible to ambient noise. Movement of patient, breathing of patient, background noise, etc… present only a few examples of what contributes to ambient noise. Filtering might come up as a suggested solution to this noise. It however would require additional signal processing at this stage of the design process. Consulting with Dr. Sabah, he suggested we continue to look into other signals that might require less processing in the first stages. Pressure Signal: The lack of literature on the use of pressure transducers in cardiovascular applications as mentioned in Section 2.4 led us back to Dr. Sabah. Dr. Sabah explained that the pressure signal would not only be difficult to pick up but would yield results that are not very accurate. As a result we opted to overlook the pressure signal. Impedance Signal: The impedance signal produced due to changes in the volume of the tissues of the thorax was the final signal we looked into and the one we selected. As mentioned in Section 2.5, impedance cardiography – which measures impedance changes of the chest -

produces a signal

proportional to the stroke volume. Even though this signal’s accuracy in determining the exact value of the stroke volume has been refuted in a number of articles, the relative changes of the stroke volume it measures have been widely accepted.

13

Furthermore, the picking up of the impedance signal is expected to present fewer obstacles than other signals since it depends on the injection of a constant current followed by the measurement of the resulting voltages. Since the impedance signal provided solutions to problems presented by earlier visited signals, it was selected for recording; thus establishing the first stage of the design process. 3.2 ICG Circuit As shown in Figure 6, the second stage we set to complete was the development of a circuit for the measurement of impedance changes in the thorax. Figure 7 below illustrates the basic idea behind impedance cardiography.

Figure 7 The basic concept governing impedance cardiography. Results can be compared to the variations seen in Figure 5.

The constant current normally applied is an AC current of the frequency 1-100 KHz and amplitude 0.8-4mA [15], [21], [22], [3]. The use of AC current is preferred over DC current since a DC current allowed to flow through the skin for a period of time would cause electrolysis of the blood and chemical burns [6]. Furthermore, the threshold for perception

14

of alternating current depends on its frequency. Currents with frequencies between 1 and 100 KHz have a perception threshold greater than 10mA (rms) [5]. The applied current in impedance cardiography is thus safe to use if maintained at mentioned levels. 3.2.1 Voltage Controlled Current Source The main component of our circuit as discussed in the previous section is a current source which provides a constant current of around 1mA at a frequency between 1 and 100 kHz. The first circuit we looked into is a typical voltage to current converter circuit shown in Figure 8.

Iin

IL

Figure 8 Voltage Controlled Current Converter. Retrieved from [23].

We performed the necessary analysis on the circuit and found that, IL =

R 2 Vin R 1 R 3 A IC2

The gain on IC2 is given by the equation A IC2 = 1+

50 R3

In order to achieve a gain A IC2 = 1 , we chose R 3 = 10MΩ

15

Setting R 1 = R 2 = 1KΩ , it followed that the generated current should be I L = Vinx10 6 A. During the first stages following the construction of the presented circuit, we carried out the testing using AC and DC inputs. The results of the tests were highly irregular, irreproducible and inaccurate. 3.2.2 Implemented Design After our attempts with the voltage controlled current source on which we experimented for over a month, we worked out the design of a custom circuit compatible with our purpose. The first issue we identified was the range of values expected at the load. Cardiograph Impedance vs. Skin Impedance Unless designed to do otherwise, a circuit intended to detect cardiac impedance inevitably picks up skin impedance too. Documented studies record skin impedance to be in the order of Meg-ohms, under DC conditions, [9] and base cardiac impedance in the range of 25.15 ± 1.74 ohms [14]. These values indicate that in order to accurately record changes in cardiac impedance, skin impedance should be eliminated. Electrode Selection As mentioned earlier, band electrodes and spot electrodes yield different results upon use for current injection and voltage measurement. Because our objective is to capture a signal which is most representative of changes in the ventricles, our design makes use of spot electrodes [21]. Furthermore, the article demonstrates that the use of the 4-electrode arrangement instead of the 2-electrode arrangement allows the elimination of skin impedance from the measurement.

16

As a result, we opted to use the 4-electrode arrangement (2 for current injection and 2 for voltage recording) as shown in Figure 9. Such an arrangement eliminates complexities in our circuit and allows us to design for a load in the order of hundreds of ohms instead of mega ohms.

Figure 9 The 4-Electrode arrangement.

Description of the Designed Circuit

3 2 1

Figure 10 Schematic of the developed circuit for current injection and voltage pickup.

-

Section 1:

This section operates as a simple current source and provides our load with a constant current when R1 >> Z Load . It was previously demonstrated that Rload is in the order of hundred ohms with variations in the order of ohms; as a result, if R1 is chosen to be >50x Rload, the resultant current through the load would have a magnitude I load =

Vin . R1

17

The initial circuit we tested consisted of Section I only and a variable resistor in place of the load. Even though the circuit proved to be functional and accurate, when it was set to inject 1 mA current the picked up voltage was in the order of mV. Sections II and III were added to the circuit in order to avoid resolution complications when the picked up signal is introduced into the computer. -

Section II:

This section consists of the op-amp LM741 in the unity-gain configuration. The op-amp exhibits high input impedance and thus when placed in parallel with the body’s relatively low impedance is assumed to allow the following section to amplify the voltage with minimal current consumption. -

Section IV:

The final section, also built from the op-amp LM741, amplifies the picked up voltage 10 times before it is measured or entered into the personal computer. -

Component Values:

•

R1 > 50x RLoad Æ R1 = 3.3kΩ

•

For Vo =

-

ILoad :

R3 V = 10Vin R 2 in

Knowing that I Load =

Æ R 2 = 1KΩ and R 3 = 10KΩ

Vin , R 1 = 3.3KΩ and that we are designing for R1

~ 1mA pk-pk at a frequency 1KHz-100KHz; we found that V should I Load = in be set to 3V pk-pk . The frequency of Vin and thus that of ILoad was selected in accordance with the operation curves of the LM741 presented in Figure 11. 18

Figure 11 Input resistance and CMRR vs. Frequency curves

Taking into consideration: a- Safety/Health limitations b- High input resistance condition and c- High CMRR condition, we set Vin and thus ILoad to work at 10KHz. Oscillator Circuit of the Input Voltage In order to avoid the usage of a function generator or AC power supply and allow easier usability, we replaced Vin by an oscillator circuit based on the XR2206 IC shown in Figure 12.

19

Figure 12 XR-2206 oscillator.

Component Values: From the XR2206 datasheet we determined that: •

A typical value for R is 10KΩ when a signal of 10 KHz is to be generated (frequency was confirmed from the datasheet to cause 2f signal . We choose f s = 10f = 100KHz . Demodulation The procedure described above is depicted in the figure below.

Figure 15 Amplitude modulation resulting from the followed procedure.

Since ω c = max) Then Time(i) = 0 amp(i) = 0 End If End If End If Next i Close #1 Close #2 sizedivide = size 'Counts the number of zero entries in the arrays For i = 0 To size - 1 If amp(i) = 0 Then sizedivide = sizedivide - 1 End If Next i ReDim amp2(sizedivide) ReDim Time2(sizedivide) Dim j As Single i=0 j=0 'Removes the zero entries in the arrays Do While (i < size And j < sizedivide) If amp(i) 0 Then amp2(j) = amp(i) Time2(j) = Time(i) i=i+1 j=j+1 Else If amp(i) = 0 Then i=i+1 End If End If Loop

46

'Calculates the average rate of decrease Dim AmpDiff, TimeDiff, RateSum As Double For i = 0 To sizedivide - 2 AmpDiff = amp2(i) - amp2(i + 1) TimeDiff = Time2(i + 1) - Time2(i) Rate(i) = AmpDiff / TimeDiff Next i For i = 0 To sizedivide - 2 RateSum = RateSum + Rate(i) Next i Dim FinalRate As Double FinalRate = RateSum / sizedivide txtrod.Text = FormatNumber(FinalRate, 3) End Sub ' This exits the program Private Sub Exit2_Click() End End Sub ' This section connects the application to labview Private Sub ICG_Click() Shell ("ICG.exe") End Sub

47