A wearable prototype of reflective sensor for non invasive measurement of heart rate
©2014
Textbook
72 Pages
Summary
This research study deals with the design and development of non-invasive and reflectance type (photoplethysmogram) wearable PPG prototype for measuring heart rate.It explains the processing of heart rate from the measured PPG signal.Illustrates the heart rate measurements at different measurement sites.Explains benchmarking of realtime measured heart rate with other standard heart rate measuring units.Presents the heart rate measurements during different actions of swimming when the prototype is mounted on the forhead of swimmer.
Excerpt
Table Of Contents
iv
2.7
Measuring heart rate of a swimmer ... 24
2.7.1
Methods to improve the swimming performance ... 24
2.7.2
Devices for monitoring heart rate of a swimmer ... 25
2.7.2.1 Suunto memory belt ... 25
2.7.2.2 Suunto ANT heart rate belt ... 26
2.7.2.3 Instabeat ... 26
3
Methodology ... 28
3.1
Measuring PPG signals ... 28
3.1.1
Proposed methods for building reflective PPG system ... 28
3.2
Improving the raw PPG signal for desired Heart Rate ... 28
3.2.1
Minimizing the DC and AC component ... 28
3.2.2
Frequency limits for the desired heart rate ... 28
3.3
Sampling the AFE- PPG signal by the microcontroller ... 29
3.4
Digitizing the sampled PPG signals ... 29
3.5
Processing digital signals to calculate heart rate ... 30
3.5.1
Technique for minimizing motion artifacts ... 30
3.6
Power supply ... 31
3.7
Measurement unit reliability ... 31
3.8
Probabilistic model of PPG sensor performance ... 32
4
Implementation and Results ... 34
4.1
Block diagram of the prototype ... 34
4.2
Hardware implementation of the PPG sensor system ... 34
4.2.1
Design models of the PPG sensor system ... 34
4.2.1.1 TCRT1000 Phototransistor-PPG sensor system ... 34
4.2.1.2 NJL5303R-PPG sensor system ... 35
4.2.2
Improving PPG signal for the desired heart rate ... 36
4.2.2.1 NJL5303R circuit simulation in NI multisim ... 36
4.2.2.2 First stage of signal conditioning ... 37
4.2.2.3 Second stage of signal conditioning ... 37
4.2.3
Strip board implementation of the circuit ... 38
4.2.3.1 Electronic component selection ... 38
4.2.4
PCB of the Analog Front End-PPG sensor system ... 40
4.3
Complete wearable measuring unit ... 40
4.4
Software implementation ... 40
4.4.1
Sampling the AFE-PPG signal ... 40
Flowchart in Figure 33 illustrates the steps needed for sampling an analog signal by
the CC430F5137 microcontroller ... 40
4.4.2
Determining frequency of the PPG signals ... 42
4.4.2.1 Minimizing motion artifacts ... 43
4.5
Algorithm for removing motion artifacts ... 47
4.6
Monitoring the measured heart rate ... 48
4.6.1
Real-time monitoring ... 48
v
4.6.2
Post user activity monitoring ... 49
4.7
Validating the measurement unit performance ... 49
4.7.1
Benchmarking ... 49
4.7.1.1 Heart-rate monitoring tests ... 51
4.8
Multiple applications ... 52
4.8.1
Wrist and finger ... 52
4.8.2
Forehead ... 54
4.8.2.1 Integrating sensor with a headband ... 54
4.9
Validating prototype reliability ... 55
4.9.1
Selection of activities and Prototype users ... 55
4.9.2
Factors affecting the measurement system ... 56
4.10
Monitoring the heart rate of swimmer ... 57
4.10.1
Testing the entire prototype during swimming ... 57
4.10.1.1 Measured PPG data ... 58
4.10.1.2 Calculated heart rate ... 59
4.11
Outcome ... 64
4.12
Future work ... 65
5
References ... 67
vi
List of Figures
Figure 1: Reduction in the size of pulse meters by the advancement in the development [1] ... 10
Figure 2: Changes in blood pressure [30] ... 13
Figure 3: Transmission mode of Photoplethysmography [1] ... 14
Figure 4: Reflection mode of Photoplethysmography [1] ... 14
Figure 5: TRCT1000 photo transistor ... 15
Figure 6: NJL5503R photo reflector [10] ... 16
Figure 7: Working principle of NJL5503R [10] ... 16
Figure 8: Methods of sensor attachments ... 19
Figure 9: IMU developed at Acreo Swedish ICT ... 20
Figure 10: Pin configuration of CC430F513x [28] ... 21
Figure 11: A still of the CCS window ... 22
Figure 12: MSP-ez430 Debug Interface ... 22
Figure 13: RF Polar H6-Heart rate monitoring device [19] ... 23
Figure 14: Polar FT7-Heart rate monitoring device [19] ... 24
Figure 15: Suunto heart rate monitoring device [21]... 25
Figure 16: Suunto ANT heart rate monitoring device [25] ... 26
Figure 17: Instabeat-Heart rate monitoring device ... 27
Figure 18: Method to read AFE-PPG signal by the microcontroller ... 29
Figure 19: Conversion of sampled PPG signals to digital signal ... 30
Figure 20: Methods to determine the frequency ... 30
Figure 21: Measurement gaps formed by the biking activity of user ... 32
Figure 22: Uptime and repair time of the sensor system ... 33
Figure 23: Block diagram of the measurement unit ... 34
Figure 24: PPG sensor system implemented with TCRT1000 ... 35
Figure 25: PPG sensor system implemented with NJL5303R ... 35
Figure 26: Output of NJL5303R phototransistor in lab view ... 36
Figure 27: Circuit simulated in NI multisim ... 37
Figure 28: Simulation result in NI multisim ... 38
Figure 29: Strip board implementation of the circuit ... 39
Figure 30: Outputs of the signal conditioning stages ... 39
Figure 31: PCB of the AFE-PPG sensor system ... 40
Figure 32: Complete wearable heart rate measuring unit ... 40
Figure 33: Method for sampling the analog signal ... 42
Figure 34: Determining frequency of the signal ... 42
Figure 35: Method to remove the frequency components of PPG signal greater than 2.34Hz ... 44
Figure 36: Result after removing the frequency components greater than 2.34Hz ... 45
Figure 37: Method to remove the frequency components of the PPG signal less than 0.7Hz ... 46
Figure 38: Result after removing the frequency components less than 0.7Hz ... 47
Figure 39: Verifying the implemented motion artifacts algorithm ... 48
Figure 40: Monitoring the real-time heart rate using USB serial communication ... 49
Figure 41: MATLAB GUI for monitoring the real-time heart rate and PPG signals ... 49
Figure 42: Heart rate measuring devices used for benchmarking [26], [27] ... 50
Figure 43: Physical environment of the connected devices ... 50
Figure 44: Comparing the heart rate measured with finger by Acreo sensor, Shimmer sensor and
AFE4490 ... 51
Figure 45: Sensor attachment methods of finger ... 52
Figure 46: Heart rate measured with finger by Shimmer sensor and ... 53
Figure 47: Integrating sensor on forehead with a headband ... 54
Figure 48:Heart rate measured on forehead by Acreo sensor and with finger by Shimmer sensor ... 55
vii
Figure 49: Integrating entire measurement unit on forehead with swimmer cap ... 57
Figure 50: Desired interval for monitoring heart rate of the swimmer ... 58
Figure 51: Heart rate measured during the entire swimming session ... 59
Figure 52: Measured PPG signals and heart rate during the walking ... 60
Figure 53: Phase1 of resting interval1 ... 61
Figure 54: Phase1 of resting interval2 ... 61
Figure 55: Phase1 of resting interval3 ... 62
Figure 56: Phase1 of resting interval4 ... 62
Figure 57: Phase1 of resting interval5 ... 63
Figure 58: Phase2 of resting interval4 ... 63
Figure 59: Measured PPG signals and heart rate during swimming interval3 ... 64
Figure 60: A smart-patch prototype for embedding the measurement unit ... 65
viii
List of Tables
Table 1: User activities ... 56
Table 2: User activities and their time duration ... 58
ix
Notation
ADC
Analog-to-Digital
Converter
AFE
Analog Front End
AFE-PPG
Analog Front End- Photoplethysmogram
ANC
Adaptive Noise Cancellation
BPM
Beats per Minute
CCS
Code Composer Studio
DST
Discrete Saturation Transform
GUI
Graphical User Interface
IAR
Ingenjörsfirman
Anders
Rundgren
ICA
Independent Component Analysis
IMU
Inertial Measurement Unit
PCB
Printed Circuit Board
PC
Personal Computer
PPG
Photoplethysmogram
RMSE
Root Mean Square Error
SOC
System on Chip
TI Texas
Instruments
USB
Universal Serial Bus
USCIs
Universal Serial Communication Interfaces
10
1
Introduction
Health monitoring has been an important field of interest over decades. The rise of
need to continuous measure and assess all standard vital signs remotely and to
monitor their trend over time is essential for the development of physiological tele-
monitoring [1]. The vital signs being the heart rate, respiratory rate, blood pressure,
body temperature, and oxygen saturation level in blood (SPO2). The emerging
advances in the field of electronics, particularly with hardware miniaturization of
devices measuring vital signs led to the development of wearable devices[Figure
1].For example, the development of compact and light-weight wearable devices
could facilitate remote noninvasive monitoring of vital signs. The main benefits of
deploying the mobile technologies in the field of medical care are [2]:
· Improve patient safety
· Decrease the risk of medical errors
· Increase physician productivity and efficiency
Figure 1: Reduction in the size of pulse meters by the advancement in the development [1]
1.1
Background and problem motivation
There have always been advances and improvements to the developed things in
every field with the growing technology. When it comes to the field of health
monitoring, the advancements in technology has led to the development of different
types of heart rate measuring devices. The available heart rate measuring devices can
be categorized into the following two types:
11
1.
Measures heart rate only when the user is steady, any kind of user's move-
ment will result in the erroneous heart rate measurements.
2.
Measures heart rate of the user during in steady and moving states by using
special techniques to detect motion artifacts and to reject them. The tech-
niques used by these devices have high computational complexity. They also
exhibit issues with the sensor attachment.
There is a real need to develop a wearable heart rate measuring unit to over-
come the limitations of available heart rate measuring devices.
1.2
Overall aim
The aim of the research work is to develop a miniaturized wearable PPG sensor
system for measuring PPG signals and a low computational complexity algorithm to
measure the heart rate of a user. And to develop a MATLAB graphical user interface
(GUI) to display the measured PPG signals and heart rate, when:
1. The measurements are carried out by connecting the measurement unit to a per-
sonal computer (PC) via Universal Serial Bus (USB) interface.
2. The measurements are stored on flash memory of IMU, which will be later read
back into a PC using the USB serial communication between the measurement unit
and PC.
1.3
Scope
The tasks needed for this research implementation are listed below:
1. Design and developing a sensor unit for measuring the PPG signals.
2. Sampling PPG signals of sensor unit by on-board microcontroller of IMU de-
veloped in-house at Acreo Swedish ICT.
3. Processing the sampled signals on IMU to calculate the heart rate.
4. Establishing a USB serial communication between the measurement unit and a
PC.
5. Developing a MATLAB GUI to display the measured PPG signals and heart
rate.
1.4
Concrete and verifiable goals
Study different heart rate measuring devices to build a low power consuming and
miniaturized PPG sensor system.
Study the effect of motion artifacts on the measurements and to implement an algo-
rithm for removing the motion artifacts. Verifying the device working with other
standard heart rate measuring devices
1.5
Outline
The following presents the summary of each of the remaining chapters:
1. Theory / Related work gives the knowledge needed for a reader.
12
2. Methodology illustrates the methods used for the development.
3. Implementation and Results presents the implementation process of both
hardware and software along with their results. It also presents the achieved
results comparison with the other pulse meter devices.
4. Conclusions and Future work provides an executive summary of the main
achievements of this project. It also discusses the future work.
1.6
Contributions
The investigations, design of PPG sensor system and mounting the components on
the designed PCB board was carried out by the author. The PCB of the PPG sensor
system was designed by a hardware engineer of Acreo Swedish ICT. The techniques
for removing motion artifacts were investigated and implemented by author. The
program used for saving the data on flash memory was developed by Acreo Swedish
ICT.
Integration of PPG sensor system with the IMU board, serial communication be-
tween the IMU and host PC for providing a user interface for monitoring the real
time heart rate was also carried out by the author.
13
2
Theory / Related work
This chapter discusses the background and introduces the reader to the related work.
2.1
Introduction to photoplethysmography
Photoplethysmography refers to the non-invasive measurement of blood volume in a
specified region. The volume of blood in a specified region increases in the systole
phase and decreases in the diastole phase during the cardiac cycle of heart as
illustrated in Figure2.1. This changing blood volume can be directly used to calculate
the heart rate and also to measure other characteristics of cardiovascular function.
The basic PPG sensing system consists of a light source to illuminate the
blood vessels and a photo detector to sense the received light that is a result of optical
absorption and scattering properties of the blood, tissue and bone. The PPG signal
consists of two components referred to as AC and DC as shown in Figure2. The AC
component is caused by the pulsatile changes in arterial blood volume and is
synchronous with the heart beat because of which it can be used as a source for the
heart rate information. The DC component is caused by the tissues and average
blood volume that superimposes with the AC component. The DC component
should be removed from the whole signal to get desired information of heartrate
from the AC component.
Figure 2: Changes in blood pressure [30]
2.2
Construction of photoplethysmography sensor system
Construction of wearable PPG sensor system depends mainly on the following two
factors:
The location of the sensor and the way it is attached to the user.
14
Figure 3: Transmission mode of Photoplethysmography [1]
A PPG sensor can be placed at any place that has a blood flow. Depending on the
location of the sensor, the construction of PPG sensor system can be made in the
following two different modes:
Transmission mode: The photo detector and LED are placed on the opposite sides of
the tissue to be measured. The photo detector measures the amount of light that was
not absorbed as illustrated in the Figure3.
Reflection mode: The photo detector and LED are placed on the same side of the
tissue to be measured. These measures the amount of light backscattered from the
skin and capillaries. This is illustrated in the below Figure4.
Figure 4: Reflection mode of Photoplethysmography [1]
2.2.1
TRCT1000-phototransistor with infrared LED
The TRCT1000 is a reflective optical sensor that has included both the infrared light
emitter of wavelength of 950nmand phototransistor side by side in a leaded package
such that it has less effect from the surrounding visible light.
15
Figure 5: TRCT1000 photo transistor
2.2.2
NJL5303R-phototransistor with green LED
The NJL5303R is a reflective optical sensor that includes both the green LED of
wavelength of 570nm and a photo transistor in a small package that are well suited
for pulse detection. In general, the green light has a higher reflective factor than the
factor of infrared light which provides more sensitive detection and high signal to
noise ratio [10].
The NJL5303R's green phototransistor has high sensitivity to the measuring pulse
waves as illustrated in Figure6; with a green circle corresponding to the wavelength
of the green wavelength spectrum.Figure7 depicts the working principle of
NJL5503R.
16
(a) Sensitivity of photo
reflector (b) Pictorial view of the photo reflector
Figure 6: NJL5503R photo reflector [10]
Figure 7: Working principle of NJL5503R [10]
2.3
Issues with PPG signal measurements
In general, measurements of PPG signals will be affected by different factors as listed
below:
2.3.1
Artifacts
In general, artifact refers to the disturbance in the measured PPG signal. The two
types of artifacts associated with the PPG signal measurements are explained below:
2.3.1.1 Ambient artifacts
The light sources other than the LED included within the PPG sensor system results
in the ambient light artifacts. The indoors fluorescent/incandescent lighting forms the
source of ambient light artifacts if the measurement is done in the laboratory
environment. The other main source will be the sun's light, either coming through a
window or from the sensor being worn while the user is outdoors [1].
17
The sources of artificial light will be generated from the electrical mains supply
having a fundamental frequency of 50Hz or 60 Hz.
2.3.1.2 Motion artifacts
Motion artifact is any corruption of the PPG signal due to the user's motion [1].
Motion artifacts will be resulted from the mechanical distortion of the optical path
between the LED and photodiode of the PPG transducer [1]. This type of mechanical
distortion comes into picture when the measurements are carried out by placing the
PPG sensor on forehead, during which there exists changes in the relative position of
the sensor with respect to the frontal bone of the skull rather than relative
movements of the sensor with respect to the skin. This mechanism results in the
changes of distribution of LEDs backscattered light reaching the photo sensor, thus
leading to the corruption of the PPG signal.
2.3.2
Pressure disturbances acting on the PPG sensor
A too low contact pressure between the PPG sensor and measurement site will result
in distorted PPG signals leading to inaccurate measurements [7]. On the other hand,
a too high contact pressure may result in comprise of blood circulation when the
measurements are conducted for a longer time thus leading to the complete loss of
PPG data.
2.3.3
Physical activity of the user
Its sources can be 1) the formation of air gaps created between the skin and sensor
during the physical activity of user, 2) variation in venous pressure resulted from the
back and forth movement of a user's physical activity[2].
2.4
Minimizing the problems associated with PPG measurements
Following describes the techniques to minimize the effect of problems associated
with PPG signal measurements:
2.4.1
Minimizing motion artifacts
According to different studies conducted to overcome the effects of movement
artifacts, suggested different methods to improve the measurement accuracy when
the user is steady while leaving out the limitations with the measurements during
motion artifacts. The following explains the ways to minimize the motion artifacts:
2.4.1.1 Measurement site
The artifacts explained in the section 2.3.1.2 are dependent on the measurement site.
The study made by Mendelson [3] states that the reflected sensor located on the
forehead provides more consistent results when the user is motionless (steady) and
when a moderate amount of motion artifacts are present as compared to the
measurement carried out in other facial regions. The study of Mannheimer [4]
reveals that the placement of the sensor directly over the eyebrow slightly lateral to
the iris also provides consistent measurement results.
18
2.4.1.2 Signal processing
Signal processing is the most common used methods to overcome the problems of
motion artifacts. There have been numerous algorithms developed for motion artifact
removal, including Independent Component Analysis (ICA) and Adaptive Noise
Cancellation (ANC) that can be applied in general to any sensing systems to remove
noise. The specific methods for solving the motion artifacts of PPG sensor unit as
explained by J.A.C.Patterson [1] are
1) Discrete Saturation Transform (DST) used by the Masimo which is a leading
commercial pulse oximeter manufacturer,
2) Wavelength method proposed by Hayes and Smith,
3) Wavelet transforms method proposed by Addison and Watson.
All the above mentioned techniques are very computationally intensive solutions for
detecting and removing the motion artifacts. There is a need to develop a low
computational complexity technique for detecting and minimizing the motion
artifacts.
2.4.1.3 Sensor attachment
Attachment of sensor on the proper measurement site also plays a major role in
minimizing the effect of motion artifacts. The three widely used methods are [Figure
8]; the first method uses an adhesive tape for attachment and the second uses a
headband for the sensor attachment. The third method is to embed the sensor into
pre-existing equipment's like a soldier's helmet or goggles. According to various
researches, the usage of compressive headband will be the optimal choice as it
presents low pressure venous pulsations and venous pooling when the user is in
Trendelenberg position [2], where the user's body will be lying flat on the back with
the feet higher than the head by 15-30 degrees.
(a)Adhesive attachment [10] (b) Headband attachment [8]
Details
- Pages
- Type of Edition
- Erstausgabe
- Year
- 2014
- ISBN (eBook)
- 9783954898350
- ISBN (Softcover)
- 9783954893355
- File size
- 4.3 MB
- Language
- English
- Publication date
- 2014 (December)
