CHAPTER 01

1
1.0 General Introduction

Ayurvedic medicine is based on the unique concept of the tridosha (Walia and Singh, 2010). The Vedic

Medical Science comes under Ayurvedic Medicine is a deep science based on diagnosis of diseases by

Nadi analysis. The ancient Ayurvedic practitioners were able to diagnose many physical and medical

conditions only by sensing the pulse at specific points in the body without using any medical equipment.

In contrast, they were able to diagnose many diseases in early stages and they could even predict the date

and time of death in some cases (Joshi, 2005).

fivanadi is the most used pulse diagnosis technique. In that case, patient’s hand is hold by the

practitioner’s left hand for support and the radial pulse is examined by the three finger tips (index, middle

and ring) of the practitioner’s right hand (Patil et al., 2015).

However, pulse diagnosis is a subjective matter and accuracy depends on the experience and practice

(Lad, 1996). To overcome the subject based problems, it is important to develop electronic devices and

algorithms. Pulse waveform acquisition and analysis was less considered earlier but now researches are

done in that area more frequently (Walia & Singh, 2010).

In this study, we aim to develop a medical device which can detect pregnancy at early stages using wrist

pulses.

1.1 Ayurvedic Medicine

Ayurvedic medicne is one of the grandest, unique and supreme gift of ancient India to the world which

is used for both diagnosis and treatments (Walia & Singh, 2010). This basis of Ayurvedic medicine is

known as “Prakrati Nidana” (Joshi, 2005). Prakarti of the person can be determined based on the

characteristics such as body size, hair, weight, appetite, voice, skin, personality and habits.

2
There are two main purposes in Ayurvedic Medicine.

1. Maintain the sound health of a healthy people by prevention.

2. Cure diseases of unhealthy people by appropriate diet, lifestyle, panchakarma and rejuvenation.

(Vaidya et al., 2015)

Ayurveda states that every living thing in the universe is made of five elements and they correspond to

the five senses in human as shown below.

Earth or Prithvi = Smell

Water or Apa = Taste

Fire or Tejas = Sight

Air or Vayu = Touch

Space or Akash = Hearing

These five elements constitute three primary life forces or doshas called Vata, Pitta, Kapha (Walia &

Singh, 2010). According to Ayurveda, vata is the energy of movement, pitta is the energy of digestion or

metabolism and kapha is the energy of lubrication and structure (Joshi, 2005). These three doshas can

be found in all the cells in every human. But the combination of three doshas is unique to each individual.

While some people are characterized by a single dosha, most people are combination of two or three.

This combination is called the concept of tridosha and it is the foundation of Ayurvedic medicine which

represents the humeral balance. The dominating type of the tridoshas determines the dosha type of that

person. Further, dosha type associated with the characteristic tendencies and according to that Ayruvedic

physician or the practitioner does the diagnosis (Walia & Singh, 2010).

Vata dominant people are active, astringent, creative, adapt to new ideas quickly, unpredictable,

enthusiasm, however, quickly to forget and anxiety. They are underweight with a light body frame of
3
thin and light muscles. Moreover, they have cool dry skin and dry hair. They have high energy in the

beginning, but get tired easily as they over use their energy. Vata people likely to have headaches,

earaches, insomnia, dry coughs, sore throats, abdominal gas, diarrhea, constipation, muscle spasms,

lower back pain, arthritis and problems related to nervous system. In addition, they tend to have

hypertension, irregular heart rhythms, premenstrual syndrome and sexual dysfunctions.

Pitta dominant people are medium size and able to maintain the weight easily. Most of the time they

have fair, reddish skin tend to have freckles and straight, blond hair. Their features such as nose and eyes

are sharp. They have a very sharp memory, short tempered, sharp-witted, self-confident and passionate.

However, they usually have medium strength and they can become aggressive. They use sharp words

when speaking. These people have a strong appetite and a strong digestion capability. They are fairly

predictable in their daily routines. Pitta dominant people likely to have heartburns, ulcers, hot sensations

in the stomach or intestines, insomnia, rashes or inflammations of the skin, acne, skin cancer, anemia.

Moreover, they may have gallbladder and liver disorders.

Kapha dominant people has strong and heavy body made of soft skin, soft hair and large soft attractive

eyes. They tend to speak in low and soft voice and gain weight easily. They need more sleep and warmth.

Usually, these people are relaxed, graceful, slow-moving, forgiving, compassionate, procrastinate,

nonjudgmental, faithful and affectionate. Their energy is made-up with all the constitutions, but it is

steady, enduring and not explosive. They learn slowly but have excellent long-term memories. Kapha

type is more resistive to diseases. Nevertheless, they are inclined to obesity, allergies, colds, headaches,

problems related to respiratory system, atherosclerosis, and painful joints (Vaidya et al., 2015, Walia &

Singh, 2010).

4
1.1.1 “Nadi Vijnanam” in Ayurvedic Medicine

The Vedic Medical Science comes under Ayurvedic Medicine is an important method of diagnostics.

The word “Nadi” refers to pulse in veins and arteries. (Joshi, 2005) Vijnanam refers to the examination

and understanding of pulses. As pulse is examined in blood vessels which carries blood through all over

the body even upto the cellular level along with nutrients, it is a perception to explore the human body

(Patil et al.,2015.). The ancient ayurvedic expertise could do the correct diagnosis of all kind of physical

and mental alignments only by sensing of “Nadis” at specific points of the body. Ayurvedic expertise

could do the perfect early diagnosis of diseases or disorders and they could predict even the date and

time of the death in some cases (Joshi, 2005). Ayurvedic pulse based diagnosis is capable of doing non-

invasive diagnosis which gives similar results to new techniques such as biopsy or CAT scan. Moreover,

early diagnosis of cancers can be done non-invasively without any side effects (Devi, 2013). Alakananda

Devi gives case studies on cancer diagnosis by pulses and modern medicine and the similarities of the

results in her paper “Pulse Reading” published in 2013.

According to the Ayurvedic literature, there are hundreds of different nadi patterns which reflect the

spectrum of internal state of the body and it’s biochemical and physiological functions. There are nine

prominent points in the human body at where these specific nadi patterns are sensed by touch of the tips

of one or more of the three central fingers (Figure 1.1). Among them, fivanadi or the radial artery

sensation is the most common one. Generally, that is known as wrist pulse analysis (Joshi, 2005). Wrist

pulse signals contain vital information of health activities and it can reflect the pathologic changes of a

person’s body condition. Therefore, the practitioners can tell the health conditions of a patient by feeling

his wrist pulses (Chen et al. 2011).

5
Figure 1 . 1 : Nine Prominent Points in the Human Body at where the Specific Nadi Patterns are
Sensed by Touch of the Tips of One or More of the Three Central Fingers accorfing to Ayurvedic
Medicine (Joshi, 2005).

6
However, there is no any significant differences between wrist pulse sigal and the finger tip pulse signal.

(Zhao and Small, 2005).

1.1.2 How the Pulse Pattern is Measured?

Pulse examination is done in the same way as it was done centuries ago. In fivanadi, patient’s hand is

hold by the practitioner’s left hand for support and the radial pulse at radial fossa is examined by the

right hand’s three finger tips, index, middle and ring finger tips (Patil et al., 2015). Three finger tips are

kept slightly pressed on the radial pulse at Radial Fossa, that is at the root of the thumb to senses the

pulse and the throbbing nature of the pulse is observed mentally by the practitioner. Vata is observed by

the first index finger tip, pitta is observed by the middle finger tip and kapha is observed by the third

(ring) fingertip (Walia & Singh, 2010). (Figure 1.2) The right hand pulse is sensed in men and left hand

pulse is sensed in female (Uebaba et al. 1993).

Figure 1 . 2 : Examination of the Radial Pulse at Radial Fossa (Root of the Thumb) of the Patient
by the Ayurvedic Practitioner’s Three Finger Tips, Index, Middle and Ring of the Right Hand.
Index Finger is Placed on Vata, Middele Finger is Placed on Pitta and Ring Finger is placed on
Kapha (Lad, 1996).

7
1.1.3 Parameters which Affect the Pulse Pattern

Gender is one of the major fact which affects the heart rate. Generally, females have higher heart rates

than male. The reason for this difference is that the females have less total muscles mass comparatively.

Therefore, females experience higher heart rates to facilitate required metabolic rates (Collora, 2015).

According to Ayurveda, pulse waveform changes depending upon the age as the doshas are changing

within the phases of life. Generally, three phases are described in Ayurveda, according to which dosha

type is disturbed dominantly (Walia & Singh, 2010).

Table 1 . 1 : Dominantly Disturbed Dosha Type (Vata,

Pitta, Kapha) with the Age according to the Ayurveda.

Age Group Dominantly Disturbed

Dosha Type

Childhood (0-16 years) Kapha

Adults (16-50 years) Pitta

Elderly people (over 50 years) Vata

Further, the maximum heart rate decreases as people get older because of physiological factors such as

telomere shortening and associated deconditioning.

8
 Table 1. 2 : Standard Pulse Readings for Various Age Groups.

Newborn Infants Infants Children Children Welltrained

(0-3 months (3-6 (6-12 (1-10 over 10 years & adult

old) months) months) years)

Adults, athletes

Including seniors

100-150 90-120 80-120 70-130 60-100 40-60

Another factor is the atmosphere temperature. This is a thermoregulatory mechanism of the body which

helps to maintain the body temperature. In hot atmosphere, heart rate increases to transfer more heat

away from the body and in cold atmosphere, heart rate decreases to preserve body temperature.

After meals, heart rate increases in order to supply more blood to the gastrointestinal tract to aid digestion.

Moreover, heart rate depends on the weight of the meals. Heavy meals increase the heart rate more than

a light meal (Collora, 2015).

Activity level of the body is another factor which affects the heart rate. In high level activities, heart rate

increases to facilitate the oxygen demand and carbon dioxide removal (All about heart rate (pulse),

2016).

Dehydration increases heart rate and it apply more stress on the heart.

Caffeine is a stimulant drug which is found in coffee, tea and sodas. It mimics the activity of natural

hormone adrenaline and influences the nervous system to increase heart rate. The drugs such as cocaine

and ephedrine are also stimulants which act in the same manner (Collora, 2015).

9
Furthermore, specific medications such as beta-blockers and calcium channel blockers decreases the

heart rate by decreasing hormonal influence on heart rate and reducing calcium supply to the heart

muscles respectively (All about heart rate (pulse), 2016).

Finally, people suffering from specific diseases specially hypertension, hypercholesterolemia and

diabetes, have decreased flexibility of vasculature hence they have increased stress in the circulatory

system. Therefore, the shape, amplitude and rhythm of the wrist pulses alter corresponding to

hemodynamic characteristics of blood flow (Chen et al. 2011).

1.1.4 Parameters Measured During Pulse Sensing

The variations of the levels of tridosha leads to the different patterns of pulse. Accordingly, the waveform

changes and variations in the amplitude, rate, rhythm, shape, regularity and hardness can be observed.

Therefore, it is important to analyze pulse patterns quantitatively as wells as qualitatively. Pulse

amplitude is the distance from the rest position to a point of the signal. Amplitude comprises vital

information of the pulse. For an example, reduced amplitude indicates the reduced blood flow.

The pulse rate is the number of pulse signals occur per minute. It differs from person to person. Moreover,

it is an important parameter of Nadi Pariksha. The peak detection algorithm can be used to measure the

pulse rate where the mean of the signal is used as the threshold value for the algorithm.

In Ayurvedic medicine, they have introduced waveform for each dosha with regard to the rhythm. In high

levels of Vata, rhythm is similar to snake’s curved scrawling. In high Pitta, rhythm relates to frog’s jump

and in high Kapha, swan’s smooth and slow rhythm takes place (Joshi, 2005). (Figure 1.3)

According to Ayurveda, there are seven levels of radial pulse readings known as Dhatus which are in-

between the superficial layer to the inner core of the body. The followings can be given as examples.

- Superficial layer is rasa and capillary layer is raktha.

10
Superficial layer is known as first layer and the most deep level where obliterate pulse is sensed, is known

as seventh layer. These levels can be felt according to the pressure applied on the radial artery. One can

feel the spikes of the pulse change as pressing deeper or shallower from one level to the other. These

levels are diagnosed by Ayurvedic practitioners according to their unique techniques developed with the

clinical practice.

In modern medicine, pulse analysis relates more towards the cardiovascular diagnosis whereas in

Ayurvedic medicine, there is a wide range of perception depends on the seven levels. These levels

elaborate more details about the sub types of doshas, status of prana, theja, ojus and conditions of seven

tissue levels.

It is important to have an idea on prakruti-vikruti paradigm. Prakurti is sensing of the seventh or the

deepest level. It is the basic constitution of a person. Vikruti is sensing of the superficial level of the pulse

which indicates the present condition of a person. For a healthy person, the reading of prakruti and vikruti

are equals (Lad, 1996).

11
Figure 1 . 3 : Waveforms assumed for Each Dosha with Regard to the Rhythm
(Joshi, 2005).

12
1.1.5 Reasons for Inaccuracy of Pulse Readings

There are several reasons for inaccuracy of pulse readings. One such reason is, inaccurate finger

positioning. To get the accurate reading, three fingers should be placed at same level. Another reason is,

placing the index finger directly on radial tubercle. In that case, any throb may not be felt. Physical

appearance can be named as another factor for inaccurate pulse reading. According to the Ayurveda,

there are specific physical parameters for each dosha type. Moreover, a thick subcutaneous fat layer can

interrupt the sensation of the pulse. In addition, premature births and umbilical strangulation also affect

the pulse of a person.

To avoid such problems, it is important to question the patient. Most importantly, pulse sensing needs

persistent and pro-longed practice (Lad, 1996).

1.1.6 Pregnancy Diagnosis by Pulse Analysis

Ayurvedic medicine is capable of diagnose pregnancy and the gender of the fetus by wrist pulse analysis

after fifteen to thirty days of conception. It can be detect by sensing the pulse of fifth level at artava pulse

of the woman both in left and right sides. Shift of nadi towards medial side or lateral side, that is little

finger (ulna border) or thumb finger (radial border) indicated pregnancy. Thus, artava pulse is very

important to detect early stage of pregnancy.

Pregnancy can be detected by sensing the pulse on the lateal or inner side at the base of the little finger

on both right hand and the left hand. The pulse become more conspicuous. If the pulse is prominent in

right side, it indicates a male fetus and if it is prominent in left side it indicates a male fetus.

Moreover, to detect the time of conception, it is important to give the complete attention to twelve vegas.

Number of times of pulse shifts in different directions (inner or outer of the border) helps to detect time

of conception. For an example, one shift indicates the first month of pregnancy and two shifts indicate

second month of pregnancy.

13
Shifts in both sides, that is to both inside and outside indicates conception of twins.

If any interruption of the Pitta or Vata spikes occurs during the shift that indicates a miscarriage or

obstructed labor (placenta previa). If the Pitta and Vata spikes exist without interruptions and the gati

under Kapha and Pitta are regular, it indicates a smooth, normal delivery. A Vata shift indicates that the

woman is pregnant for short period of time and gandakala to the fetus. A Vata spike under Kapha finger

denotes ovulation and the number of beats between two shifts indicates the number of days since

ovulation.

To understand the prakruti of the fetus, have to sense the central parts of the fingers on fifth level. Shift

occurs under index finger, middle finger or the ring finger denotes the prakruti Vata, Pitta, Kapha

respectively. If the shifting occurs under two finger, the fetus is dual doshic, which is Vata-Pitta, Pitta-

Kapha or Vata-Kapha.

It is important to understand the difference between a spike and a shift. Spike is a vertical movement

whereas shift is a horizontal movement.

However, there is another way to confirm pregnancy rather than sensing the wrist pulses. If the pulse

sensation on the lateral and inner side of the base of the little fingers on the right and left hands is

prominent, it indicates pregnancy. If the right hand sensation is more prominent, it indicates a male fetus

and if left hand sensation is more prominent, it indicates a female fetus.

However, to detect pregnancy by pulse analysis more practice, patience and complete attention are

necessary requirements (Lad, 1996).

14
1.2 Pregnancy

Pregnancy occurs due to the sexual intercourse or it may occurs as a result of assisted reproductive

technology (Boundless, 2016).

The sperm from the male and the egg from the female, unite in one of the two fallopian tubes and the

fertilized egg is known as a zygote. The cell division of the zygote begins approximately after 24 to 36

hours when the sperm and the egg unite and continues at a rapid rate to a blastocyst. The blastocyst

moves towards the uterus and attaches to the uterine wall. This takes about one week and known as the

implantation (OMICS International, 2016).

The first trimester of pregnancy is from week 1 to week 12. In this period, conception occurs and risk of

spontaneous abortion or the miscarriages are high (Boundless, 2016). Further, minute ventilation is

increased by 40% and many symptoms and discomforts of pregnancy like nausea and tender breasts

appear in the first trimester. The womb will grow to the size of a lemon by eight weeks. Though

the fetus grows to a recognizable human shape and starts the movement during the first trimester, the

movement cannot be felt by the mother until the second trimester, more specifically until the 19th-21st

weeks (OMICS International, 2016).

1.2.1 Signs of Pregnancy

It’s not necessary to perform the pregnancy tests to know whether the woman has become pregnant

because naturally her body shows some symptoms or discomforts (Mom Junction, 2014). Early

pregnancy includes symptoms such as missed periods, tender breasts, nausea and vomiting, hunger and

frequent urination.

These symptoms and discomforts occurs because of pregnancy, may not significantly interfere daily

routine of the pregnant woman and these are not a threat to the health of the mother or the fetus. However,
15
there is no clear splitting up of symptoms, discomforts and complications. Further, depending on the

severity, some cases may characterized as either a discomfort or a complication (OMICS International,

2016).

Missed Periods:

If the regular periods in menstrual cycle is delayed by more than ten days, it may be a sign of pregnancy.

However, missing menstrual cycle is not an accurate sign of pregnancy as it may occur due to medications

like contraceptives and eating disorders like anorexia or bulimia.

Spotting:

If a woman notices a few drops of blood about between 1-2 weeks of her last period, it could be

implantation bleeding. It occurs when the embryo implants itself into the wall of the uterus. However,

this is not common for all women.

Nausea and/or Vomiting:

Nausea is the most visible symptom of pregnancy. If it occurs quite often without any apparent reason,

it is a definite sign of pregnancy which occurs due to hormonal levels rising. Most probably, women

experience nausea and/or vomiting at about three weeks into their pregnancy. Any smell or taste can

trigger such a response to pregnancy.

Bloating:

Women may feel an abdominal bloating due to the hormonal changes during pregnancy. They may also

experience cramps along with bloating.

16
Soreness of Nipples:

Changes in breasts such as tenderness and painful to touch are also an early visible signs of pregnancy.

Further, breasts become bigger and fuller, areolas and nipples become darker and there are small bumps

on the nipples.

Frequent Urination:

Frequent Urination, especially occurs in the first trimester in every 20-30 minutes. It is a certain sign of

pregnancy and occurs as a result of the increased water retention and blood flow in the body. In contrast,

blood flow through the kidneys increases due to hormonal changes. As a result, the bladder becomes full

speedily. (Mom Junction, 2014) In other words, increased intravascular volume and elevated GFR

(glomerular filtration rate) causes frequent urination. (OMICS International, 2016) The fetus may also

exerts pressure on the bladder which causes frequent urination. (Mom Junction, 2014) Furthermore,

urinary tract infections can be seen among pregnant women. (OMICS International, 2016)

Backache:

Lower back pains occur similar to before/during a period, could imply pregnancy. Some women

experience back ache throughout the day and it does not recover with rest or massage. Pregnancy back

pain occurs around the sacroiliac joint. It occurs due to many reasons as explained below.

1. Weight Gain

Pregnant woman undergo a sharp rise in weight between 25-35 pounds. As the spine has to

support the weight, back pain occurs in pregnant women. Further, acute back pain may occur as

the expanding uterus exerts pressure on the blood vessels, nerves, pelvis and back. (Mom

17
 Junction, 2014) Striae gravidarum, stretch marks related to pregnancy also can be seen because

of the expanding uterus.

2. Posture Change

Due to the sudden weight gain, the gravitational force acts on the body changes. Therefore, the

woman start adjusting her posture and there are changes in her movements which causes back

pain.

3. Hormonal Changes

A hormone called Relaxin is secreted to relax the ligament in the pelvic region during pregnancy.

The joints also become loose as a preparation for the delivery. These ligament changes cause the

instability. (Mom Junction, 2014) Further, it may cause pelvic girdle pain. (OMICS International,

2016)

4. Muscle Separation:

The rectal abdominal muscle, which is parallel sheet of muscle that extends from the rib cage to

the pubic bone, starts separating at the center due to the expanding uterus. It may cause intense

pain.

5. Stress

Both, the body and mind undergo stress during pregnancy and it leads to back pain or spasms.

(Mom Junction, 2014)

Edema (swelling):

One of the common complaints made by the pregnant women is, edema, which is caused by compression

of the inferior vena cava (IVC) and pelvic veins by the uterus leads. It increases hydrostatic pressure in

lower extremities (OMICS International, 2016).

18
Headache:

Headache is a common symptom of pregnancy due to the physical changes the body goes through.

Irritability/Mood Swings:

During pregnancy, level of neuro-transmitter or chemical messengers to the brain are affected by

hormonal changes. It results in heightened emotions, both good and bad. Most commonly, irritable,

moody behaviors can be observed due to pregnancy hormones.

Food Aversions/Cravings:

Pregnancy leads women to a dreadfully yearning for things at peculiar hours of the day. It is a common

sign of pregnancy. Pregnant women react to food in different manners due to the increased level of

estrogen in the body and the heightened sense of smell.

Disrupted Sleep Cycle:

Sleep cycle of the pregnant women gets disrupted due to the physical changes.

Fatigue:

The increasing level of progesterone causes unusual fatigue or exhaustion. Vomiting and frequent

urination take place as a result of fatigue (Mom Junction, 2014).

Some of the pregnant women may face to constipation and feel tiredness.

Though most of the time the above mentioned symptoms and discomforts indicate pregnancy, it should

be confirmed with a pregnancy test (OMICS International, 2016.)

19
1.2.2 Pregnancy Detection Methods of Western Medicine

Human Chorionic Gonadotropin

Human Chorionic Gonadotropin, which is better known as hCG (Figure 1. 5 ) is a hormone secreted once

the fertilized egg implants to the uterine wall. (Human Chorionic Gonadotropin (HCG), 2014) hCG

hormone secretion begins within the first seven days after implantation of the fertilized egg by the

developing placenta cells to the blood. (Pregnancy testing, 2016) It nourishes the fertilized egg. hCG

hormone in the blood can be detected about 6 days after conception and it can be detected in urine about

12-14 days after conception. (American Pregnancy Association, 2012)

Figure 1. 4 : Chemical Structure of Human Chorionic Gonadotropin (hCG), hormone which

helps to diagnose pregnancy via a Urine or Blood Analysis of the Women who are Suspecting
Pregnancy. (American Pregnancy Association, 2012)

hCG level increases until 8-11 weeks of pregnancy and reached to it’s maximum value. (Association,

2012) But, according to Human Chorionic Gonadotropin (HCG), 2014, it is about 14-16. Then it

decreases gradually. (Figure 1.5) 85% of the normal pregnancies doubles the hCG level in every 48-72

hours (American Pregnancy Association, 2012). Later on, it takes around 96 hours as hCG concentration

to be doubled is high (Human Chorionic Gonadotropin (HCG), 2014).

20
 Figure 1 . 6 : Hormone which Prepares the Uterine Walls or Lining for Embryo Growth (hCG)
Concentration Changes in Urine or Blood During Early Stage of Pregnancy. It Increases
during First 10 Weeks and then Stabilizes Eventually for the Entire Pregnancy after Placenta
Takes over the Progestorone Production (NewKidsCenter, 2016).

The hCG hormone level is given in milli-international units per milliliter (mIU/ml). If the hCG level is

less than 5mIU/ml, it reflects negative pregnancy and if the hCG level is greater than 25mIU/ml, it

reflects positive pregnancy. hCG level of each and every woman rises differently. Therefore, it is

not necessarily the level that matters, but rather the change in the level. (American Pregnancy

Association, 2012)

However, a single hCG test result is not sufficient to confirm pregnancy in most cases. But multiple hCG

tests are performed to make an accurate assessment in questionable health regarding pregnancy.

Nevertheless, to date a pregnancy, the hCG hormone levels should not be used as it might vary in large

scales (American Pregnancy Association, 2012). But according to White, Zieve, and Ogilvie, 2014, a

quantitative HCG measurement can determine the exact age of the fetus.

21
hCG test are performed for multiple purposes. Most common purpose is pregnancy confirmation.

(Human Chorionic Gonadotropin (HCG), 2014) It can be used to diagnose abnormal pregnancies, such

as ectopic pregnancies and molar pregnancies. (White, Zieve, and Ogilvie, 2014) Sometimes, it is used

to monitor the treatment of a molar pregnancy. (Human Chorionic Gonadotropin (HCG), 2014)

Furthermore, it can be used to diagnose possible miscarriages too. (White, Zieve, and Ogilvie, 2014).

Moreover, hCG test together with other screening tests, is performed to see whether there are any birth

defects such as Down syndrome.

Besides pregnancy, hCG can be performed to diagnose or check the treatment of germ cell cancers,

abnormal tissue growing or cancers in the uterus known as choriocarcinoma. Germ cell cancers are

cancers of the ovaries or testicles which are developed from an egg or a sperm. In such cases, hCG test

is performed along with the alpha-fetoprotein test. Not only in women, cancers in the testicles of men

also can be diagnosed by a hCG test (Human Chorionic Gonadotropin (HCG), 2014).

hCG Urine Test for Pregnancy

Most widely used routine pregnancy test is hCG urine test. It uses antibodies of hCG for detection. (Gnoth

and Johnson, 2014). hCG urine test only detects whether hCG presents in the urine sample or not. It does

not give an exact amount of hCG presents in the urine sample. The most widely available home

pregnancy tests are hCG urine tests (Human Chorionic Gonadotropin (HCG), 2014).

Home Pregnancy Test

After the egg is fertilized and implants to the uterine wall, hCG secretion starts and gradually increases

the concentration. hCG hormone enters to the blood stream and it is filled by the kidneys to the urine.

But in the first several weeks, hCG concentration in the urine may not be sufficient to be detected and

give a positive result. For an accurate result, hCG hormone test should be performed when the woman

22
misses her first menstrual cycle. By that time, hCG level increases upto 50 mlU/ml. The accuracy of the

test increases with each passing week. Sometimes, the first hCG urine test performed after the missed

menstrual cycle may become negative. But that result does not imply that the woman is not pregnant. If

the pregnancy is still suspected, another urine test should be done after two weeks.

Any of the urine sample is appropriate for the hCG urine test. But, the first morning urine is the best

sample as it’s concentration of hCG is high and during the day urine sample is diluted (Pregnancy testing,

2016).

hCG Blood Test

hCG blood test can be performed to detect pregnancy or hCG related to certain cancers. Further, the hCG

level in the blood can be used to screening of birth defects. More often, this tests are performed to

diagnose down-syndrome in first trimester along with an ultrasound measurement of fetus’s neck and a

test for protein called pregnancy-associated plasma protein A (Human Chorionic Gonadotropin (HCG),

2014).

There is no any special preparation to be done for the test (White, Zieve, and Ogilvie, 2014).

There are two types of hCG blood tests.

1. Qualitative hCG test measures whether hCG is present in the sample/not.

2. Quantitative hCG test (or beta hCG) test measures the specific level of HCG in the blood.

(American Pregnancy Association, 2012)

There are several minor risks and discomforts related to hCG blood tests as given below.

Excessive bleeding

Fainting or feeling light-headed

Blood accumulating under the skin (hematoma)

Infection (White et al., 2014)

23
Low hCG level may indicate, miscalculation of pregnancy dating, threatened spontaneous abortion,

incomplete miscarriage, blighted ovum or ectopic pregnancy. Ecotopic pregnancy is the implantation of

fertilized egg in a place such as fallopian tube rather than implants in uterus (American Pregnancy

Association, 2012). But, low hCG level always not a sign of a problem. Normal pregnancies which ends

with a perfectly healthy baby delivery may also show low hCG level.

High hCG level may indicate, miscalculation of pregnancy dating, molar pregnancy or multiple

pregnancy such as twins or triplets (American Pregnancy Association, 2012). Further, it may indicate

choriocarcinoma of the uterus, hydatidiform mole of the uterus, ovarian cancer or testicular cancer (only

in men) (White, Zieve, and Ogilvie, 2014).

If the hCG level given is high/low, another test should be performed within 48-72 hours.

Table 1 . 3 : hCG Level Variations with Gestational Age.

Gestational Age hCG level (mlU/ml)

3 5-50

4 5-426

5 18-7340

6 1080-56 500

7-8 7650-229 000

9-12 25 700-288 000

13-16 13 300- 254 000

17-24 4060-165 400

25-40 3640-117 000

24
Ultrasound Testing

The result of an ultrasound after 5 -6 weeks gestation is more accurate than hCG testing. However, a

diagnosis should not be made by ultrasound test until the hCG level has reached at least 2,000.

1.2.3 False Pregnancy Detections

hCG based pregnancy tests can give false negative/positive results due to several cases.

hCG tests give negative results if the test is performed too early. In that case hCG levels are too low to

be detected. The urine is too diluted to contain sufficient hCG (Pregnancy testing, 2016).

hCG tests give false positive results in blighted ovum cases which is known as anembryonic pregnancy.

In such cases, though the fertilized egg implants in the uterus, it does not develop into an embryo. But

the gestational sac develops. The poor quality of the egg or the sperm or abnormal cell division leads to

this no embryo pregnancy. Often, it causes early pregnancy failure or miscarriage. Women with bighted

ovum experience pregnancy signs and positive pregnancy test results as hCG level increases. Therefore,

ultrasound scanning is necessary to diagnose the empty placenta.

Further, hCG test results affect by cancers and early miscarriages. Some antibodies and hCG

containing medications often used in fertility treatments, may also affect the test results

(Blighted Ovum: Causes, symptoms, and more, 2005).

25
1.3 Computational Pulse Analysis

Computational analysis is needed for further development of captured raw data.

1.3.1.Plethysmography

Plethysmography is a non-invasive method based on the graphical recording of changes in body with

respect to heart pulses. In other words, this is a technique used to screen arterial and venous pathologies.

Photoplethsmography (PPG) comes under plethsymography is based on the difference between the

incident light wave and the reflected light wave which depends on the number of red blood cells in the

cutaneous microcirculation.

The original wave captured is known as the volume capacity photoplethysmography. It is difficult to

analyze the course of wave change of volume capacity photoplethysmography. The first derivation of

the volume capacity photoplethysmography is known as velocity capacity photoplethysmography.

However, this wave patterns are difficult to understand. Thus, the second derivative of the velocity

capacity photoplethysmography is taken which is known as accelerated photopleythsmography. It is the

best wave patterns to analyze (Singh & Bansal, 2014).

1.3.2 Pulse Wave Velovity

During a single systole blood propagates through the arteries due to energy transfer of potential energy

of stretched arteries to kinematic energy. Accordingly, changes in blood pressure, blood flow, blood

velocity and blood profile occur. Thus, pulse Wave Velocity (PWV) in the arteries of blood circulation

can be used as an important phenomena to obtain the above mentioned changes. It is a quick and objective

analysis of vascular system functioning. PWV system determines the shape of the digital volume pulse

26
waveform using high fidelity photo-plethysmography transducer with signal conditioning circuit (Pulse

wave analysis, 2014, Pulse wave velocity (PWV) analysis, 1988).

1.3.3 Heart Pulse Sensors

Heart pulse sensing by using electronic equipment is important in many fields such as health care, athletic

training and etc. for monitoring the pulse activity of the heart. The heart pulses are directly related to the

physical conditions of the person. Thus, it is important to provide an accurate measurement of pulses.

Electronic unit worn on the wrist is one of the most useful devices which is used to analyze heart pulse

(Bryars, Cavanaugh, and Corporation, 1995).

There are different type of sensors used in wrist pulse analysis. US. Pat. No. 4,120,269 (Prinz) describes

a digital plethysmography which utilizes an infrared light transducer.

US. Pat. No. 4,059,118 (Stupay) and US. Pat. No. 4,224,948 (Cramer) describes a device which uses a

piezoelectric crystal.

US. Pat. No. 4,409,983 (Albert) describes a complex arrangement of piezoelectric sensors to develop a

relatively noise free signal which is presented to the input of a microprocessor.

US. Pat. No. 4,781,195 describes a device which uses an optical sensor system for cancelling the ambient

light signal in a blood oxygenation monitor (Britton and Corporation, 1995).

27
Infrared Sensors

Infrared sensors are a type of highly sensitive electronic blood pulse sensors. The sensor emits pulsed

infra-red light waves with specific wave lengths which can be absorbed by blood constituents. A light

detector is used to detect the reflected waves (Sweeney & Unisen, 1976).

Figure 1 . 7 : Technique of Capturing the Pulse Signal Using an Infrared Light Emitter
Diode and a Photo Transistor.

28
1.3.4 Arduino

Arduino is a microcontroller-based development board. It is an open-source prototyping platform. A

variety of simplified arduino development boards come to the market.

Arduino IDE, that is getting started page for installation, utilizes the C programming language. Hence, a

massive Arduino Library which is continually enhancing due to open-source community can be accessed.

Arduino Uno
Arduino Mega
Figure 1. 8 : Outward Appearance of Different Types of Arduino Boards
(Arduino, 2015)

There are different types of Arduino boards namely Arduino UNO and Arduino Mega (Arduino, 2015).

(Figure 1.6)

1.3.5 Methods of Signal Enhancing

Wrist pulse measurements contain artifacts due to the body motions. Therefore, the accuracy is lessen

due to the noise. Hence, it is important to filter out and electrically cancel the noise maximally in the

heart pulse frequency band. It is effective when considerable signal-to-noise ratio is obtained (Bryars,

Cavanaugh, and Corporation, 1995).

29
MATLAB stands for “MATrix LABoratory” which is a matrix based tool. (Anon n.d., 2012) Further, it

is a high-performance language for technical computing. It was initially written as an easy access to

softwares developed by the LINPACK (linear system package) and EISPACK (Eigen system package)

projects. (Houcque, 2005) It can be used for numerical computation and visualization. MATLAB is

generally fast to write and run programs manipulated on array-based data. MATLAB is an advanced

software over conventional computer languages such as C or FORTRAN. But unfortunately, MATLAB

is not appropriate for very large arrays or lots of computations.

Sophisticated data structures, built-in editing and debugging tools and object-oriented programming are

the factors which make MATLAB an excellent tool for research. Further, it is more collaborative as it’s

basic data element is an array that does not require dimensioning.

The MATLAB software is commercially available since 1984. Currently, it is considered as a standard

tool worldwide (Houcque, 2005)

Processing is a flexible software sketch book and a language for learning how to code within the context

of the visual arts. Since 2001, Processing has promoted software literacy within the visual arts and visual

literacy within technology.

1.4 Currently Available Systems for Pulse Analysis

Quantification of the dynamics of physiological signals namely EMG, ECG and EEG are done by number

of developed nonlinear methods and meaningful results have been achieved where pulse waveform

acquisition was less considered. Nevertheless, now researches are done in that area broadly (Walia &

Singh, 2010).

Aniruddha Joshi et al. designed Nadi Tharangini, consist of three flexible diaphragms element provided

with strain gauge and a high grade industrial transmitter cum amplifier of output 4-20 mA, is used to
30
acquire analog wrist pulse signals on vata, pitta and kapha. Sample rate and the quantizer of the acquired

signal are 500 Hz and 16-bit quantizer respectively. The 16-bit digitizer is used to convert the output of

the amplifier to 2-10 V by using a 500 ohm resistor. The data acquired using Nadi Tharangani is

frequently corrupted because of the explicit and implicit electrical and electronic noise. Nevertheless, the

developed system with proper shielding reduces the noise level to a negligible value (Joshi et al., 2007).

Figure 1 . 9 : Hardware Components of “Nadi Tharangani”

which is Developed to Acquire Wrist Pulses (Joshi et al, 2007).

Ashok Bhat et al. describe a developed Nadi Tharangani system (US 20100152594 A1). This system is

able to capture small pressure units accurately and reproducibly. They have obtained waveforms which

are clear of noise and leads to an accurate diagnosis. Six pulse waveforms are acquired, three from each

hand of the subject. To capture the exact values, a slight air gap is introduced between the each sensing

element and the skin of the subject. Rigorous machine learning algorithms are used to extract the exact

pulse waveform from the general physiological characteristics such as rhythm, self-similar nature and

chaotic nature present in the pulse. Subsequently, pulse waveforms are classified into different types and

sub-types according to the Ayurvedic Literature. They suppose that their invention is helpful to eliminate

the human errors occur in manual Nadi-pariksha. Furthermore, they think that it is helpful in obtaining
31
accurate and quantitative information for diagnosis. Finally, they mention that their invention eliminates

subjective constituents in the diagnostics (Bhat et al., 2008).

Figure 1 . 10 : Comparison between Vata Component of the Pulses of Normal Healthy Person,
Person with Back Pain, Person with Fever and Person with Stomachache Obtained from
Develobed “Nadi Tharangani” System (Patent US20100152594 ). y-axis Shows the Normalized
Amplitude of the Signals and x-axis Shows the Samples (Bhat et al., 2008).

32
The Nadi Yantra designed by Mahendra Kumar et al. is another objective approach towards the Pulse

Diagnosis Science which can be used to analyze pulse waveforms. It consists of a mechanical set-up of

three identical piezo based sensors, amplifiers, data acquisition subsystem used to record the wrist pulse

signals on vata, pitta, kapha and a filtering circuitry. It has three fingers like protrusions and their

positions are adjusted at the tip region as required. It helps to figure out the pre-eminent points for signal

capturing. To get rid of natural damping at the tip, they have used springs attached to protrusions. The

raw signals are collected, then filtered and amplified. Afterwards, these signals are sent to the computer

using BioPac 150TM whose sampling frequency is 1000 Hz (Kumar et al., 2008).

Figure 1 . 11: Flexibity of the Nadi Yantra
System (Kumar et al., 2008).
Figure 1 . 12 : Three Fingers like
Protrusions whose Position can be
Adjusted at the Tip Region, Attached
with Springs to Get Rid of Natural
Damping at the Tip of Nadi Yantra
System(Kumar et al., 2008).

33
S. A. Gangal et al. developed Nadi Parikshan Yantra, a three point radial pulse examination system. Nadi

Parikshan Yantra is also consist of three identical data acquisition channels. Pressure data of radial pulse

is captured at vata, pitta and kapha. To capture three signals simultaneously, three identical pressure

sensors are arranged on an acrylic module. Diameter of the sensor is 10 mm and adjacent sensors kept

approximately 6mm apart. These three sensors are connected to three identical data acquisition channels

by using coaxial cables and three different channels are used to display the acquired signal on the

computer screen (Kalange et al., 2012).

Figure 1. 11 : Sensors Attached to Subject's Wrist and the Other

Hardware Componenets of the Nadi Parikshan Yantra (Kalange et al.,
2012).

N. Roopini et al designed a system for pulse examination by using optical pulse sensors to acquire the

pulse signals from the subject. The pulse sensors are placed side-by-side on the subject’s arm

appropriately and the acquired pulse signals are sent to the microcontroller ATMEGA328 for further

34
processing. Analog-to-digital conversion occurs here and LabVIEW software is used to display the data.

To pre-process the pulse signals, Butterworth low-pass filter with a flat frequency response is used.

Lower cutoff frequency is .0.125Hz

Upper cutoff frequency – 180Hz

Order- 8th order

Cut-off frequency – 180 Hz (max frequency of pulse signals are approximately 160 Hz)

Pre-processed signals are further developed by feature extraction. The considered features for

classification of Vata, Pitta and Kapha of are pulse repetition rate, frequency, amplitude, mean and

standard deviation. They suppose that this system helps the physicians for Nadi pariksha and further

diagnosis and the prognosis of cardiac related disorders. Furthermore, it can be used as a home based

health monitoring system as it is a portable wearable device (N, Shivaram, and Shridhar, 2016).

Kulkarni et al. suggested a system which captures human wrist pulses using three pressure sensors based

on piezoelectric principle as pressure obtained from the sensors are proportional to the electrical signal.

That is, sensors are placed on artery and the pressure exerted on the artery is measured. NI USB-6210

multifunction data acquisition card is used for data acquisition. It has an interface with the computer.

Then, the acquired signals are exported in MATLAB. Next, the amplification, filtering and feature

extraction is done by using digital signal processing techniques. Back propagation algorithm is used to

minimize the errors in classification. This system can be used to detect diabetes either by tridosha

analysis or ANN as soft computing tool. The real time PC based system helps to check the prakriti of the

subject at any time according to the heart rate variability (Kulkarni, 2016).

Singh & Bansal carried out a research using 36 samples to check the accuracy of the automatic feature

extraction in acceleration plethysmography. In this research, they have used three modules of PPG100C

and PPG100C-MRI on index finger, middle finger and ring finger to acquire data and BIOPAC MP
35
System and Acqknowledge Software are used to record the data. The data was collected after 1 ½ hours

after breakfast (before 10.30 am), immediate before and after 15 minutes of lunch (1-2 pm). Data was

recorded for 30 seconds. First, they calculated the parameters manually. That process is time consuming

and susceptible to errors. Thus, an automated method based on an algorithm is developed to detect the

relevant features of PPG in the form of algorithm. For noise removal, they have used digital IIR band

stop filter at default line frequency of 50 Hz and quality factor is 5. Filtered data derivation was done

twice at sampling rate 25. They suggest that their algorithm gives 100% accurate results for PPG of

normal young subjects. Thus, they concluded that automatic feature extraction is much more efficient

and accurate rather than manual process (Singh & Bansal, 2014).

According to Kalange & Gangal, there are various types of pressure sensors such as piezoresistive and

piezoelectric which can be used for human pulse detection. They have used strain gauge transducer to

their study and observed noisy waveforms and dc shift due to the holding pressure which dependent on

change in holding pressure. Moreover, it needed a power source for it’s operation. Even so, piezoelectric

transducer is an active transducer and had a good dynamic response. It didn’t show a dc shift.

Accordingly, they have used piezoelectric transducer and a signal processing circuit for capturing and

processing human pulse. They used a small size of sphygmomanometer cuff to place the sensors. Since

the data was in mV range amplification was done and filtering circuitry was used to cancel out the noise.

Cut-off frequency and the gain of the filter/ amplifier circuit is 100 Hz and 10 respectively. Power line

interference noise was removed by using a notch filter of 50 Hz. They compared their data sample with

the standard pulse pattern for Vata, Pitta and Kapha in literature. Finally, they consulted Ayurvedic

practitioners and discussed. They conclude that Vata, Pitta and Kapha can be classified by observation

of pulse repetition rate of waveforms (Kalange & Gangal, 2007).

Chen et al. proposed a new wrist pulse signal time series analysis. They followed a normalization process.

Accordingly, a reference value is selected, using a newly collected data sample of healthy persons. Then,
36
they constructed an autoregressive (AR) model with the selected reference value. As the next step, the

residual error (difference between the actual measurement of the new signal and the by reference signal)

was calculated and considered as the disease sensitive feature. To demonstrate the applicability of the

system, wrist pulse signals were acquired using a Doppler ultrasound device and a database was created.

The analyzed results showed the accuracy over 82% in distinguishing healthy persons from patients with

acute appendicitis and more than over 90% for other diseases. Thus, they concluded that wrist pulse

analysis can be used to indicate patients of specific diseases among healthy subjects (Chen et al., 2011).

37
CHAPTER 02

38
2.0 Materials and Methods

The details of the materials used and the experimental methods followed for detecting the human pulse

at radial fossa and the signal conditioning methods have been given in this chapter.

According to the literature review and the power analysis we selected our sample size as 100, 50 non-

pregnant healthy women and 50 pregnant healthy women within 0-8 weeks of pregnancy in the 20-30

age group.

2.1 Materials

Heart pulse sensors, 16-bit analog to digital 4 channel converter, micro sd card and holder, real time

clock and Arduino UNO board are used to assemble a small device to capture the wrist pulse signal of

subjects. Here, each component is described briefly.

2.1.1 Heart Pulse Sensor

First implementation and developments of the heart pulse sensor as an open source hardware project, is

done by Joel Murphy and Yury and it is used in the device developed to capture wrist pulse signals after

further modifications.

It is a very small sensor (Diameter ~16mm, overall thickness ~3mm, cable length 60cm) (Figure 2.1).

The voltage range is 5V. Current consumption of the sensor is approximately 4mA at 5V. Ambient light

sensor Avago (APDS-9008) and green super bright reverse mount LED from Kingbright

(AM2520ZGC09) is used. Moreover, filtering and amplification circuitry and a normalizer around

reference point V/2 are added to make heart pulse sensor more effective.

39
 Figure 2 . 1 : Heart Pulse which is Used to Measure the Rhythm of Blood Flow
on Radial Artery.

Built-in hysteresis for rising and falling thresholds which can be adjusted as required is introduced to the

sensor. Analog signal from the pulse sensor rises above mid-point when capillary tissues is filled with

surge of fresh blood and then it falls below mid-point.

Op Amp circuit that invariably uses a photodiode as a current source is the current to voltage converter.

Ambient light sensor (Avago APDS-9008) used in heart pulse sensor sensitive to the visible spectrum

and very small. Super-bright Green LED (Kingbright) is used because the sensitivity of this sensor is 565

nm. Specialty of this sensor is that it is a surface mount LED which is shorter than the sensor and does

not cause saturation. Only the sensor sides the skin through a small hole (~2mm). Thus, the sensor can

40
be placed very close to the skin, blocking out ambient light. Moreover, it becomes more immune to noise

resulting from movement.

2.1.2 16 Bit Analog to Digital Converter - ADS1115

ADS1115 is précised and easy to implement, analog-to-digital converter with 16 bits of resolution. It is

offered in an ultra-small, MSOP-10 package and it consists of onboard reference and oscillator.

ADS1115 is operated by a single power supply ranging from 2.0V – 5.5V. It works optimally under 40°C

to +125°C.

Captured data is transferred through a compatible serial interface of 12C where four 12C slave addresses

can be selected. Moreover, ADS1115 can achieve conversions upto sampling rate of 1000 samples per

second (SPS). ADS1115 allows large signals as well as small signals to be captured with high resolution

as it has an onboard PGA with input ranges from the supply to as low as ±256mV.

There are two different modes of operations, namely continuous conversion mode and single-shot mode.

Single-shot mode is very efficient in current consumption during idle periods, as it automatically powers

down after a conversion.

Figure 2. 2 : The Outward Apperance of the 16 Bit Analog to Digital Converter - ADS1115.

41
2.1.3 Arduino UNO

Arduino UNO board with 32kb memory is used in our device which is developed to capture the wrist

pulse signals. It is powered via the USB connection or with an external power supply and the power

source is selected automatically. Arduino Uno reads data from ADC - (ADS1115). Then it sends data to

computer via a USB port (virtual serial port). Further processing is done by using this digitized data.

Software in Arduino is used to program the Arduino board by using a PC.

Arduino UNO provides an extra layer of protection from short currents and over currents to the USB port

of the personal computer by it’s resettable poly-fuse. This resettable poly-fuse disrupt the connection

automatically if more than 500mA current is passed until the short current or the over current is removed.

The optimum external voltage supply to the Arduino UNO board is 7-12 V. However, if the power is

supplied less than 7V, the 5V pin supplies less than five volts. Thus, the board becomes unstable.

Moreover, if the power is supplied more than 12V, the voltage regulator may damage due to overheating.

Fig 2.4 Arduino Uno board with 32kb memory

Figure 2 . 3 : Outward Appearance of Arduino UNO.

42
2.1.4 Real Time Clock (DS3231)

Real time clock (DS3231) is a high quality 12C real-time clock which has a high accuracy. It contains

an integrated temperature-compensated crystal oscillator (TCXO) which improves long term accuracy

and reduces the number of components of the production line, a crystal and a battery input. Thus, it can

be disconnected from the main power supply and use as a accurate timekeeper. DS3231 can perform in

commercial and industrial temperature ranges. Furthermore, it carries the informations such as seconds,

minutes, hours, day, date, month, and year. DS32321 can be operated either in 24 hours time format or

12 hours time format. It consists with two configurable alarm clocks. The calendar can be set to a square

wave output. I2C bidirectional bus is used to serial transfer of address and data. Moreover, it consists of

a very precise digital temperature sensor with the accuracy is ± 3 ° C. Additionally, if the main power

fails, accurate timing and temperature is maintained by connecting to a external battery power supply.

Arduino board is used with ArduinoIDE with a SD card to complete the library card initialization and

read-write directly.

43
2.2 Methods

The experimental methods followed for detecting the human pulse at radial fossa and the signal

conditioning methods is described under this sub topic.

Initiation

Device to Collect Pulse Data from the

Subject Group

Device Development

Data Collection

Signal Enhancing

Feature Extraction

Figure 2 . 4 : Flow Chart of the Methodology.

2.2.1 Initiation

First and foremost, two Ayurvedic practitioners were contacted and discussed about the preganancy

detection methods. According to them, there are four places where pulse can be sensed namely, wrist,

base of the little finger, underarm near the breast and lower abdominal. However, wrist is the most

suitable and comman place where pulse is sensed to detect pregnancy. Moreover, it is difficult to acquire

44
the data from the under arm near the breast and lower abdominal in this stage due to the ethical

considerations. Thus, wrist pulse sensing is studied and precautions to be taken before acquiring the data

is noted. Additionally, it is decided to acquire the finger tip signal simultaneously as finger tip signals

contains less noise and base of the little finger is also used to diagnose pregnancy. Accordingly, a

questionnaire was prepared to be filled by Ayurvedic doctors (See the annex v - Questionnaire to

Ayurvedic Practitioners - English). After data collecting from Ayurvedic practitioners, a questionnaire

was prepared to be filled by the subject group with the aid of feedbacks of Ayurvedic practitioners as

well as western medicine practitoners (See the annex viii - Questionnaire to volunteer participants -

English).

2.2.2 Device to Collect Pulse Data from the Subject Group

A small device consists with an Arduino Uno board, heart pulse sensor and an USB cable is designed to

capture the wrist pulse signal and the finger tip pulse signal.

The circuit of the heart pulse sensor is shown in the annex xii.

There are three connections in between the sensor and the Arduino Uno board, analog signal, positive

and negative. In our device, a purple colour wire is used for analog signal, white color wire is used for

positive and black colour wire is used for negative.

The analog signal captured by the heart pulse sensor is sent to the Arduino Uno and it reads the analog

value for every milisecond. That is, the sampling rate of the device is 1 kHz. We considered the accurate

result obatainable range for the healthy subjects with high heart rate variability when setting this value.

Generally, the sampling rate of the heart pulse is greater than 125 Hz to fulfil the above requirement.

As the raw data contains noise and the signal values are too small to analyze, electronic filtering and

amplification is needed.

45
Band Pass Filter

Passive low pass-filter is formed for noise reduction as shown in figure 2.6. The maximum frequency we

obtained in the raw data is approximately equals to 2.0 Hz. Thus, cut-off frequency is set as ~4.64Hz by

using R2 and C2/C1. Accordingly, amplitudes of the signal component that are above the cut-off

frequency is subtracted. However, when R2 and C2 combines with the load impedance as shown in the

figure, a high pass filter is formed. Thus, the final result is a band-pass filter. The center frequency of

this circuitry is about 2.1Hz.

Figure 2 . 5 : Schematic Diagram of Passive Low Pass-filter is Formed for

Noise Reduction.

OP-AMP

OP-AMP used in this stage of signal enhancing is an active band-pass filter with a gain. Though the gain

in isolation is 47dB, the overall result reduces to 37.5dB when the two stages are combined.

46
 Figure 2 . 6 : Schematic Diagram of OP-AMP.

Figure 2 . 7: Difference between the Raw Signal Captured by the Heart Pulse Sensor and
the Signal Obtained after Filtering and Amplification.

47
ArduinoIDE is used to program the Arduino. Cut-off Values are set and heart rate calculation is done

using program. To set the cut-off Values, device is placed on the subject’s finger and the output value is

observed for 10 subjects. Accordingly, upper cut-off ADC value is set as 560 and lower cut-off value is

set as 400. Three sub programs are written using ArduinoIDE as follows.

Time Interrupt

Sampling time is set to 1ms.

Data Storage

Captured data is store in SD card when use as standalong device.

Serial Handling

To transfer the captured real time data and off line data to the computer.

2.2.3 Device Development

As we observed that the sensor gives different values at vata, pitta, kapha, our device is modified using

three heart pulse sensors to capture the pulse signal simultaneously at vata, pitta, and kapha. Moreover,

sensors were integrated to a non-inflatable sphygmomanometer cuff whose size can be adjusted

according to the subject’s hand size to minimize the noise. As one of our sensors is damaged due to

contact with moisture of the skin, three sensors were covered with a small thin film of vinyl to seal the

sensor side. This vinyl layer does not affect to the data capuring.

Figure 2 . 8: Arduino UNO Board and the Three Heart Pulse Sensors of the Assembled Device.

48
In the developed device, ADS1115 is used instead of Arduino Uno in built ADC as the ADC resolution

is high (16bit). The analog signal captured by the heart pulse sensor is sent to the ADC. 1ms is used as

the sampling time of the device to acquire the maximum information in the data sample. Then, ADC

converts analog signal to a digital signal. Though our output should laid in between 0-65, 535 as we used

16 bits, our output laid in between 0-32,767. The reason for the above mentioned factor is that the output

of the ADS1115 is a signed integer. Therefore, 1 bit is assigned for sign and only 15 bits are used to

communicate the magnitude of the measurement. Thus, the output value is reported as a positive or

negative value.

Accordingly, upper cut-off value is set as 28 800 and lower cut-off value is set as 9300.

The following sub programs were written to acquire data through ADS1115 and DS3231.

ADC Value Reading

Digitaliezed 16 bit ADC Value read from ADS1115 using I2C communication.

Real Time Clock Reading

Real time data value read from DS3231 Clock using I2C communication.

ArduinoIDE is used program the device with a 20s resting time period and 1 minute of data recording

time period. That is, once the program is started running, data recording starts after 20s automatically

and continued recording for 1 minute. However, that was not successful as the accuracy of the assumption

of the subject’s hand become stable within 20s is low and there was no any method to confirm it. Thus,

the Processing software is used.

2.2.4 Programming the “Processing” Software

To overcome this problem, real time pulse visualizer window is designed using “processing” software.

Pulse pattern can be observed on the window and we can easily decide whether hand is satable or not.
49
When the hand becomes stable, pulse wave shows a repetitive similar pattern. After seeing that kind of

repetitive similar pattern, the record button on the window is selected to start the data recording and after

1 minute of recording t the stop button on the window is selected to stop the data recording. After

selecting the stop button, the data sample automatically saved in the folder whose path is introduced

before with the given name for that particular data sample.

Figure 2. 9 Processing Software Development

2.2.5 Data Collection

Data is collected from non-pregnant, healthy 50 subjects of women in the age group of 20-30, at the wrist

of the left hand and the little finger tip of the left hand

First, the consent of participating is taken from each participant of the sample group (See the annex

ii/iii/iv- Informed Consent for the Volunteer Participants – English/ Sinhala/ Tamil). Then the

questionnaire is filled by each participant (See the annex viii/ix/x - Questionnaire to volunteer

participants – English/ Sinhala/ Tamil).

50
As the next step, the wrist pulse of the subject is sensed by one of our group members according to the

technique we practiced with Dr. Jeewani Krisanthi, Ayurveda General Hospital, Rajagiriya. Accordingly,

the sensors are attached on the most suitable place with the aid of the cuff. Then, the data recording is

started for 1 minute after observing the stable pulse pattern on the real time pulse visualizer programmed

in “Processing” siftware. After, one minute of data recording, the digitized data is automatically saved

in the database. Thus, the data could be used for further processing and feature extraction later.

Figure 2 . 10: Sensors Attached to the Subject's Wrist with the Aid of
Non-inflatable Sphygomamometer Cuff and Arduino UNO Board
Connected to PC.

After obtaing the wrist pulse signal, subject’s little finger tip was placed on one of the heart pulse sensor

and data sample is saved in the data based using the same procedure.

51
2.2.6 Signal Enhancing and Feature Extraction

Wrist Pulse Signal Analysis

MATLAB software is used to do the feature extraction of the data saved in text format in the database.

Import data to a
Import data Select the text file in
MATLAB as a
the database
matrix

Figure 2 . 11: Importing Data in MATLAB

Data in the matrix form is used for further processing. First, the data is analyzed in time domain. The

raw data is plotted in MATLAB using the “plot” command. Three colours are used to plot the raw data,

one colour for each sensor. Sensor one signal is plotted in red colour, sensor two signal is plotted in green

colour and sensor three signal is plotted in blue colour.

Figure 2 . 12: Colours of the Three Sensors with Respect to the Location

Then the data is processed in the frequency domain by converting data into frequency domain using Fast

Fourier Transformation.

Fast Fourier Transformation

Fast Fourier Transformation was applied to each data. Sampling frequency (Fs) is set as 1000Hz and

signal length is set as 4000. As the next step, all the graphs are plotted in frequency domain. The

maximum amplitude component of each sensor and the respective frequency component is noted.

52
It is noted that the wrist pulse signals contains significant amount of noise. Thus, the several types of

filtering techniques are used to filter out the noise in time domain.

Filtering Techniques

1. Gaussian Filter

1 (𝑥−𝜇)2
−
𝑓(𝑥 ) = 𝑒 2𝜎2
𝜎√2𝜋

f (x) – Fuction

𝜎 – Standard deviation of the distribution

x – Input signal value

𝜇 – Mean value

2. Avarage Filter

Average filtering is used as it is a simple, intuitive and easy to implement method of smoothing.

However, there is a disadvantage of filtering as the most important data which is difficult to differentiate

from noise can be lost. Thus, advanced filtering techniques are being studied to filter out the data without

losing important information.

53
Finger Tip Signal Analysis

First, the data is imported to MATLAB and they are plotted in time domain. Plots of the finger tip signals

are compared with the plots of wrist pulse signals in time domain. Then the Mahanalobis distance

comparison technique is used to group the data samples.

Mahanalobis Distance Comparison

𝑫𝟐 = (𝒙 − 𝒚)𝑻 𝑪−𝟏 (𝒙 − 𝒚)

𝑫𝟐 - Mahalanobis Distance

x – Mean vector of data group A

y – Mean vector of data group A

𝑪−𝟏 - Inverse covariance matrix of independent variables

T – Indicates vector should be transported

54