Modern Pulse Meter for Traditional Indian Medicine

1

A

Seminar Report

On

―Modern Pulse meter for Traditional

Indian Medicine‖

Session: 2011-2012

Submitted in partial fulfillment of the requirement

For the award of the

Degree of

Bachelor of Technology In

Electronics & Communication Engineering

Submitted by Under The Supervision Of Dinesh Kumar Rathore Mrs. Vandana Talk

C.RN. 08/089 Mr. Deepak Bhatia

Department Of Electronics Engineering

RAJASTHAN TECHNICAL UNIVERSITY KOTA-324010

2

Acknowledgement

I take this opportunity to express my deep regards and sincere gratitude

for the valuable, expert guidance rendered to me by guides Mrs. Vandana Talk

& Mr. Deepak Bhatia, Electronics Engineering Department, University College

of Engineering, Rajasthan Technical University. Their guidance by going

through manuscript critically and holding informal discussion is grateful

acknowledge. They shall always be a constant source of inspiration for me.

My sincere thanks are to Respected Mr. Rajeev Gupta (Head of Electronics

Department) for providing necessary facility for my work.

My sincere heartfelt gratitude to my family whose prayers, best wishes, support,

concern and encouragement has been a constant source of inspiration to me.

Dinesh Kumar Rathore

C.RN.08/089

Batch: EC-1

3

University College of Engineering

Rajasthan Technical University

Certificate

This is to cert i fy that the seminar ent i t led ― Modern Pulsometer for

Tradi t ional Indian Medicine‖ is submit ted in part ial fulf i l lment of the

requirement for the award of B.Tech (Electronics & Communicat ion

Engineering) by Dinesh Kumar Rathore(08/089 ) of Final B.Tech.

Guides:

Mrs. Vandan Talk

Mr. Deepak Bhat ia

Department of Elect ronics Engineering

Universi ty Col lege of Engineering

Rajasthan Technical Universi ty Kota -324010

4

Table of Contents

University College of Engineering Rajasthan Technical University ...................................................... 3

Abstract ................................................................................................................................................... 8

CHAPTER 1 INTRODUCTION .......................................................................................................... 11

Pulse diagnosis .................................................................................................................................. 11

Mechanism of Pulse Formation ........................................................................................................ 11

Normal pulse rates ............................................................................................................................ 12

Calibration and validation of the arterial pulse measurement system ............................................... 14

Traditional Chinese Pulse Diagnosis ................................................................................................ 19

CHAPTER 2. POISEUILLE'S LAW; HEMODYNAMICS; PERIPHERAL PULSES; FORMATION

OF RADIAL PULSE; DISEASES OF ARTERIES; FACTORS AFFECTING THE PULSE

PRESSURE AND CLINICAL EXAMINATION OF PULSE ............................................................. 23

Poiseuille‘s Law and Its Application ................................................................................................ 23

A Novel Pulse Measurement System By Using Laser Triangulation And A CMOS Image Sensor 30

Pre Processing Of Pulse Series ........................................................................................................ 31

A. Irregular Pulses in the Pulse Series ....................................................................................... 31

B. Intricacy with Pulse Feature Identification ........................................................................... 32

CHAPTER 3.KNOWLEDGE MANAGEMENT SYSTEM ................................................................ 33

Electret Microphone As A Sensor .................................................................................................... 35

A Novel Pulse Measurement System ................................................................................................ 35

Wrist Pulse Characteristics and Acquisition ..................................................................................... 40

A. Wrist Pulse Characteristics .................................................................................................... 40

B. Wrist Pulse Acquisition .......................................................................................................... 41

CHAPTER 4. EXPERIMENTAL RESULTS ...................................................................................... 42

Outlier Pulse Identification and Pulse Feature Derivation ................................................................ 46

Dynamic Time Warping Algorithm ............................................................................................. 46

Nadi Tarangini Discussion ................................................................................................................ 47

A. Comparison with earlier systems. ............................................................................................ 47

B. Important Properties of Pulse Waveform. ................................................................................ 48

C. Varying Pressure ...................................................................................................................... 49

D. Variations with Age. ................................................................................................................ 50

5

T h e d e s i g n o f P u l s e C o l l e c t i n g T e r m i n a l ( P C T ) ...... 51

CHAPTER 5. CONCLUSION .............................................................................................................. 53

REFERENCES ..................................................................................................................................... 61

6

Table of Image

Figure 1 Mechanism of Pulse Formation the piston and tied into an artery. Suppose, if piston

is pushed forward towards the right into the tube the vessel distends locally. This occurs

simply because the blood is incompressible.. This immediate incompressible column of blood

produces ................................................................................................................................... 11

Figure 2Signal processing flowchart for the proposed arterial pulse measurement system .... 14

Figure 3The schematic drawing of the linearity calibration of the APM system. ................... 15

Figure 4Calibration for the pulsation amplitude measurement of the proposed APM system:

(a) the results for linearity calibration experiments and (b) the one standard deviation of

amplitude calibration experiments. .......................................................................................... 16

Figure 5The schematic drawing of the frequency calibration of the APM system. ................ 17

Figure 6Frequency calibration of the APM system: (1) the spot variation versus time (time

domain) and (2) the frequency spectrum analysis of the data in (1) using FFT method. ........ 18

Figure 7(a) Deep pulse (Cheng Mai) images, (b) Superficial pulse images .............................. 21

Figure 82 (a) Trend of superficial pulse images, (b) Trend of deep pulse images ................... 22

Figure 9Pulse obtained from in our setup. ............................................................................... 27

Figure 10 Hardware setup of our system. ................................................................................ 28

Figure 11Time domain features ............................................................................................... 29

Figure 12Wrist pulse series ...................................................................................................... 32

Figure 13First Derivative of Wrist pulse series ....................................................................... 32

Figure 14(b) Enlarged First Derivative of Wrist pulse series .................................................. 32

Figure 15Beat-to-Beat alteration.............................................................................................. 34

Figure 16Geometrical layout of the arterial pulse measurement system. ................................ 36

Figure 17The actual implementation of the proposed APM system........................................ 37

Figure 18The contours of laser spots after ET values of the CMOS image sensor were set to:

(1) 40 and (2) 1. ....................................................................................................................... 39

Figure 19Wrist pulse of a healthy person ................................................................................ 40

Figure 20Laser spot centers measured by APM system: (a) the original data of centroid

variation of laser spot center and (b) the data after enlarged amplitude scale and with

filtering. .................................................................................................................................... 43

7

Figure 21. Full spectrum analysis of the data in Figure 9: (a) the results of without using filter

and (b) the results of using filter. ............................................................................................. 44

Figure 22Illustration of palpation positions for pulse diagnosis used in traditional Chinese

medicine. .................................................................................................................................. 45

Figure 23Warping Path Calculation......................................................................................... 47

Figure 24Pulses as presented in selected earlier works, arranged in ascending published

date. ......................................................................................................................................... 47

Figure 25Pulse waveforms using our methodology on patients with various disorders. Only

one of the three acquired waveforms have been given for each type. .................................. 49

Figure 26The changes observed in the pulse waveform as the applied pressure increases

from Left to Right. .................................................................................................................... 50

Figure 27Pulse Collecting Terminal .......................................................................................... 51

Figure 28Photo of PCT in measurement and the sensors ....................................................... 52

Figure 29Pulse information measured at the Chun point on the left hand (small intestine) of

the tested subject: (a) 30 min before a meal (to be continued). ............................................... 54


the tested subject (continued): (b) 30 min after a meal. ........................................................... 56

Figure 31Pulse information for the Guan point on the left hand (liver) of the tested subject:

(a) before staying up late (to be continued). ............................................................................ 57

Figure 32Pulse information for the Guan point on the left hand (liver) of the tested

subject (continued): (b) after staying up late. .......................................................................... 58

8

Abstract Extensive research has been done to show that heart- beats are composed of the

interaction of many physiological components operating on different time scales, with

nonlinear and self-regulating nature. The more direct, and easily accessible manifestation of

the heartbeat is the pulse; however, it has not been studied any- where near as extensively.

In this paper, we establish the relevance of the multi- fractal formalism for the arterial

pulse, which has long been used as a fundamental tool for diagnosis in the Traditional Indian

Medicine, (Ayurveda). The finding of power-law correlations through defriended fluctuation

analysis indicates presence of scale-invariant, fractal structures in the pulse. These fractal

structures are then further established by self-affine cascades of beat-to-beat fluctuations

revealed by wavelet decomposition at different time scales. Finally, we investigate how this

pulse dynamics change with age, and disorder. The analytic tools we discuss may be used on

a wide range of physiological signals.

Abstract-Siddha is a natural treatment and the oldest medical system of healing in the

world. nadi which is a pulse-based diagnosis method which is the skill of feeling the pulse,

and IS known to dictate all the salient features of a human body. In this paper, we provide a

complete spectrum of details of our procedure for obtaining three different pulses based on

time series. This system contains a strain gauge equipped with a diaphragm element, a

transmitter and also an amplifier, a digitizer which quantifies the analog signals. The system

is designed with 16-bit accuracy in such a way that it provides no interference noise and no

external electronics. Compared with the prior systems like ECG, the system provides a

9

detailed classification of the nadi pulses which produces the waveforms with respect to

abnormalities.

The varying pressure given to the pulse analyzer classifies vadha, pitha, and kapha

based on the abnormalities captured from a single artery. The obtained output from this

module is been fed to the knowledge management system that identifies the diseases based

on body type. The designed system is being evaluated by siddha practitioners as a computer-

aided diagnostic tool

This paper presents a novel, non-invasive, non-contact system to measure pulse

waveforms of artery via applying laser triangulation method to detect skin surface

vibration. The proposed arterial pulsation measurement (APM) system chiefly consists of a

laser diode and a low cost complementary metal-oxide semiconductor (CMOS) image sensor.

Laser triangulation and centroid method are combined with the Fast Fourier Transform (FFT)

in this study. The shape and frequency of the arterial pulsation can be detected rapidly by

using our APM system. The relative variation of the pulse at different measurement points

near wrist joint is used as a prognostic guide in traditional Chinese medicine (TCM). An

extensive series of experiments was conducted to evaluate the performance of the designed

APM system. From experimental results, the pulse amplitude and frequency at the Chun

point (related to the small intestine) of left hand showed an obvious increase after having

food. In these cases, the peak to peak amplitudes and the frequencies of arterial pulsations

range from 38 to 48 µm and from 1.27 to 1.35 Hz, respectively. The height of arterial

pulsations on the area near wrist joint can be estimated with a resolution of better than 4 µm.

This research demonstrates that applying a CMOS image sensor in designing a non-

contact, portable, easy-to-use, low cost pulse measurement system is feasible. Also, the

10

designed APM system is well suited for evaluating and pre-diagnosing the health of a human

being in TCM clinical practice.

Wrist pulse analysis for identification of health status is found in Ancient Indian as

well as Chinese literature. The pre- processing of wrist pulse is necessary to remove

outlier pulses and fluctuations prior to the analysis of pulse pressure signal. This paper

discusses the identification of irregular pulses present in the pulse series and intricacies

associated with the extraction of time domain pulse features. An approach of Dynamic

Time Warping (DTW) has been utilized for the identification of outlier pulses in the wrist

pulse series. The ambiguity present in the identification of pulse features is resolved with

the help of first derivative of Ensemble Average of wrist pulse series. An algorithm for

detecting tidal and dichotic notch in individual wrist pulse segment is proposed.

11

CHAPTER 1 INTRODUCTION

Pulse diagnosis In medicine, one's pulse represents the tactile arterial palpation of the heartbeat by trained

fingertips. The pulse may be palpated in any place that allows an artery to be compressed against a

bone, such as at the neck (carotid artery), at the wrist (radial artery), behind the knee (popliteal

artery), on the inside of the elbow (brachial artery), and near the ankle joint (posterior tibial artery).

The pulse can also be measured by listening to the heart beat directly (auscultation), traditionally

using a stethoscope.

Mechanism of Pulse Formation

The mechanism responsible for the generation of pulse can be better explained

through the interpretation of diagrams. As shown in the diagram (Fig. 9), there is a tube

fastened with

Figure 1 Mechanism of Pulse Formation the piston and tied into an artery. Suppose, if piston

is pushed forward towards the right into the tube the vessel distends locally. This occurs

simply because the blood is incompressible.. This immediate incompressible column of blood

produces

local increase of pressure that has little effect on the advancing piston, which is in fact a mass

of incompressible blood. Instead the next section of the artery is stretched so that a wave of

pressure travels along the vessel wall without involving actual transmission of blood.

Similarly the largest arteries, such as the aortic arch, innominate and sub Clarian arteries are

highly distensible and so, the expelled blood is really accommodated in them. This wave of

distension is transmitted along the arteries and what we feel as the pulse. This means that the

12

pulse is actually a wave set-up in the walls of the vessels by the ―systole of the ventricle and

it is not due to the passage of blood along the arteries (Fig. 10).

In the figure, it has been shown that sudden ejection of large quantity of fluid into a

distensible tube, such at point A causes bulging of the tube at the point of ejection, and the

pressure in the area of the tube rises, immediately after the vessel wall is distended at point A

the elevated pressure at this point forces a small amount of fluid along the vessel, as

illustrated by the arrows. This causes the vessel wall to distend at point B, while-at the same

time the vessel begins shrinking at point A. Consequently, the vessel wall at point B

continues to distend and the pressure continues to rise while the vessel at point A continues to

shrink and the pressure continues to fall. After point B is well distended, the elevated

pressure at this point causes fluid to flow to point C, distending the vessel at point C and

relaxing the vessel wall at point B. This process is repeated by small increments along the

entire length of the vessel until the pressure wave reaches the end of the vessel. So it is very

clear that the elastic property of the arterial wall takes the part in the generation and

transmission of the pressure wave of the pulse.

Normal pulse rates

Normal pulse rates at rest, in beats per minute (BPM):[2]

newborn

(0-30 days

old)

infants (1 -

11 months)

children (1

- 10 years)

children over 10

years & adults,

including

seniors

well-rained

adult

athletes

regular

adults

70–198 80–120 70–130 60–100 40–60 60–100[3]

The pulse rate can be used to check overall heart health and fitness level. Generally lower is

better, but bradycardias can be dangerous. Symptoms of a dangerously slow heartbeat include

weakness, loss of energy and fainting

Peterson (1952 and 1954), the first scientist who raised the question against the

problem that, if there would be an appreciable lag time between the pressure pulse and the f

aid displacement or movement of pulse wave from segment to segmeqjt. In his article first

published, he introduced the idea of ‗large lag.‘ Spencer (1958) stated that in the upper aorta

http://en.wikipedia.org/w/index.php?title=Pulse&printable=yes#cite_note-1

http://en.wikipedia.org/w/index.php?title=Pulse&printable=yes#cite_note-NHS-AF-2

http://en.wikipedia.org/wiki/Bradycardia

13

pressure and flow start together, but it was not supported by any figure. Again Remington

(1963) hypothesized and supported the view of Peterson that there appears to be a true lag of

about 5 millisecond, and further stated that after initial delay between vertical and ascending

aorta, the pressure pulse seems to be propagated at a steady rate through the aorta. Gregg

(1967) states that almost simultaneously the acceleration of blood occurs with the rise of

pressure pulse in the aorta and arterial tree. This is to be naturally expected since the latter

occasions the former. However, whereas the pressure wave travels in the term of meters per

second i.e. 7 meters per second the movement of the red cells or plasma is much slower at 10

to 20 cm. per second. Thus although the pressure wave or pulse may reach the vessel of the

foot in 0.2 second, it requires several heart beats for the blood which enters the ascending

aorta to reach the foot vessel. This is because the speed of the blood depends on such factors

as the blood pressure gradient, viscosity and cross sectional area of the blood vessel. So far as

rate of flow versus cross section is concerned; it varies inversely as the total cross section of

the vascular bed. And as the sum of the cross sectional area is concerned, it increases

progressively as the arterial system divides and redevised. So naturally the velocity of blood

in aorta is 0.8-1.0 meter per second and about 0.5-1.0 millimeter per second in capillaries.

As regards velocity of the ‗pulse it is determined almost entirely by the elasticity of

the wall. Also, it would be appreciable here to throw light on the velocity of the pulse wave.

Bramwell and Hill (1922) demonstrated that the velocity of the pulse increases with the age

closely following the rise in blood pressure, which occurs as one grows older. In the same

year Hickson McSwiney showed that the pulse wave velocity varied with respiration.

Bramwell et al (1923) pointed out that pulse wave velocity varied with the pressure within

the artery and with the extensibility of the arterial wall. Bazette and Dreyer ( 1923 )

demonstrated that pulse wave velocity was slower in larger vessels—4 meter per second in

artery as brachial compared with 8.5 meters per second or more between the elbow and the

wrist. Starling (1968) says that the pulse wave velocity is independent of the output of the

heart with each stroke and so is a better indication of the elastic behavior of the arteries than

is the pulse wave.

Here one should be very clear in his mind as also foregoing discussion under the head

‗mechanism of pulse formation‘ proves that the phenomena taking part in the formation of

pulse chiefly concern with the mechanism of hemodynamics of cardiovascular system, and

this mechanism can be explained in the light of Poiseuille‘s law—a physical law concerns

14

with the branch of hydrodynamics which shows the interrelation between the pressure, flow

and the resistance in a rigid tube.

Calibration and validation of the arterial pulse measurement system

The image data were recorded by a CMOS image sensor and transmitted to a personal

computer for further analysis. The images were saved in bitmap format for later image

Figure 2Signal processing flowchart for the proposed arterial pulse measurement system

processing and frequency Spectrum analysis. Fifteen frames per second were captured

by the CMOS image sensor. The pictures captured by the CMOS image sensor lasted for 10

or 20 seconds in each measurement. The more frames are recorded; the better resolution of

the image data can be achieved.

The calibration of the APM system was very straightforward and easy. A precise

translator was used to calibrate the linearity of the APM system. A white paper was placed

above the translator as a reference panel, and then the shift of the laser spot can be calibrated

by adjusting the elevation of the translator step-by-step. The adopted step size of the

translator that can mimic pulsation amplitude measurement is 20 μm. The schematic drawing

of the linearity calibration of the APM system is shown in Fig. 5. The amplitude calibration

15

of pulsation height measured by the APM system was conducted on an isolated optical table.

The results for the linearity calibration experiments are shown in Fig. 6(a).

The one standard deviation (1σ) is obtained by conducting the measurements 30

times, and the results are as shown in Fig. 6(b). Examining Fig. 6(b) indicates that the one

standard deviation of the APM system achieved is less than 0.04 pixels, i.e., approximately

equivalent to 3.8 μm. This fact demonstrates that the APM system provides pretty good

performance on measurement stability.

The frequency calibration was conducted by comparing the experimental results with

the drumhead movements of a loudspeaker driven by a high-precision function generator

(LFG-1300, Leader, Inc.). In this experiment, we used a function generator with a specific

frequency to drive the loudspeaker. So, the drumhead variation of the loudspeaker referred to

a standard frequency can be provided as the

Figure 3The schematic drawing of the linearity calibration of the APM system.

16

Figure 4Calibration for the pulsation amplitude measurement of the proposed APM system:

(a) the results for linearity calibration experiments and (b) the one standard deviation of

amplitude calibration experiments.

reference of the frequency calibration of our APM system. The schematic drawing of the

frequency calibration of the APM system is shown in Fig. 7. For the loudspeaker operated at

1.0 Hz, the amplitude variation of the loudspeaker drumhead movements in the time domain

and its frequency spectrum are shown in Fig. 8(a) and 8(b), respectively. The calibration

results, from 0.6 to 2.0 Hz, are shown in Table 1. In this test, the frame capture rate of CMOS

image sensor of the APM system is set

17

Figure 5The schematic drawing of the frequency calibration of the APM system.

to 15 frames/sec. The accuracy of pulse measurement could reach 2.5%, i.e. the error

in the pulse rate would be less than 1.5 pulses per minute.

The accuracy of the frequency measurement of our APM system was also validated

by comparing the experimental results with the data obtained from a blood pressure monitor

(Model No.: OS-512, OSIM, Inc.). Ten healthy volunteers participated in this study. The

experimental results are summarized in Table 2. The difference between the measured results

obtained from the APM system and the blood pressure monitor was no more than 2.8%, i.e. 2

pulses/min.

Table 1. Frequency calibration of the APM system from 0.6 − 2.0 Hz with a function

generator and a loudspeaker.

Standard frequency

(Hz)

Standard frequency

(Hz)

Frequency obtained

from the APM system

Error (%)

Frequency obtained from the APM

system system

Erro

r (%) 0.6 0.585 2.5

0.7 0.703 0.4

0.8 0.820 2.5

0.9 0.879 2.3

1.0 0.995 0.4

1.1 1.113 1.2

1.2 1.230 2.5

1.3 1.288 0.9

1.4 1.406 0.4

1.5 1.523 1.5

1.6 1.581 1.2

1.7 1.698 0.2

18

1.8 1.816 0.9

1.9 1.874 1.4

2.0 1.991 0.4

Figure 6Frequency calibration of the APM system: (1) the spot variation versus time (time

domain) and (2) the frequency spectrum analysis of the data in (1) using FFT method.

19

The position variations of the laser spot centers measured in the time domain are

illustrated by the thin line in Fig. 9(a). This variation curve contains a full region of

information associated with vibration frequencies and amplitudes of the arterial pulses,

breathings, hand movements, and

Table 2 Comparison of the frequency measurements made by the APM system and a

blood pressure monitor (Model No.: OS-512, OSIM, Inc.).

Volunteer No.

OSIM OS-512 (pulse

rate)

APM system (pulse

rate)

Error

(%) 1

Volunteer No

69 68 −1.

4 2 76 77 1.3

3 75 76 1.3

4 84 86 2.4

5 90 92 2.2

6 82 81 −1.

2 7 78 78 0.0

8 72 74 2.8

9 94 96 2.1

10 78 78 0.0

involuntary body tremors. Generally, the human pulse is about 0.7 to 2 Hz. The

measurement data after being filtered with a band pass filter is depicted by a thick line as

shown in Fig. (a). The amplitude of vibration of laser spot center (in terms of pixels) has been

enlarged in Fig. (b). It can be seen that the change of the center of the laser spot is

proportional to the change in distance to the skin as calculated in Eq.(1). In Fig.(b), the peak

to peak values of the spot center variation curve are almost fallen within 0.6 pixels, i.e., the

arterial pulse amplitude is approximately equal to 57 μm. The full spectrum in frequency

domain for Fig. (a) is shown in Fig. (A). An inspection of both Fig. 9(a) and Fig. 10(a)

indicates that there was low frequency vibration caused by noise, hand movements, and

breathings.

Traditional Chinese Pulse Diagnosis

TCPD, one of the four diagnostic methods of TCM, is to judge disease by means of

fingertips palpating patient‘ pulse image shown in the superficial arteries. Many western

people may consider that pulse waveform is just the same as electrocardiogram (ECG) and

the patient‘s ECG analysis is enough. The signal of ECG acquired through several electrodes

20

only reflects the bioelectrical information of body. Having analyzed the pressure fluctuation

signal of pulse, doctors can detect and predict some symptoms that ECG cannot. TCPD can

not only deduce the positions and degree of pathological changes, but also is a convenient,

inexpensive, painless, bloodless, noninvasive and non-side effect method promoteded by

U.N. 2

The substance of pulse is the blood and the power of pulse is the heart. The heart

pumps and blood into all parts of the body through vessels and then the blood enter viscera

inward and reach limbs & skin outwards incessantly. Besides, the blood circulation also

depends on other viscera, which coordinates the heart. The lung meets all vessels and the

blood circulation all over the body should converge into the lung; the liver stores blood and is

in charge of its conducting; the kidney stores essence. Thus through the vessels, all visceral

state and disease condition can be understood by means of pulse diagnosis. 3,4 Pulse

diagnosis is to palpate pulses with fingertips and then to understand and judge the disease

condition through the process of diagnostician‘s comprehension. It also named pulse-

palpating, pulse-feeling, pulse-touching, pulse-reading, pulse examination or pulse taking.

Pulse taking is the common word. To sum up the ancient Chinese Medicine, the significances

of TCPD research today are as follows:

1. The physical examinations for the people of special careers such as students, pilots,

athletes and some others, especially for the workers in chemical plants;

2. The researches of drug‘ s functions and effects on blood vessels & heart;

3. The monitoring of patients, pregnant women, fetus and so on;

4. The important reference for the doctors to recognize the exterior and interior of

disease, to judge the deficiency and excess, to ascertain nature of disease, to identify

the cause of disease, to predict the prognosis and to inspect the disease mechanism;

5. The medical education and training for medicos;

6. The researches on the circulation system, nerve system, body fluid regulation, the

emotions and so on; 5

21

7. The researches on fitness and exercise (checking the effects and revising the exercise

plan);

8. The surveying of psychology and the detecting of liar; 6-9

Figure 7(a) Deep pulse (Cheng Mai) images, (b) Superficial pulse images

Since ancient times, doctors have been paying great attention to pulse taking and have

accumulated rich experiences. Taking pulse in TCPD, involving counting the number of

beats and identifying its form & pattern, does not just mean the identifying of pulse

waveform as the researcher of modern medicine did. 10 It should be borne in mind that each

of the types doesn‘t represent just one aspect of a given pulse. For example, floating and

sinking describe the depth; slow and rapid describe the rate, whereas surging and fine

22

describe the size of the pulse. Actually, these parameters occur in combinations. In most

cases, a patient‘s pulse is described with a composite term such as floating, slippery, and

rapid, or sinking, wiry, and thin.

Figure 82 (a) Trend of superficial pulse images, (b) Trend of deep pulse images

According to the theory of TCPD, we use different pressure to acquire the pulse

image and then judge the pulse whether floating or sinking, whether excess or deficiency and

so on. Pulse shape varies with pressures. When the pulse waveform amplitude is the highest

among those pulse

23

CHAPTER 2. POISEUILLE'S LAW; HEMODYNAMICS;

PERIPHERAL PULSES; FORMATION OF RADIAL PULSE;

DISEASES OF ARTERIES; FACTORS AFFECTING THE PULSE

PRESSURE AND CLINICAL EXAMINATION OF PULSE

Poiseuille’s Law and Its Application

In the branch of hemodynamics it was the Poiseuille (1884) and Hagen ( 1939 ) who

invented that pressure, flow and resistance have definite relation to each. Other and showed

in vitro that volume ( F ) of a fluid flowing through a capillary tube increase with the pressure

head (P), and decreases with resistance (R) to flow as indicated by the equation which

After extension can be written in the form of F = (Px-P2).

Pa is the pressure at a downstream point in the tube; Thus P1-P2 represents the

pressure head or pressure gradient, is 3.1416. The is important in the formula because we are

dealing with a cylindrical tube. The 8 arose in the process of Hagen‘s integration. V is fluid

viscosity, finally, after further dilatation of the above formula it was established that flow

varies directly and resistance inversely with the fourth power of the radius. And flow varies

inversely and resistance directly with the viscosity of the fluid. Thus considering only the

fourth power of the radius, if other things such as viscosity and pressure remain unchanged, a

decrease to half a radius will actually decrease the flow to a sixteenth of the original value. In

other words the resistance to flow is increased sixteen times.

While applying Poiseuille‘s law to the circulating blood in vivo, we should bear in our

mind that it is mostly applicable in vitro to the Newtonian fluid flow through a rigid tube

under steady pressure head. By definition Newtonian fluid means a simple viscous fluid

which viscosity is unaffected by flow rates and remains constant at different rates of

streamlined flow. So, in vivo Poiseuille‘s law have but restricted application. The reason-is

that in living body the heart beats rhythmically and ejects blood physically into a system of

elastic tubes and moreover the blood is not a perfect fluid but a two phase system of liquid

and cells. Thus we may say that the physical principles are of value only when used as an aid

24

to understanding what goes on in the body rather than as an aid in themselves. In the living

subject, so far as steady head of the pressure, which fulfills one of the criteria of Poiseuille

observation, is concerned it is achieved approximately by the combined effect of elasticity of

larger vessels and resistance offered by the arterioles. And thus the pulsatile ejection of the

heart is converted into a steady outflow. Yet there are some factors which would seem to be

immediately complicating to help in achieving this goal.

Considering about Newtonian fluid in relation to blood we know that it is made of

both liquid and. cells and has anomalous viscosity as observed in vitro. In vitro when

measured at moderate or high shear rates, blood shows a relative viscosity of 4-5. And when

measurements are made at very low shear rates the viscosity is greatly increased. But the

condition in vivo is rather different, because variation of temperature affects the viscosity.

Cooling raises the viscosity of blood. There are other certain metabolic conditions which

increases the viscosity of blood and they are: (1) large amount of fat in the blood, ( 2 )

polycythemia, (3) acidosis. (4).hyperglycemia, and (5) hypocalcaemia. But it is to be noted

here that in normal conditions throughout most of the physiological range of flow, viscosity

does not change appreciably with flow. In general, viscosity of blood in vivo is lower and in

capillaries unaffected by temperature changes, it is similar to that of plasma. And the blood

behaves approximately as a Newtonian fluid. Haynes (1957) also proved by his experiment

that in the physiologic range of blood flow, blood behaves as if it were a Newtonian fluid.

Thus we see that in vivo the criteria such as ‗steady pressure head‘ and ‗Newtonian fluid‘ of

Poiseuille experiment is mostly satisfied at the normal physiologic range.

From foregoing discussion it is now evident that Poiseuille‘s law may also be

applicable in vivo. And its application to the circulation depends on whether resistance is

independent of pressure and flow. In the vitro the resistance would be constant because both

the viscosity and the geometry of the tube were unchanged and did not change with the rate

of flow of pressure. But in circulating system peripheral resistance is a measure of the totality

of the factors affecting blood flow. These include change in apparent viscosity (which is

known to occur with increase in perfusion pressure, occasioned by the movement of the red

cells to form a central axial rod. Also, perfusion pressure in relation to blood flow

numerically means the mean intraluminal pressure at the arterial end minus the mean pressure

at the venous end). Besides viscosity other factors are the existence of stream-lined versus

turbulent flow, the length of the vessels and the cross sectional area of the blood vessels

25

which is determined by the extravascular pressure provided by surrounding tissue; by

mechanical dilatation with perfuming pressure; by opening of new capillaries and vessels

with change in metabolism and with rising perfusion pressure; and by active change in the

state of contraction of the muscular walls through vasomotor nerves, hum oral substances and

metabolic products.

We know that in the body arterioles play major role in causing the peripheral

resistance and the smooth muscle in their media having the quality of distensibility also

reacts actively to stretch by contracting. In the case of vessels such as arteries and veins

which are having smaller resistance, given pressures head would increase their internal radius

and would correspondingly cause a raised flow at that pressure. This is not the situation in

arterioles due to their myogenic property. So the pressure flow relationship in distensible

vessels and the vessels having myogenic element will be different. And therefore, flow-

pressure curves obtained with blood in vascular beds are quite different from those obtained

in the artificial system used in Poiseuille‘s experiment. These differences are described

diagrammatically

In (A) when the tube is rigid and the viscosity and the length changes are ignored the

relation between pressure and flow is linear as would be defined by Poiseuille‘s formula in

vitro. But in (B) where the tube is elastic such as arteries the increase in flow as a result of a

rising pressure head is. Greater than in A. In it the initially collapsed vessels are distensible.

In B it is obvious that the vessels cannot be ultimately distensible because the adventitia etc.

contain inextensible fibro-collagenous element. So at high pressure the flow would again be

linear. In (C) the myogenic contractile response to stretch is depicted as affecting the ‗elastic‘

effects exerted by the pressure rise. The curve is concave to the pressure axis. Curve (D)

shows the result when the myogenic elements of the wall even exceed the ‗elastic‘ effects of

a raised pressure. Superimposed can the physical features are the additional effects caused by

an increased sympathetic vasoconstrictor discharge to the resistant vessels. Such discharge

constricts the lumen of the innervated vessels mainly the arterioles and thereby decreases the

flow of a given pressure head.

Wright ( 1971 ) says that two possible explanations may be given for the deviations of

the curves and they may be ( 1 ) the geometric factor of Poiseuille‘s law is not constant but

varies with distensibility of the blood vessels, and ( 2 ) blood has anomalous viscosity.

26

As far as role of anomalous viscosity in pressure-flow relationship is concerned,

apparently it is a minor factor in large' (non-capillary) blood vessels. As it has also been

pointed out already that throughout most of the physiological range of flow blood behaves as

if it were Newtonian fluid. Now the only remaining factor in consideration to curve deviation

is the distensibility of the blood vessels. In this context it can be said that while experimental

evaluation is difficult, it does appear that vessel distensibility is the major factor responsible

for the deviation of in vivo pressure-flow curves from Poiseuille‘s law. This in mrn is related

to the active tension or contraction of smooth muscle in the vessel walls and collagenous

fibers in the architecture of the vessels.

In summarizing the entire discussion, we can say that the resistance and distensibility

of the vessels have definite relations to the flow of the blood throughout the body; and as it

has been shown in the diagram of pressure-flow relationship that the factor of overwhelming

importance is the dispensability of the blood vessels for the shape of aptual flow-pressure

curve in vascular bed.

In this section, we describe the salient features of traditional Indian medicine. The

pulses are sensed by the pressure exerted by the artery. The nadi pulses are sensed by the

fingertip. The pulses obtained are very minute to analyze and very challenging too.

27

For this, the experiments were carried out using three pressure sensors, „Millivolt

Output Medium Pressure Sensor ‟[Mouser Electronics, Inc.] With tiny diaphragm at the

center, which has the pressure range of 0–4 inch H2O‟?

Figure 9Pulse obtained from in our setup.

As shown in the above Fig,, The three location pulses namely vata, pitha, kapha are

sensed by the set of three pressure transducers which is mounted on the wrist. The digitized

output from the pressure sensing element is experienced by using the 16-bit multifunction

data acquisition card NI USB-6210 (National Instruments, TX, and USA). It has been

interfaced with the personal Computer .The electrical signal is proportional to the pressure

experienced, in differential modes.

28

Figure 10 Hardware setup of our system.

A sampling rate of 500 Hz (higher than the Nyquist criteria) is been captured for a

length of time which is predetermined. The strain gauge transducer is approximately of

29

Dimension 1cm×1cm. A flexible diaphragm is placed at the center, which is a force-

gathering element.

The deformation is being done by arterial pressure waves. Three constant resistors are

present in a Wheatstone bridge circuit along with the variable resistor transforms the strain

output into an electrical signal which is proportional to i.e. micro machine is been utilized

by the sensor gives a stress concentration to the enhanced structure that provides linear output

to the pressure measured.

(C)DE noising Using Wavelets

Figure 11Time domain features

To provide conditioned and linearized signal a standard transmitter and an amplifier

with 4-20mA is used. The 4-20mA is converted to 2-10 V through a resistor of 500 ohms

while connecting to 16-bit digitizer. The span and zero adjustments are calibrated with zero

atmospheric pressure is provided to adjust the zero. The output from the zero pressure is

4mA. The output signal is delivered as an amplified signal which has been adjusted by the

span. Due to explicit and implicit electrical and electronic noises, the data obtained are

corrupted. In our system the noise level is negligible because of proper shielding.

30

A Novel Pulse Measurement System By Using Laser Triangulation And A CMOS Image

Sensor

The waveform of arterial pulsation is considered a fundamental indicator for the

diagnosis of cardiovascular disease, which can guide therapeutic decisions in complex

clinical situations. Abnormalities of the waveform shape and frequency of the arterial

palpitation are indicators of certain cardiovascular disorders. Thus, how to distinguish arterial

pulse waveforms without distortion has become an important issue in biomedical signal

processing. In addition, pulse diagnosis is one of four kinds of diagnostic methods used in

TCM clinical practice to determine the physiological condition of the patients [2]. The most

commonly used clinical methods to measure the behavior of the pulse include the

stethoscope, sphygmomanometer, and Doppler-based instrumentation. Recently, some

practitioners in TCM use a pulse diagnosis machine or other device to record changes in the

pulse [3]. A set of three pressure transducers for sensing the pulses at three locations has

already been developed. Lu et al. analyzed the harmonics of the frequency spectrum of

arterial pulse waves and correlated some illness conditions to certain harmonics. Hong et al.

described a non-touch pulse measurement method based on optical interferometer [6], which

could estimate skin vibration velocity. However, the devices used in the above-mentioned

studies could either interfere with the measurement results because of making contact with

the body, or were more costly because of having additional sensors for probing multiple

points. An optical non-contacting technology, which is based on optical triangulation, is

proposed in this study. Laser triangulation is a well-known method in thickness and contour

measurement, and has been applied to many industrial fields. It was also used to examine the

vertical movements of the vocal folds during phonation.

Laser triangulation is normally used in conjunction with light centroid measuring

devices, e.g. position sensing detectors (PSDs) and charge coupled devices (CCDs). Since the

manufacturing technology behind the CMOS image sensors has now been advanced

sufficiently to achieve good stability and low cost, the CMOS image sensors have become

increasingly significant in industrial optical sensors. The CMOS image sensors possess

several programmable features including electronic exposure (ET) control, continuous or

single frame capture, and progressive or interlaced scanning modes. The first of these

features is very important to our experiments, especially for reducing noise and locating

measurement point of laser spot.

31

In general, the relationship between pulse waves and physiological or healthy

conditions of the tested subjects is quite complex. Such relationship might exhibit nonlinear

characteristic and might also vary person to person due to characteristics of the artery, deep

or shallow, healthy or hardened, etc. Basically, the more pulse waveforms are obtained, the

more information including pulse rates and harmonics related to the diagnosis in TCM can be

achieved. In TCM clinical practice, an experienced TCM physician can do the pulse

diagnosis by palpation treatment conducted on multiple measurement points (i.e., Chun,

Guan, and Chy points) to determine the relationships between the organs health and the wave

patterns of pulses. These facts have been demonstrated in many literatures [13-14]. Of

course, a more complex and accurate arterial pulse measurement system is necessary if more

factors are taking into consideration in TCM clinical practice.

This paper describes the design of the arterial pulsation measurement (APM) system,

and presents the results of tests conducted to verify the pulse measurement accuracy. The

pulsation rate was derived from the frequency spectrum of the laser spot vibration, and

showed great consistence with data taken from loudspeaker movement driven by a function

generator at a specific frequency. Frequency validation was also conducted by comparing

the experimental results with data obtained from a standard blood pressure monitor. The

amplitude and frequency variation at each point measured on the tested subject‘s wrist is an

important symptom for some illness during the patient‘s medical examinations in TCM

clinical practice. The evaluation of pulse variation gives us some valuable information

concerning about the tested subjects‘ health.

Pre Processing Of Pulse Series

A. Irregular Pulses in the Pulse Series

Due to physiology of the human body, wrist pulse series shows beat to beat variations.

For the variability study of pulse parameters like pulse peak amplitude, tidal wave

amplitude, dichotic wave amplitude, their timings and pulse rate, they must be derived

from each pulse segment. But due to motion artifacts in the pulse acquisition, few

pulses lose their character and hence it becomes difficult to assign pulse features

with them. Such outlier pulses need to be removed prior to further analysis as they have

adverse effects on the analysis. An example of such pulse series is shown in Fig. 2 wherein

the fifth pulse is losing its character.

32

Figure 12Wrist pulse series

B. Intricacy with Pulse Feature Identification

Since wrist pulse is a combination of forward wave, generated by the heart

and a reflected wave from the peripheral organs, it undergoes various slope changes

in the diastole phase of the pulse. The first derivative of the pulse

Figure 13First Derivative of Wrist pulse series

Figure 14(b) Enlarged First Derivative of Wrist pulse series

33

CHAPTER 3.KNOWLEDGE MANAGEMENT SYSTEM In the siddha system of medicines the nadi is sensed at the wrist of the patient

with varying pressure in different amplitudes and energies etc. This was correlated

with the body conditions. The pulse vata, pitta and kapha sensed by the sensors are sent

to the system for further remedies. The knowledge managemenet system has the

classification of diseases according to the body type.

The vata, pita and kapha type diseases are stored in the databases. The vaiability in

the pulse indicates that the dominant pulse is analysed and the rectification for which the

symptoms are matched.

These matching are done with the knowledge management system. The symptoms

have been identified by the knowledge management system which in turn gives the

rectification and the food items to be followed by the patient are displayed. This can

further be send to the centralized system where the clinicians can verify and the

prescriptions are generated. Variations in pressures and the mapping to KMS As per our

previous observations done the changes in the pulse waveforms for 31 patients in our

database based on their age. They are classified in age groups „below 25‟, ―25–50‟ and

‟above 50‟. We have observed that the „below 25‟ pulses are more dominant in secondary

peaks. ―25–50‟ group is relatively stable, while older pulses are not regular in nature. We

have given samples of healthy nadi of 3 groups in of both the hands. As can be seen, the

pulse duration goes up (rate decreases) as the age increases and also the patterns a re

d i f fe ren t . Again, t hese observat ions are consistent with [4], [15]; the changes are

indicative of an age-dependent reduction in large artery (capacitive) compliance and in

small artery (oscillatory or reflective) compliance.

However, we consider all above findings to be only a preliminary suggestion to

investigate further with much larger dataset. While applying machine learning algorithms

on pulse dataset for identifying pulse type, we need to consider all the above dimensions, i.e.

age, pressure applied, disorder and many more given in the literature.

Our system observes the pulses in all age groups, finds the variations and locates the

correct disorders based on the symptoms observed. This can be accurate compared with the

prior work done. Disorders may differ from person to person which can be captured

through the analyzer and the disease deification is done by our knowledge management

system.

34

Our system analyzes the waveforms obtained. Sample waveforms are obtained from

the left hand artery of a patient. Section III shows the wave forms obtained which has been

given in Fig.3. The respective doshas vata, pitha and kapha are obtained from three different

pressure transducers. This has been represented in three different colors. The sampling

frequency is 500 Hz. The zoomed waveform is shown in Fig 3(b). It gives us the number of

points per pulse cycle. It has been represented in the form of secondary peaks.

The extreme high component is used to remove the noise by wavelet demising . So

that the obtained pulse is clean shows the component. The demised pulse, the solid line

followed by the dotted line. This line indicates the pulse captured from sensors.

Figure 15Beat-to-Beat alteration.

As shown in the Fig. (b), The Time domain features are Percussion wave(P), valley

(V), tidal wave (T) and dichotic wave (D). These three important time domain features are

obtained from the pulse waveform. A standard pulse wave form with definite amplitude

and time duration must be there in the wave-parts. This indicates that the body organs and the

heart are functioned proper or not. The prior systems were compared with the collected

waveform which is rich in harmonics and it is superior too.

35

Electret Microphone As A Sensor

A low-cost Electret microphone (e.g. Rapid 35-0192) will provide an output of

around 1mVrms from normal speech at a distance of around 60 cm. So you can expect signal

levels from the microphone of around 1 - 3 mV. An electret microphone is a special type of

Field Effect Transistor (FET), which converts vibrations (from sound or physical contact)

into an electric signal, which is then amplified internally by the FET. The electret is

therefore polarized, and must be connected the correct way around - the 0V (negative) leg

can be identified as it is connected to the external metal can by a visible tag.

A Novel Pulse Measurement System

The proposed APM system combines the Fast Fourier Transform (FFT), the centroid

method, and the optical triangulation method. The frequency spectrum of the arterial pulse

waveforms measured at the specified point is obtained by FFT method. The calculations were

conducted by MATLAB 7.0 and Origin 6.0. It is possible by using the FFT theory [15] to

build a variety of non-sinusoidal waveforms consisting of many sinusoidal waveforms. In

other words, a non-sinusoidal waveform can be decomposed into many sinusoidal waveforms

with different frequencies, amplitudes, and phases. Due to the speed limitation of the CMOS

image sensor, we discuss only the fundamental sinusoidal waveform in this paper.

The laser triangulation method is simple in structure. It makes possible to measure the

subject‘s arterial pulse waveforms in a non-contact way. The experimental data show that

changes in the arterial pulse waveforms can be detected by analyzing the centroid movements

of a laser spot. The changes of the centroid of the laser spot, which is measured at certain

points on the wrist, can be transformed into the changes in magnitude of relative height

caused by skin vibration.

The basic operation principle of the proposed APM system is described as follow. A

laser diode, a laser driver, and a CMOS image sensor are used to establish an optical non-

contact pulse measurement device. The laser diode emits laser light onto the measurement

site of skin surface where its arterial pulsation needs to be determined. The laser spot is

formed on the skin surface of the wrist of tested subject and the variation of the spot image is

captured by the CMOS image sensor and then projected onto the scattered points that

represent arterial pulsations. These scattered light points are processed by FFT method to

determine the amplitude and frequency of arterial pulses of tested subjects. In this work, the

proposed APM system adopts a simple structure based on optical triangulation. The

36

geometrical layout of the designed APM system is depicted in Fig. 1. In Fig. 1, X represents

the distance between the target and the collimated lens of laser diode and δX is the small

fluctuation (i.e., the distance between measured points A and B) of skin surface due to

arterial pulsation. The target distance X is measured continuously. Using a simple

triangulation principle, the measured X target coordinates are mapped onto the detection

position d on the CMOS sensor, as shown in Fig. 1. The target distance X is given by

Figure 16Geometrical layout of the arterial pulse measurement system.

where L is the distance between the laser and the CMOS image sensor, d is the distance

between the two spots mapped onto the CMOS image sensor, f is the focal length of the lens,

Z is the distance between the measured point A and the center C of the lens of the CMOS

image sensor, α is the angle between the axis and the measured point A, δX′ is the distance

between the measured point A and the optical axis of the lens, and θ0 is the angle between

the two axes of the CMOS image sensor and the laser. In our APM system, these parameters

are X = 94 mm, L = 110 mm, Z = 144.7 mm, and f = 16 mm (the focal length of the lens of

the CMOS image sensor). The diameter of the lens in front of the CMOS image sensor is 6

mm. Differentiating Eq. (1) with respect to the measured distance and rearranging the result

yields

37

Where δX is also regarded as the resolution of the designed APM system.

For the experiments conducted in this study, the smallest resolvable amplitude change to a

sub- pixel size of d = 0.8 μm on the CMOS image sensor can be achieved. After calibration,

this value corresponds to a measurement resolution of 9.5 μm achieved by the designed APM

system, i.e., δX = 9.5 μm in Eq. (2). Such a measurement resolution is sufficient to detect the

vibration of human pulsation.

The actual implementation of the APM system is shown in Fig. 2. The sampling area

in Fig.2 was located by a TCM physician. Using simple triangulation method, the

displacement δX of the variation of the laser spot mapped onto the CMOS image sensor

can be determined. The amplitude and frequency of the arterial pulse can be obtained by

analyzing the spot position. The centroid method has been widely used to locate a light spot

with respect to various types of image features, to sub-pixel accuracy. The resolution of the

APM system can be increased by

Figure 17The actual implementation of the proposed APM system.

38

introducing sub-pixel processing technique. The threshold level of image pixel is deliberately

set so that in the captured laser spot image a pixel with intensity below the threshold level

will be neglected. In our experiments, the threshold of the gray level of image pixel is set to

20% below the highest full range level. Also, note that the vibration frequency of arterial

pulsation can be obtained by processing the recorded image by means of FFT method.

A CMOS image sensor with a 5.3 × 3.8 mm2 active area (HV7131D, manufactured

by Hynix Semiconductor Incorporated.) was used to detect the laser light spot with high

accuracy and stability. The sensor has a 648 × 488 pixels array and each compact active pixel

element has high photo- sensitivity (3150 mV/lux-sec). It can convert the photon energy to

analog voltage signal with a resolution of 8 μm. The CMOS sensor utilizes three On-chip 8-

bit Digital to Analog Converts (DAC) and 648 comparators to digitize the pixel output.

The output power of the adopted laser diode (Model no.: QL63d5sA, MORETEC,

Inc.) is 1.3 mW, its wavelength is 650 nm, and its spectral width is about 20 nm. The

diameter of the laser spot on the skin surface is approximately 1 mm, giving a spot on the

CMOS sensor of tens of pixels in both directions.

At normal incidence of laser light, about a 4 ~ 7% power reflection occurs due to the

differences in the refractive indices of the skin layers [18]. Also, some photons are scattered

by superficial skin. The light scattered from the skin surface is the most important signal for

this pulse measurement. However, photons penetrating into skin layers are also scattered,

providing diffuse reflections which are of lower power than the first kind of light scattered by

the superficial skin. The exposure time (ET, equivalent to gain level) of the CMOS sensor

can be adjusted to different levels for different situations. This means that the signal-to-noise

ratio (SNR) can be improved by setting the gain level.

There is a large amount of stray light due to the diffusely scattering nature of the skin

tissue. By decreasing the ET of the CMOS sensor, the undesirable light can be eliminated.

For example, Fig. 3(a) shows the experimental result when the ET value of the CMOS image

sensor was set to 40. In the case of Fig. 3(b), the ET value of the CMOS image sensor was set

to 1. The contours of the laser spot

39

Figure 18The contours of laser spots after ET values of the CMOS image sensor were set to:

(1) 40 and (2) 1.

shown in Fig. 3(b) are obviously smaller than that in Fig. 3(a). In a normal image capturing

situation, as shown in Fig. 3(a), the previously measured point can be approximately

relocated by comparing the two consecutively captured frames. The signal processing

flowchart is shown in Fig. 4. The software packages of MATLAB 7.0 and Origin 6.0 were

used to develop the signal processing program. The program we developed can calculate the

40

amplitude and frequency of the arterial pulse of the tested subject via examining the relative

movements of the measured laser spots.

Wrist Pulse Characteristics and Acquisition

A. Wrist Pulse Characteristics

The pulse pressure signal monitored on the radial artery follows a specific pattern. A wrist

pulse of a healthy person is shown in Fig. 1. A steep rise in pulse pressure signal is observed

due to systole phase of the heart. This action forms a percussion wave, which achieves the

peak in the signal strength. Percussion wave is the forward wave generated in the arterial

structure by pumping action of the heart. After this systolic wave, a secondary peak known as

Tidal wave is observed. In the descending part of the wrist pulse a dicrotic wave appears due

to shock of closure of the aortic valve, indicating start of diastole phase of the heart. The

incisures present due to this is also referred as dichotic notch. The peak amplitude H1 and its

timing T1, amplitudes H2, H3 and their respective timings T2, T3 associated with Tidal wave

as well as amplitudes H4, H5 and their respective timings T4, T5 associated with dicrotic

wave of individual wave are helpful in the diagnosis. The pulse of an abnormal health status

person shows different characteristics in terms of rate, rhythm as well as absence of tidal

and/or dicrotic notch [8]. Hence identification of these features is essential in wrist

pulse analysis.

Figure 19Wrist pulse of a healthy person

41

B. Wrist Pulse Acquisition

A pressure sensor capable of measuring 350 mmHg of pressure is used for acquiring

wrist pulse. The signal is amplified and filtered to obtain a clean pressure signal. The signal

is then sampled with sampling frequency of 500 Hz with the help of USB-6009 Data

Acquisition system from National Instruments. Since the wrist pulse contains most of the

signal in the lower frequency band, the signal is filtered with the help of a low pass filter

having appropriate cut off frequency. The signal also contains baseline fluctuations due to

motion artifacts and respirations of the person. This baseline is removed using discrete

Meyer wavelet transform.

42

CHAPTER 4. EXPERIMENTAL RESULTS In TCM there are four diagnostic methods: inspection, auscultation, questioning, and

palpation. Among these methods, the pulse diagnosis by palpation is the most important and

also the most difficult one. According to Chinese medical literature [13], there are in total of

29 wave patterns of the arterial pulse, each having a specific name. However, it is too

difficult and too subjective for most people to distinguishing 29 different wave patterns with

the finger tips. The TCM physician usually needs to use an auxiliary device to determine

these patterns. The relationships between the organs health and the measurement points are

illustrated in Fig... This is especially meaningful in TCM clinical practice, because for pulse

diagnosis the physician places his index, middle, and ring fingers on the patient‘s wrist, in

accordance with the three locations called Chun, Guan, and Chy [14]. In our experiments, we

found that variations in the pulse at relative measurement points on the wrist mentioned-

above could reveal something about the physiology of the tested subjects.

Some of the measurement results for pulse amplitude and frequency obtained under different

conditions are shown in Figs, respectively. In Fig the pulse amplitude and frequency

measured at the Chun point of left hand (related to the small intestine) showed an obvious

increase after having food. Fig.(a) shows results observed before the meal and Fig. (b) are

those after meal,

43

Figure 20Laser spot centers measured by APM system: (a) the original data of centroid

variation of laser spot center and (b) the data after enlarged amplitude scale and with

filtering.

respectively. The measurements were conducted 30 minutes before and 30 minutes after

eating the meal.

In Fig. (a), the peak to peak value of pulsation amplitude is approximately within 0.4

pixels, i.e., the maximum variation in pulse amplitude is approximately equal to 38 µm. In

Fig. (b), the peak to peak value of pulsation amplitude is approximately ranged from − 0.32

to 0.28 pixels. This means that the maximum variation in pulse amplitude of the tested

subject after meal 30 min will increase to 48 µm. An examination of Figs. (a) and (b) also

observes that the pulse frequency measured at the Chun point of left hand of the tested

subject is changed from 1.27 to 1.35 Hz.

44

Figure 21. Full spectrum analysis of the data in Figure 9: (a) the results of without using filter

and (b) the results of using filter.

45

Figure 22Illustration of palpation positions for pulse diagnosis used in traditional Chinese

medicine.

The other test showed that staying up late caused changes in the amplitude and frequency of

the pulse at the Guan point of left hand. The pulsation at the Guan point of left hand is

closely related to the liver activity. Without staying up late, the amplitude and frequency of

the pulsation of the tested subject measured at Guan point were normal, as shown in Fig.

13(a). But after staying up late, it showed an apparent increase in the amplitude of

measurement data, as shown in Fig. 13(b). In Fig.

(a), the peak to peak amplitude of pulsation measured at the Guan point on the left

hand of the tested subject varies in small range, approximately equal to 0.2 pixels

(i.e., equivalent to 19 µm). In Fig.

(b), for the tested subject staying up late the pulse variation in peak to peak amplitude

measured at the Guan point enlarged approximately to 0.6 pixels (for most portion),

46

Outlier Pulse Identification and Pulse Feature Derivation

Dynamic Time Warping Algorithm

Dynamic Time Warping (DTW) algorithm is used conventionally to align two time

series of unequal lengths. The wrist pulse segments are derived from the wrist pulse

series and two such segments are represented with time series P and Q .

P = P1 , P2 ,......... Pn

Q = Q1 , Q2 .........Qm

A matrix‗d‘ with nxm dimension is constructed to align two time series P and Q where ith

and jth element of matrix is the distance between Pi and Qi element of the series given by

contiguous warping path ‗W‘ through matrix elements is given as W = w1 , w2 ,.......wk

w1 = d (1,1) and wk = d (n, m)

Where x(m, n) ≤ k < m + n − 1 .

In addition it follows the monotony condition

i(k ) ≤ i(k + 1) , j(k ) ≤ j(k + 1)

and the continuity condition

i(k + 1) − i(k ) ≤ 1 , j(k + 1) − j(k ) ≤ 1 .

With boundary conditions as

Out of possible multiple paths an optimal warping path is selected based on minimizing

following cost function

wi , j = d (i, j) + min{wi−1, j , wi , j −1 , wi −1, j −1}

47

Figure 23Warping Path Calculation

Nadi Tarangini Discussion

A. Comparison with earlier systems.

Fig. 4 shows few of the pulse waveforms from earlier works. These waves are of

different patients, of different age groups, using various techniques and are of different

resolutions. The subfigures from (a) through (o) are from [3], [11], [12], [4], [13], [14], [15],

[16], [17], [18], [19], [20], [21], [8], [22] respectively. We observe that there has not been

substantial improvement in the pulse waveforms from

1950s to the current methodologies in a form that permits diagnosis.

Feature extraction followed by machine learning methods

depend not only on clinician‘s experience but also on the

Figure 24Pulses as presented in selected earlier works, arranged in ascending published date.

48

quality of the pulse waveforms. We now provide some of our normalized pulses from

various patients with different disorders, and age groups for the vata dosha in Fig. 5. It can be

observed that there are distinctly observable patterns in each of the six pulses. For example, a

healthy pulse has a main peak and 2 secondary peaks with regular behavior, whereas a fever

pulse is very irregular in nature. Further, various other disorders are also reflected in the

pulse waveforms as shown. We can observe variations in amplitudes, the rising & falling

slopes, systolic & diastolic energies, velocities, and so on. Machine learning algorithms can

be applied on these waveforms to distinguish major types of nadis defined in Ayurveda [1].

We find there is a strong temporal inter-beat similarity or correlation between

successive beats. Heart rate variability (and now variability in pulse) more appropriately

emphasizes the fact that it is the variations or the intervals between consecutive beats that

is sometimes more important than the heart/pulse rate or the average values. Variability in

various time-domain, frequency domain morphology-based features; such as amplitude,

energies, angles, entropies, velocities have been explored [23], [8]. Such variability patterns

can also be seen in the pulses captured by Nadi Tarangini.

displays cross-correlation between two doshas (vata and pitta) of left hand, and one

dosha (vata) of both the hands of a patient. The phase between them can be determined

using the phase difference between the corresponding harmonics of the FFT. These cross-

correlation features could also reveal some information about body conditions.

B. Important Properties of Pulse Waveform.

The complex pulse signals should be able to provide reproducibility, accuracy and

precision. For our system, we checked the reproducibility and completeness as follows:

• In order to check the reproducibility, pulse waveforms of a single healthy person

were recorded at different times from 10.30am to 5pm for five consecutive days with the

same settings. It was observed that the Ap- proximate Entropy (ApEn) of the waveforms, and

the number of data points per pulse cycle remained (almost) constant for the same timings for

49

the 5 days. There were some slight changes in the pulse shape, as the nadi is also sensitive to

the mental status, stresses, thoughts, and so on.

Figure 25Pulse waveforms using our methodology on patients with various disorders. Only

one of the three acquired waveforms have been given for each type.

• In order to check the completeness of the acquired time series, nadi was acquired

with the same sensor but of a digitizer having an accuracy of 8-bit, 12-bit and 16- bit for

same set of patients. The details captured by 8-bit digitizer were less as compared to the 12-

bit. As there was no significant new information from 12-bit to 16-bit upgradation, we claim

that all the details have been captured. In all the further experiments a digitizer having

accuracy of 16-bit is used.

C. Varying Pressure

In Ayurveda, the nadi is sensed by the Ayurvedic practitioners at the wrist with

varying pressure. At different applied pressures, different amplitudes, energies etc. are sensed

which are then correlated with the body conditions. Further, traditionally in TCM, the pulse

has been classified simply as floating or sinking, according to whether the force exerted to

detect the pulse is small or large [13]. We followed similar methodology of applying varying

pressure using our system, and were able to confirm the desired behavior as shown in Fig. 7.

50

As the contact pressure of the sensor over the pulse point increases, the amplitude of

the pulse signal first increases, reaching a maximum, and then decreases. After a particular

threshold value, the pulse dies. All these observations are consistent with the Ayurvedic

literature [1],[2]. At each pressure, the pulse gives different insights about the body. However

at this point, we consider this finding to be only a appropriate observation which necessitates

further investigation with much larger population samples.

(a) Two doshas of left hand (b) vata dosha of two hands

Cross-correlation in pulse waveforms.

Figure 26The changes observed in the pulse waveform as the applied pressure increases from Left to Right.

(a) Left Hand (b) Right Hand

Comparison in a pulse cycle for various age groups.

D. Variations with Age.

We now describe the changes in the pulse waveforms for 31 patients in our database

depending upon their age. They are classified in age groups ‗below 25‘, ‗25–50‘ and‘above

50‘. We observed that the ‗below 25‘ pulses are more dominant in secondary peaks. ‗25–50‘

51

group is relatively stable, while older pulses are irregular in nature. We have given samples

of healthy nadi of 3 groups in Fig. of both the hands. As can be seen, the pulse duration

increases (rate decreases) as the age increases and also the patterns are different. Again, these

observations are consistent with [4], [15]; the changes are indicative of an age-dependent

reduction in large artery (capacitive) compliance and in small artery (oscillatory or reflective)

compliance.

However, we consider all above findings being only a preliminary suggestion to

investigate further with much larger dataset. While applying machine learning algorithms on

pulse dataset for identifying pulse type, we need to consider all the above dimensions, i.e.

age, pressure applied, disorder and many more given in the literature.

The design of Pulse Collecting Terminal (PCT)

The PCT is the basic unit of WNCTs including a sensor module, signal regulating module,

control and processing module (MCU), wireless module, power control module and user

interface. The module structure of the PCT is shown in Figure

Figure 27Pulse Collecting Terminal

The Sensor module converts the biological signal to electronic signal with flexible

pressure control (similar as the Chinese doctor’s operation) by controlling the pump and

solenoid valve accurately. The Signal regulating module regulates the data for the next

operation and it plays a key role in improving signal’s performance.

52

The MCU converts the physical pulse signal to the digital data and carries the digital

rocessing for the transmission. The wireless module is used to send the processed data to

the sink node wirelessly following a transmission protocol. User interface provides a GUI for

the users. We will describe the design details about those parts as follows.

Figure 28Photo of PCT in measurement and the sensors

53

CHAPTER 5. CONCLUSION

The Modern sensors reflect the feeling information which is being presently used in

Indian ways of Siddha or the regular Traditional Chinese Medicine (TCM). We have taken

the pressure sensing based methodology and the current progress in the instrumentation

technologies in the design of high quality pulse acquisition system. Our Pulse Waveforms, in

the form of time series, has high details, as compared to the former reported systems.

We have proved our waveforms are reproducible and complete. We also showed

variations with respect to applied age and pressure groups which are consistent with Siddha

literature and prior work. Based on this, we believe that our system can be used by a larger

number of lay persons. We are currently evaluating, with the help of Siddha practitioners the

use of Nadi Tarangini for diagnostic purposes. Strong machine learning algorithms can be

tried on these waveforms to differentiate them into major types of nadi defined in Siddha

literature.

Based on optical laser triangulation theory, a non-invasive and non-contact arterial pulsation

measurement (APM) system to detect micro-vibration on skin surface is developed in this work. The

APM system consists chiefly of a laser diode and a CMOS image sensor, and the implementation cost

is pretty low. An extensive series of experiments to evaluate the performance of the APM system was

conducted. The pulse waveforms of the tested subject can be detected by our APM system easily. The

APM system achieves a measurement resolution of μm order. Experimental results also show

that the amplitude and frequency of the pulse of tester have been changed under different conditions.

These tests demonstrate the performance of the proposed APM system for measuring micro-pulsation

on skin surface is pretty good. If a speedier CMOS or CCD image sensor, such as 200 frames per

second or more and a smaller pixel size can be used, the pulse waveform obtained by our APM

system would be more accurate and clearer. To reduce the speckle effect of the laser, a non-coherent

light could be

54


the tested subject: (a) 30 min before a meal (to be continued).

56


the tested subject (continued): (b) 30 min after a meal.

57

Figure 31Pulse information for the Guan point on the left hand (liver) of the tested subject:

(a) before staying up late (to be continued).

58

Figure 32Pulse information for the Guan point on the left hand (liver) of the tested

subject (continued): (b) after staying up late.

59

Our experimental results have shown the feasibility of using the optical triangulation method

and a CMOS image sensor to measure arterial pulsation. Although the demonstrated examples are not

yet sufficient to clinical bearings, they serve to test the method and evaluate the performance of the

proposed APM system. In the future, this arterial pulse measurement system can be improved by

using 3 light sources to simultaneously check the pulsation of three or six different TCM points on the

wrist (i.e., Chun, Guan, and Chy points on one hand or two hands, respectively). And the relation

between the pulse signal and the healthy condition of the subject will be established. We hope that we

can report the investigation results in future.

Modern pressure sensors can reflect the ―feeling‖ information used in the Indian

system of Ayurveda, or in Traditional Chinese Medicine (TCM). We have adopted pressure

sensing based methodology and the recent developments in instruct mentation technologies in

designing a high quality pulse acquisition system. Our pulse waveforms, in the form of time

series, have high details, as compared to earlier reported systems. We showed that our

waveforms are reproducible and complete. We also showed variations with respect to applied

pressure and age groups which are consistent with Ayurveda literature and prior work.

Based on this, we believe that our system can be used by a larger number of lay

persons. We are currently evaluating, with the help of Ayurvedic practitioners the use of Nadi

Tarangini for diagnostic purposes. Rigorous machine learning algorithms could be applied on

these waveforms to classify them into major types of nadis defined in Ayurvedic literature

In this section, we analyze the waveforms obtained from Nadi Tarangini. Sample

waveforms from the left hand of a patient using our system, as described in Section III are

shown in the Fig. 3.

The three different colors in Fig. 3(a) represents the data obtained from the three

different pressure transducers for vata, pitta and kapha doshas respectively. The zoomed

waveform in Fig. 3 (b) gives an idea of the number of points per pulse cycle (sampling

frequency 500Hz) and also that it contains details in the form of secondary peaks. Though the

pulse form is clean, we use wavelet denoising to remove the extreme high frequency

component (Fig. 3(c)). The solid line (pulse after denoising) almost follows the dotted line

(pulse captured from sensor). Fig. 3(d) depicts the denoised signal at better resolution.

60

A pulse waveform is usually composed of important time domain features: percussion

wave (P), tidal wave (T), valley (V) and dichotic wave (D), as shown in Fig. 3(b). These

wave-parts should be present in a standard pulse waveform with definite amplitude and time

duration to indicate proper functioning of the heart and other body organs. These collected

pulse waveforms are rich in harmonics and appear superior as compared to previously

developed systems.

.

61

REFERENCES

1. Www.wikipedia.org

2. Www.ieee.org

3. Aniruddha joshi, anand kulkarni, sharat chandran, v. K. Jayaraman and b. D.

Kulkarni: nadi tarangini: a pulse based diagnostic system, pune, 2007.

4. V. Lad, secrets of the pulse: the ancient art of ayurvedic pulse diagnosis.

Motilal banarsidass, delhi, 2005.

5. S. Upadhyay, nadi vijnana (ancient pulse science). Chaukhambha sanskrit

pratishthan, delhi, 1986.

6. H. Warner, h. Swan, d. Connolly, r. Tompkins, and e. Wood, ―quantitation of

beat-to-beat changes in stroke volume from the aortic pulse contour in man,‖

journal of applied physiology, vol. 5, pp. 495– 507, 1953.

7. G. Mcveigh, c. Bratteli, d. Morgan, c. Alinder, s. Glasser, s. Finkelstein, and j.

Cohn, ―age-related abnormalities in arterial compliance identified by

pressure pulse contour analysis: aging and arterial compliance,‖ hypertension,

vol. 33, pp. 1392–1398, 1999.

8. O.lau and a.chwang, ―relationship between wrist- pulse characteristics and

body conditions,‖ in proc. Of the 14th engineering mechanics conf., 2000.

9. A.brumfield and m.andrew, ―digital pulse contour analysis: investigating age-

dependent indices of arterial compliance,‖ physiological measurement, vol.

26, pp.599–608, 2005.

10. l.xu, d. Zhang, k. Wang, n. Li, and x. Wang, ―baseline wander correction in

pulse waveforms using wavelet-based cascaded adaptive filter,‖

computers in biology and medicine, vol. 37, pp. 716–731, 2007.

11. L.xu, k. Wang, l.wang, and n. Li, ―pulse contour variability before and after

exercise,‖ ieee cbms, 2006.

12. M.aritomo and y.yonezawa, ―a wrist-mounted activity and pulse recording

system,‖ in proc. Of the first joint bmes/embs conf., vol. 2, p. 693, 1999.

13. h.lee, s. Suzuki, y. Adachi, and m.umeno, ―fuzzy theory in traditional chinese

pulse diagnosis,‖ in proc. Of int. Joint conf. On neural networks, vol. 1, pp.

774–777, 1993.

14. J.remington and e. Wood, ―formation of peripheral pulse contour in man,‖

journal of applied physiology, vol. 9, pp. 433–442, 1956.

62

15. J.s. k.h.wesseling, j.r. jansen and j.schreuder, ―computation of aortic flow from

pressure in humans using a nonlinear, three-element model,‖ journal of

applied physiology, vol. 74, pp. 2566–2573, 1993.

16. Y. Yoon, m. Lee, and k. Soh., ―pulse type classification by varying contact

pressure,‖ ieee engineering in medicine and biology magazine, vol. 19, pp.

106–110., 2000.

17. N. Linton and r. Linton, ―estimation of changes in cardiac output from the

arterial blood pressure waveform in the upper limb,‖ british journal of

anaesthesia, vol. 86, pp. 486–496, 2001.

18. J.allen and a.murray, ―age-related changes in the characteristics of the

photoplethysmographic pulse shape at various body sites.‖ Physiological

measurement, vol. 24, pp. 297–307, 2003.

19. C. Vlachopoulos, n. Alexopoulos, d. Panagiotakos, m. O‟rourke, and c.

Stefanadis, ―cigar smoking has an acute detrimental effect on arterial

stiffness,‖ american journal of hypertension, vol. 17, pp. 299–303, 2004.

20. D. Duprez, d. Kaiser, w. Whitwam, s. Finkelstein, a. Belalcazar, r. Patterson,

s. Glasser, and j. Cohn, ―determinants of radial artery pulse wave analysis in

asymptomatic individuals,‖american journal of hypertension, vol. 17, pp. 647–

653, 2004.

21. L. Xu, k. Wang, and l. Wang, ―pulse waveforms classification based on

wavelet network,‖ ieee embs, pp. 4596–4599, 2005.

22. E. Treo, m. Herrera, and m. Valentinuzzi, ―algorithm for identifying and

separating beats from arterial pulse records,‖ biomedical engineering online,

vol. 4, 2005.

23. L. Wang, k. Wang, and l. Xu, ―lempel-ziv decomposition based arrhythmic

pulses recognition,‖ ieee embs, pp. 4606–4609, 2005.

24. H. Wang and y. Cheng, ―a quantitative system for pulse diagnosis in

traditional chinese medicine,‖ ieee embs, pp. 5676–5679, 2005.

25. I. Wilkinson, c.mceniery, and j. Cockcroft, ―atenolol and cardiovascular risk:

an issue close to the heart,‖ the lancet, vol. 367, pp. 627–629, 2006.

26. L. Li and z. Wang, ―study on interval variability of arterial pulse,‖ in proc. Of

the first joint bmes/embs conf., vol. 1, p. 223, 1999

63

27. S.ramchurn and a.murray multiracial analysis of the day and night

characteristics of heart rate variability. Computers in cardiology, pages 421–

424, 2002

28. S.christopher and p.raymond gender-related differences in the central arterial

pressure waveform. American college of cardiology, 30:1863–1871, 1997.

29. W.hwang and s.mallat characterization of self- similar multiracial with

wavelet maxima. Applied and computational harmonic analysis, 1:316–328,

1994.

30. Department of biomedical engineering, ming chuan university, no. 5,

deming rd., gweishan

31. Township, taoyuan 333, taiwan.

32. 2 department of optics and photonics, national central university, no. 300,

jung-da rd., chung-li

33. City, taoyuan 320, taiwan.3 department of bio-industrial mechatronics

engineering, national taiwan university, no. 1, sec. 4, roosevelt rd., taipei

106, taiwan.

Modern Pulse Meter for Traditional Indian Medicine

Documents

Transcript of Modern Pulse Meter for Traditional Indian Medicine