P13071: Non-Invasive Blood
Glucose Monitor System Level Design Review
Jared Bold, Yongjie Cao, John Louma, Andrew Rosen, Daniel Sinkiewcz
S21 Antenna
Cable to LabVIEW
Device Under Test (DUT)
KGCOE MSD Technical Review Agenda
KGCOE MSD Page 1 of 2 Technical Review Agenda
P13071: Non-Invasive Blood Glucose Monitor
Meeting Purpose:
1. Overview of project 2. Confirm Customer needs and Specifications 3. Review variety of concepts designs 4. Propose our selected design 5. Generate new ideas
Materials to be Reviewed:
1. Project Description and Objective 2. Work Breakdown Structure 3. Customer Needs 4. Customer Specifications 5. Functional Decomposition 6. Concept selection process 7. Risk Assessment 8. Project Plan
Meeting Date: April 5, 2013
Meeting Location: Room 09-4435
Meeting time: 10:00 – 11:30 AM
Timeline:
Meeting Timeline
Start time
Topic of Review Required Attendees
10:00 Introduction for the Project Prof. Slack, Dr. Venkataraman
10:10 Work Breakdown Structure Prof. Slack, Dr. Venkataraman
10:15 Customer Needs Prof. Slack, Dr. Venkataraman
10:25 Customer Specifications Prof. Slack, Dr. Venkataraman
10:35 Functional Decomposition and Block Level Diagrams Prof. Slack, Dr. Venkataraman
10:55 Concept Development and Chosen Design Prof. Slack, Dr. Venkataraman
11:10 Risk Assessment Prof. Slack, Dr. Venkataraman
11:20 Project Plan Prof. Slack, Dr. Venkataraman
KGCOE MSD Page 2 of 2 Technical Review Agenda
Project Description . Project Background: Blood glucose monitoring is a valuable tool not only used by diabetic individuals to maintain a healthy lifestyle, but also by physicians caring for patients. Today's accurate monitoring systems are invasive, requiring direct access to a patient's blood for accurate analysis. The blood glucose meter pricks the finger in order to obtain a drop of blood from which a discrete glucose level can be determined. Continuous glucose measurements can be obtained by placing a glucose sensor subcutaneously, which delivers measurements to an external system. There is a distinct lack of noninvasive glucose monitoring alternatives that can provide the same level of accuracy.
Problem Statement:
Develop a non-invasive real time monitoring system that measures blood glucose. The system should use a microstrip antenna to measure reflection and transmission of a synthesized signal. These measurements should be comparable to a network analyzer.
Objectives/Scope:
Improve accuracy of current noninvasive glucose monitoring system
Provide a method of real time monitoring of patient's blood glucose levels
Archive measurement data for future analysis
Obtain a transmission signal and perform a vector measurement
Deliverables:
Improved accuracy of reflection vector measurements
Compact printed circuit board layout
Introduce calibration system to further increase accuracy measurements
Visual representation of data through LabVIEW graphical user interface
Expected Project Benefits:
Current blood glucose monitors are invasive and discourage multiple measurements. There are many benefits to having a well designed non-invasive blood glucose monitor. It will allow patients to continuously monitor their blood glucose and will encourage the patients to monitor their levels more often. The system will also allow for data archival, so a history of blood glucose levels can be analyzed.
Core Team Members:
Jared Bold
Yongjie Cao
John Louma
Andrew Rosen - Project Manager
Dan Sinkiewicz
Strategy & Approach .
Assumptions & Constraints:
The team will use a two antenna system to measure both the transmission and reflection coefficients of the user's limb. The system must be designed to minimize undesired effects on the antenna system, such as coupling. A calibration system for the measurement circuit will be designed to improve measurement accuracy. The data must be processed and displayed through LabVIEW and compared real time to automated Network Analyzer measurements to verify accuracy.
Issues & Risks
Difficulty measuring transmission through arm
The antennas might couple or the "creeping wave" effect could alter measurements
Peripheral devices could be difficult to integrate successfully
Patient could be at risk of slight shocks if the system short circuits
Summary of Ben Freer’s Thesis
The present work on blood glucose monitors focuses on the possibility of a monitor that
non-invasively measures blood glucose levels using electromagnetic waves. The technique is
based on relating a monitoring antenna’s resonant frequency to the permittivity and conductivity
of blood which in turn is related to the glucose levels.
Using the Agilent 85070E dielectric probe and an Agilent 8720B network analyzer, the
dielectric permittivity and conductivity of twenty different blood samples was measured over a
frequency range of 1GHz – 10GHz. The Cole-Cole model was modified through curve fitting to
in-vitro data that includes a factor representing glucose level. The desired frequency sweep range
for the monitor was then determined to be 200MHz to 2GHz.
An antenna was been designed, constructed and tested in free space. A simulation model
of layered tissue and blood together with an antenna was created to study the effect of changing
glucose levels. It is noted that the antenna’s resonant frequency increases with increase in
glucose levels. An analytical model for the antenna was developed, which was validated with
simulations. A measurement system was developed to measure the resonant frequency of the
antenna. A frequency synthesizer generates an RF signal over the desired frequency range of
200MHz to 2GHz. This signal is sent to the antenna through a directional coupler that generates
forward and reflected signals. These voltages are measured and the reflection coefficient is
calculated with a microprocessor.
As an experimental verification, two antennas were strapped one on each leg of a patient
with one antenna connected to the PNA and the other to the measurement system. As the patient
ingested fast acting glucose tablets, the blood glucose level was measured by a traditional
glucose meter. At the same time, a comparison of the resonant frequency of the antenna
measured by the PNA and by the measurement system showed good agreement. Further, it is
seen that the antenna resonant frequency increases as the glucose level increases, which is
consistent with the simulation model.
Non-Invasive Blood Glucose Monitor
Data Collection Calibration
Power SystemRegulation
Data Measurement
Data Analysis
Power Source
Voltage Regulation
Battery Pack
3.3/5V
Short Protection
Measure open, short, load, through
Low Power Mode/Indication
Transfer/store reference points
TX RF Signal (PLL)
RX RF Signal (S11) (REFLECTION)
Transmit Phase and Magnitude change to
μController
Filter RF Signal
Scale Signal
Filter RX Signal
Store
Transmit to computer
Labview GUI
USB
Calculate resonant over time
Plot Data
Archive Data
Calibration command
RX RF Signal (S21) (TRANSMISSION)
Filter RX Signal
Determine Change in Phase and Magnitude
Determine Change in Phase and Magnitude
ADC
Andrew Rosen-Project Manager
Power Unit
Computer
Measurement System
Microcontroller
John Louma (backup)
Jared Bold
Dan sinkiewicz
Yongjie Cao
PCB Layout
RF Transmission S21
Andrew Rosen (backup)
Jared Bold (backup)
RF Path
Antenna
Vector Measurement
Andrew Rosen
John Louma
John Louma(backup)
Dan Sinkiewicz
Andrew Rosen(backup)
John Louma(backup)
RF Transmission S11
John Louma
Yongjie Cao
John Louma(backup)
Yongjie Cao(backup)
Revision #:
Customer
Need #Importance Description Comments/Status
Measurement System
CN1 5 More accurate resonant frequency measurement Explore other options
CN2 5 Perform frequency sweep Other synthesizers?
CN3 5 Calibration System Short, open, load
CN4 3 Incorporate dual antenna system Measure transmitted and reflected signals
CN5 5 Verify antenna performance in real time Labview and network analyzer
CN6 4 Resolution (time) Sample rate
Microprocessor Data Handling
CN7 5 Communication with PC Wired VS. wireless
CN8 5 Controlling frequency sweep
CN9 3 Resolution (bit size)
PC Data Analysis
CN10 4 Parse data Labview
CN11 5 Real time update Labview
CN12 5 Display data Labview
CN13 2 Archive data
Physical Attributes
CN14 5 Compact PCB
CN15 5 Non-invasive
Functionality
CN16 4 Battery power Portable
CN17 2 Low power notification
CN18 3 Ergonomic fit on a limb Arm or leg?
CN19 4 Power conservation Low power when not used
Cust. Need #: enables cross-referencing (traceability) with specifications
Importance: Sample scale (1=must have, 2=nice to have, 3=preference only), or see Ulrich exhibit 4-8.
Description: organize as primary and secondary needs (hierarchy) -- Ulrich exhibit 4.8
Comment/Status: allows tracking of questions, proposed changes, etc; indicate if you are meeting the need ("met") or not ("not met")
Spec. # Importance Source Function Specification (metric)Unit of
Measure
Marginal
ValueIdeal Value
Comment
s/Status
Measurement System
S1 5 CN1 Antenna Performance Accuracy of measurement system compared to Network Analyzer KHz 15 10
S2 5 CN4 Power absorbtion Power of transmitted signal mW 200 150
S3 4 CN2 Frequency Sweep Check for resonant frequency between 200 MHz and 4 GHz MHz 200-4000 200-2000
S4 5 CN3 Calibration Time required for calibration Seconds 60 1
S5 4 CN4 Resonant Frequency Determination Bandwidth of Narrowband Antenna MHz 15 10
S6 5 CN4 Resonant Frequency Determination Bandwidth of Wideband Antenna MHz 200-2000 100-3000
S7 4 CN4 Power absorbtion return loss of Narrowband Antenna dB -20 -15
S8 5 CN4 Power absorbtion return loss of Wideband Antenna dB -20 -15
S9 4 CN4 Phase change Max phase change degrees 180 180
S10 3 CN6 Sampling Intervals between measurements Seconds 60 15
Microprocessor Data Handling
S11 3 CN7 Data Transmission Time required for packet transmission Seconds 2ms 1us
S12 3 CN9 Analog to Digital Converter Bit resolution Bits 10 16
S13 5 CN11 Update Time Time between microprocessor transmission and data display Seconds 10ms 8ms
PC Data Analysis
S14 2 CN12 Data Display Amount of data points Data Points 20 240
S15 2 CN13 Data Storage Amount of data stored Bytes 1MB 1KB
Physical Attributes
S16 3 CN14 Printed Circuit Board Layout Size Inches 3x3 Less than 3x3
Functionality
S17 3 CN16 Power Source Battery Voltage Volts 5 3.3
S18 3 CN17 Low Power Notification Time before battery death Minutes 30 5
Spec. #: enables cross-referencing (traceability) and allows mapping to lower level specs within separate documents
Source: Customer need #, regulatory standard (eg. EN 60601), and/or "implied" (must exist but doesn't have an associated customer need), constraint
Description: quantitative, measureable, testable details
*This table can be expanded to document test results
Non-Invasive Blood Glucose Model
Incident Signal
Reset
Transmitted Signal
Resolution
P13071 Non-Invasive Blood Glucose MonitorLevel 0 Functional Block Diagram
Device Under Test (DUT)
Computer ( LabVIEW)Data to computer
Reflected Signal
Open, Short, Load, Through
USB
Battery PackPower Regulator
Hard Drive(Archive)
Plot Data
Math Manipulation
Port Choice
USB
Physical Button
P13071 Non-Invasive Blood Glucose Monitor
Level 1 Functional Block Diagram
Resolution
Arm/Sample
RF Synthesizer
Antenna (S11)
Vector Measurement
(Scattering Matrix)
Antenna (S21)
Vector Measurement
(Scattering Matrix)
Directional Coupler
Directional Coupler
RF Splitter
Inci
den
t
Inci
den
t
Tran
smit
ted
Reference
Reference
Reference
Reflected
Ref
lect
ed
MIcrocontroller
POR (Power-on Reset)
Calibrate RF PathTransmit Calibration
Data to HostInitialize Synthesizer
Begin sweep command from PC?
Enter Low Power Mode
Transmit Synthesizer Signal
Measure S11Measure S21Send S Matrix to
PC / Store S Matrix in Flash
Sweep Complete?
No
No
Sweep Interval Timer Expired
Enter Low Power Mode
No
Yes
Reset
Brown Out
Stop
Microcontroller Level 0 Functional Block Diagram
P13071 Non-Invasive Blood Glucose Monitor
Screening - GUI
Concepts
A B C D E
Matlab C++ Mobile Platform (Reference)
On board
Screen/processing
Selection Criteria LabVIEW
USB interface 0 0 - D -
Ease of programming + - - A -
Ease/ability to Interface with peripherals 0 - - T -
Familiarity + + - U -
M
Sum + 's 2 1 0 0 0
Sum 0's 2 1 0 0 0
Sum -'s 0 2 4 0 4
Net Score 2 -1 -4 0 -4
Rank 1 3 4 2 4
Continue? Yes
P13071 Non-Invasive Blood Glucose Monitor
Screening - RF Synthesizer
Concepts
A B C D E G
VCO Crystal Reference DDS RF Mixer
Selection Criteria PLL
Frequency range - - d - -
Deviation from set frequency - + a 0 0
small footprint 0 - t 0 0
Input signal 0 + u 0 -
m
d
a
t
u
m
Sum + 's 0 2 0 0
Sum 0's 2 0 3 2
Sum -'s 2 2 1 2
Net Score -2 0 0 0 -1 -2
Rank 5 2 1 3 4
Continue? No No No Yes No No
P13071 Non-Invasive Blood Glucose Monitor
Screening - Calibration
Concepts
A B C D
Mux with on chip loads External loads (Reference)
Selection Criteria None
Automated + 0 D
Ease of implementation - + A
Programming complexity - 0 T
Calibration - + U
Usability + - M
Space required + -
Sum + 's 3 2 0
Sum 0's 0 2 0
Sum -'s 3 2 0
Net Score 0 0 0
Rank
Continue? Yes No No
P13071 Non-Invasive Blood Glucose Monitor
Screening - Communication Interface
Concepts
A B C D E G
USB Bluetooth Zigbee RS-232
Selection Criteria (Ben's Thesis)
Power Usage during transmission + + + d
Power Usage during idle 0 + + a
Transfer speed + + + t
Power Supply + 0 0 u
Error rate 0 - - m
Distance of communication 0 0 0
Ease of implementation 0 0 0 d
Cost to implement - - - a
Space to implement - - - t
Ease of computer interface - - - u
m
Sum + 's 3 3 3 0
Sum 0's 4 3 3 10
Sum -'s 3 4 4 0
Net Score 0 -1 -1 0
Rank 1 4 3 2
Continue? Yes No No No
P13071 Non-Invasive Blood Glucose Monitor
Screening - Processor
Concepts
A B C D E G
MSP430 Stellaris ARM C2000 Reference Hercules ARM Arduino
Selection Criteria PIC
I2C 0 0 0 D 0 0
UART 0 0 0 A 0 0
SPI 0 0 0 T 0 0
USB 0 - 0 U - -
ADC + - - M - -
Flash Memory 0 - 0 + -
Clock Speed 0 + + + -
GPIO - - - - -
Sum + 's 1 1 1 0 2 0
Sum 0's 6 3 5 0 3 3
Sum -'s 1 4 2 0 3 5
Net Score 0 -3 -1 0 -1 -5
Rank 1 3 2 1 2 4
Continue? Yes No No No No No
P13071 Non-Invasive Blood Glucose Monitor
Screening - On Device Storage
Concepts
A B C D E
Removable SD card uC Flash Flash mem chip (Reference) Flash drive
Selection Criteria None
Storage Size + + + D +
Storage Speed + + + A +
Ease of Access 0 0 0 T 0
Ease of Implementation - 0 - U -
Cost - 0 - M -
Sum + 's 2 2 2 0 2
Sum 0's 1 3 1 0 1
Sum -'s 2 0 2 0 2
Net Score 0 2 0 0 0
Rank 2 1 2 2 2
Continue? No Yes No No No
P13071 Non-Invasive Blood Glucose MonitorScreening - Microstrip Antenna Designs
Concepts
A B C D E
Patch Dipole vivaldi Reference Bowtie
Selection Criteria planar
inverted f
size 0 0 0 D 0
groundplane + + - A +
bandwidth + - 0 T +
ease of design + + - U -
M
D
A
T
U
M
Sum + 's 3 2 0 0 2
Sum 0's 1 1 2 0 1
Sum -'s 0 1 2 0 1
Net Score 3 1 -2 0 1
Rank 1 2 5 3 4
Continue? Yes No No No No
MSD Project Risk Assessment
ID Risk Item Effect Cause Like
liho
od
Seve
rity
Imp
ort
ance
Action to Minimize Risk Owner
1
Compensating for Electrical Delay
Calibration won’t be accurate
Calibration 2 2 10 Develop a check in code to adjust for electrical delay
Andrew
2 Transmission through arm won’t work Inability to measure S21
Arm causes too much attenuation 3 3 5
Use appropriate power in design for microwave sensors Andrew
3 Coupling between antennas
Accuracy of measurement S11, S21
Antennas are place too close together 2 3 10
Place device in area this is unlikely to occur Andrew
4 Noise being introduced in Channel
Bad measurement of S11, and S21
Movement, body type, environment 3 2 9 Designed to reduce transience Jared
5 Synthesizer lock time Lengthen/restrict resolution
Switch time on synthesizer 1 1 6 Synthesizer choice Dan
6 Peripheral configuration issues
Peripheral devices won’t perform as intended
Inappropriately set registers 2 3 10 Dev boards for all peripherals Dan
7 Dielectric gel could affect measurement Could shift resonance
Dielectric not modeled properly 1 2 8 Research dielectric gel Andrew
8 Patient electrocution Could harm patient Short power to device 1 2 10 Power isolation Yongjie
9 Program hangup Loss of functionality Poor programming 1 3 10 Spend time reviewing code John
10 Not accurate reference signal
S11 and S21 will not be correct
RF splitter doesn’t work 1 3 10 Validate power splitter Andrew
12 PCB layout Not knowing how to do an RF pcb
13 Directional coupler Not working at right freq
Likelihood scale Severity scale
1 - This cause is unlikely to happen 1 - The impact on the project is very minor. We will still meet deliverables on time and within budget, but it will cause extra work
2 - This cause could conceivably happen 2 - The impact on the project is noticeable. We will deliver reduced functionality, go over budget, or fail to meet some of our Engineering Specifications.
3 - This cause is very likely to happen 3 - The impact on the project is severe. We will not be able to deliver, or what we deliver will not meet the customer's needs.
“Importance Score” (Likelihood x Severity) – use this to guide your preference for a risk management strategy
Prevent Action will be taken to prevent the cause(s) from occurring in the first place.
Reduce Action will be taken to reduce the likelihood of the cause and/or the severity of the effect on the project, should the cause occur
Transfer Action will be taken to transfer the risk to something else. Insurance is an example of this. You purchase an insurance policy that contractually binds an insurance company to pay for your loss in the event of accident. This transfers the financial consequences of the accident to someone else. Your car is still a wreck, of course.
Accept Low importance risks may not justify any action at all. If they happen, you simply accept the consequences.