The Pennsylvania State University
The Graduate School
Department of Bioengineering
ULTRASONIC TRANSDERMAL GLUCOSE MONITORING AND INSULIN
DELIVERY USING CYMBAL TRANSDUCER ARRAYS
A Dissertation in
Bioengineering
by
Eun-Joo Park
2010 Eun-Joo Park
Submitted in Partial Fulfillment of the Requirements
for the Degree of
Doctor of Philosophy
May 2010
ii
The dissertation of Eun-Joo Park was reviewed and approved* by the following:
Nadine Barrie Smith Associate Professor of Bioengineering Dissertation Advisor Chair of Committee
Andrew G. Webb Professor of Bioengineering
Ryan S. Clement Assistant Professor of Bioengineering
Richard J. Meyer Jr. Associate Professor of Materials Science and Engineering
Herbert H. Lipowsky Professor of Bioengineering Head of the Department of Bioengineering
*Signatures are on file in the Graduate School
iii
ABSTRACT
As a practically portable ultrasound system for transdermal drug delivery, cymbal
transducer arrays have shown promise in several in vivo experiments using small animals.
The goal of this study was to evaluate the potential of the cymbal transducer array as a
pre-clinical application for transdermal insulin delivery and glucose sensing. In order to
achieve this goal, several aspects have been investigated, such as the feasibility of
ultrasonic transdermal insulin delivery and glucose sensing in large animals, and a
closed-loop feedback controller for regulating glucose levels. To evaluate further the
practicability of the cymbal transducer array as a pre-clinical application, in vivo
experiments with transdermal insulin delivery were performed in large pigs (~ 200 lbs).
The results of the ultrasound exposure group indicate that the glucose level decreased to
-74 ± 5 mg/dl at 60 minutes and continued to decrease to -91 ± 9 mg/dl at 90 minutes. In
addition to the insulin delivery, the cymbal transducer array has been used in determining
the glucose concentration of interstitial fluid. From the perspective of application in
humans, the in vivo studies using large pigs were designed to determine the feasibility of
the cymbal transducer array in ultrasonic transdermal glucose sensing. After 5 minutes of
ultrasound exposure at Isptp = 100 mW/cm2, the glucose levels determined by the
ultrasound system and glucose meter were 127 ± 16 mg/dl and 131 ± 5 mg/dl,
respectively. Based on the results of transdermal insulin delivery and glucose sensing in
large pigs, the ultrasound systems for both applications were combined using a closed-
loop feedback controller. Four in vivo experiments were designed with large pigs to
demonstrate the feasibility of the combined ultrasound system in noninvasive glucose
iv
control. Based on the glucose level measured by ultrasound and biosensors at 20-minute
intervals, the ultrasound system of insulin delivery was shown to regulate the glucose
level. During the 120-minute experiments, glucose levels were controlled in the range of
104-137 mg/dl as determined by the ultrasound system. Through the in vivo experiments
of transdermal insulin delivery and glucose, the ultrasound system using the cymbal
transducer array has shown the positive potential of a practically portable device for a
clinical application of diabetes care.
v
TABLE OF CONTENTS
LIST OF FIGURES .....................................................................................................vii
LIST OF TABLES....................................................................................................... ix
Chapter 1 INTRODUCTION......................................................................................1
1.1 Motivations .....................................................................................................1 1.1.1 Diabetes ................................................................................................1 1.1.2 Glucose monitoring and insulin delivery .............................................4
1.2 Specific Aims and Scope ................................................................................5 1.3 Dissertation Outlines ......................................................................................7
Chapter 2 BACKGROUND - LITERATURE REVIEW ...........................................8
2.1 Transdermal Drug Delivery............................................................................8 2.2 Glucose Monitoring ........................................................................................13 2.3 Closed-loop Control for Diabetes Care ..........................................................14
Chapter 3 BACKGROUND - ACOUSTIC THEORY ...............................................17
3.1 Design of cymbal transducer ..........................................................................17 3.2 Acoustic Characteristics of Cymbal Transducer ............................................18 3.3 Acoustic Cavitation ........................................................................................20
3.3.1 Definition..............................................................................................21 3.3.2 Cavitation Threshold ............................................................................22
Chapter 4 MATERIALS AND METHODS - APPRATUS .......................................25
4.1 Construction of Cymbal Transducer Arrays...................................................25 4.1.1 Fabrication of cymbal arrays ................................................................25 4.1.2 Construction of matching circuits ........................................................29
4.2 Exposimetry: Acoustic Characteristics of Cymbal Arrays.............................30 4.3 Fabrication and Calibration of Electrochemical Glucose Sensors .................32
4.3.1 Fabrication of glucose sensors..............................................................33 4.3.2 Calibration of glucose sensors..............................................................34
4.4 Closed-Loop Feedback Controlled Ultrasound System for Insulin Delivery and Glucose Sensing.......................................................................35
Chapter 5 MATERIALS AND METODS – IN VIVO EXPERIMENTS ...................38
5.1 Transdermal Insulin Delivery .........................................................................39 5.2 Ultrasonic Transdermal Glucose Sensing.......................................................43 5.3 Closed-Loop Feedback Controlled Ultrasound System for Glucose
Control ...........................................................................................................46
vi
Chapter 6 RESULTS...................................................................................................50
6.1 Acoustic Pressure Distribution of Cymbal Transducer Arrays ......................50 6.2 Transdermal Insulin Delivery ........................................................................51
6.2.1 In Vivo Experiments in Large Pigs .......................................................51 6.2.2 Analysis of Insulin Concentration in Transdermal Insulin Delivery....53 6.2.3 Ultrasonic Glucose Sensing..................................................................56 6.2.4 Glucose control in large pigs using a closed-loop feedback
controlled ultrasound system..................................................................58
Chapter 7 CONCLUSION ..........................................................................................62
7.1 Transdermal Insulin Delivery .........................................................................62 7.2 Ultrasonic Glucose Sensing............................................................................64 7.3 Closed-Loop Feedback Controlled Ultrasound System for Glucose
Regulation......................................................................................................66 7.4 Future Work....................................................................................................68
BIBLIOGRAPHY........................................................................................................71
Appendix A FUNDAMENTAL RESONANCE FREQUENCY OF CYMBAL TRANSDUCER – FEA ........................................................................................80
A.1 In Air..............................................................................................................80 A.2 In Water .........................................................................................................81
Appendix B DETAIL INSTRUCTION OF CYMBAL ARRAY CONSTRUCTION ...............................................................................................83
B.1 Metal Endcaps................................................................................................83 B.2 Cymbals .........................................................................................................85 B.3 Polyurethane Base..........................................................................................85 B.4 Connections....................................................................................................86 B.5 Casting ...........................................................................................................87
Appendix C A COMPARISON BETWEEN INTRAPERITONEAL INJECTION AND TRANSDERMAL DELIVERY..................................................................91
Appendix D MATLAB CODES.................................................................................98
D.1 Low pass π-circuit for impedance matching..................................................98 D.2 Feedback controller........................................................................................100
vii
LIST OF FIGURES
Figure 3.1: Diagram of cymbal transducer in cross sectional view............................18
Figure 4.1: PZT-4 disk and a brass endcap.................................................................26
Figure 4.2: Dimensions of endcaps and PZT-4 disk...................................................27
Figure 4.3: Lightweight cymbal arrays were constructed using four transducers in a two-by-two pattern, (left) or nine cymbal transducers in a three-by-three pattern (right). The two-by-two array was used for ultrasonic glucose sensing, while the three-by-three array was used in the transdermal insulin delivery. ......29
Figure 4.4: Electric impedance matching network using a π-type circuit. After constructing the circuit based on the theoretical values, actual value of each component was adjusted to have maximum acoustic output................................30
Figure 4.5: Equipment setup of the exposimetry to characterize the cymbal arrays. Acoustic pressure field of each cymbal array was measured in the water tank filled with degassed water. ...................................................................................31
Figure 4.6: Photograph of an enzyme based electrochemical biosensor which consists of three electrodes: working/sensing, counter, and reference electrodes ..............................................................................................................34
Figure 5.1: Photograph of a transdermal insulin delivery experiment with pig placed in a lateral recumbent position with the array attached.............................40
Figure 5.2: (a) Layout of the glucose sensor incorporated with the cymbal transducer array placed on the skin surface. (b) Photograph of the ultrasonic glucose sensing on a pig with the array attached to the skin and the reservoir filled with PBS solution........................................................................................44
Figure 5.3: (a) The layout of the glucose sensing system incorporated with the insulin delivery system as placed on the skin surface. After five minutes of ultrasound exposure, the biosensor determined the glucose concentration of the interstitial fluid. Based on the glucose level, the insulin delivery system was automatically operated. (b) Photograph of the ultrasound systems placed on the pig for insulin delivery and glucose sensing. The ultrasound system was controlled by a closed-loop controller. ..........................................................47
Figure 5.4: Schematic diagram shows the experimental setup with the closed-loop feedback controller for noninvasive glucose regulation. The ultrasound operation for both the glucose sensing and insulin delivery systems was automatically operated..........................................................................................49
viii
Figure 6.1: Acoustic intensity profiles of the two-by-two (left) and three-by-three (right) arrays were determined by measuring acoustic pressure and plotted in dB. The spatial-peak-temporal-peak intensities were 100.8 ± 0.9 mW/cm2 for the two-by-two array and 100.0 ± 0.5 mW/cm2 for the three-by-three array. ............................................................................................................51
Figure 6-2: Change of blood glucose level in the ultrasound mediated transdermal insulin delivery on pigs.........................................................................................52
Figure 6.2: (a) The concentrations of human (, µU/l) and porcine (◊, ng/l) insulin were determined for 90-minutes periods. While porcine insulin concentrations stayed in the rage of 12.23 - 12.38 ng/l, human insulin concentration increased by 0.5 and 23.7 µU/l after 15 30 minutes ultrasound exposure, respectively. (b) Human insulin concentrations were determined in ng/l by unit conversion and compensated for the cross-reaction between porcine and human insulin for ELISA tests. The cross-reaction ratio was obtained at 13.3%. ................................................................................................55
Figure 6.3: The total insulin concentrations in ng/L were presented with the change of blood glucose level in mg/dl over 90-minute experiments. As insulin concentrations increased faster, the glucose level decreased at higher rate. .......................................................................................................................56
Figure 6.4: The glucose concentrations of the interstitial fluid determined using the ultrasound system and a conventional glucose meter. Asterisks above the data bar were used if there was no statistical difference between the ultrasonic glucose sensing method and the conventional method based on the ANOVA with the 0.05 or 0.01 level of significance. ...........................................57
Figure 6.5: (a) – (d) Glucose concentrations of four pigs, which were regulated by the ultrasonic transdermal glucose sensing and insulin delivery systems combined by a feedback controller. ......................................................................60
Figure C.1: Photograph of a transdermal insulin delivery experiment on a rat place in a dorsal decubitus position with the array attached. ...............................92
Figure C.2: Changes of blood glucose by the ultrasonic transdermal insulin delivery and subcutaneous injections of insulin. ..................................................94
ix
LIST OF TABLES
Table 1.1: Comparison of Type 1 and Type 2 diabetes. .............................................3
Table 2.1: Enhanced skin permeability by ultrasound.................................................10
Table 4.1: To control the ultrasound system using cymbal arrays, a simple proportional controller was used. .........................................................................37
Table 5.1: The detail information of ultrasound exposure conditions. .......................45
Table A.1: Comparison of the in-air resonance frequency of standard size brass-capped cymbals for various PZT types.................................................................80
Table A.2: Comparison of the in-air resonance frequency of standard size cymbals for various endcap materials...................................................................80
Table A.3: Comparison of the in-air resonance frequency of standard size brass-capped cymbals for different endcap thickness. ...................................................81
Table A.4: Comparison of the in-air resonance frequency of standard size brass-capped cymbals for different cavity depths. .........................................................81
Table A.5: Comparison of the in-water resonance frequency of standard size brass-capped cymbals for various PZT types. ......................................................81
Table A.6: Comparison of the in-water resonance frequency of standard size cymbals for various endcap materials...................................................................82
Table A.7: Comparison of the in-water resonance frequency of standard size brass-capped cymbals for different endcap thickness. .........................................82
Table A.8: Comparison of the in-water resonance frequency of standard size brass-capped cymbals for different cavity depths. ...............................................82
x
ACKNOWLEDGEMENTS
I would like to thank my family for their countless love and support. They have
been with me all the time and made me get through this long journey. Without them, I
wouldn’t be able to overcome all the obstacles. Also, I would like to thank Dr. Nadine
Barrie Smith. She was my role model, mentor, big sister, and a good friend although I
did not have a chance to tell her how much I was happy to have her. She was, is, and will
be my advisor.
Chapter 1
INTRODUCTION
1.1 Motivations
More than 23.6 million Americans suffer from diabetes – the seventh leading
cause of death in the United States (American Diabetes Association 2007). Diabetes is a
disease in which the body does not properly produce or utilize insulin, a hormone that
converts food (such as sugar and starches) into energy needed for daily life. Diabetes
costs the nation in excess of $170 billion per year and accounts for approximately 25% of
all Medicare expenditures. In addition, it is a risk factor for heart disease, stroke, and
birth defects and shortens average life expectancy by up to 15 years. Although much
research has been dedicated to finding the cause and a cure, the cause of diabetes remains
unknown.
1.1.1 Diabetes
Diabetes is a metabolic disorder characterized by chronic hyperglycemia with
disturbances of carbohydrate, fat, and protein metabolism resulting from defects in
insulin secretion, insulin action or both. Although the symptoms are often not severe or
may even be absent, diabetes may present characteristic symptoms such as thirst, polyuria,
weight loss, and blurred vision. Complications stemming from diabetes include long-
term damage, dysfunction and failure of several organs, especially the eyes, kidneys,
2
heart, blood vessels and teeth (Beaser & Staff of Joslin Diabetes Center 2007). In the
absence of effective treatment, the complications can ultimately lead to death.
When an individual has symptoms such as thirst, polyuria, unexpected weight loss,
drowsiness or coma, and marked glucosuria, diagnostic screening using a fasting plasma
glucose test or an oral glucose tolerance test (OGTT) is required. Because of its
simplicity, the fasting plasma glucose test is recommended by the American Diabetes
Association (ADA). Under normal conditions, the plasma glucose concentration after an
eight-hour fast is less than 100 mg/dl while the 2-hour postload plasma glucose level is
lower than 140 mg/dl. If the fasting plasma glucose level is higher than 126 mg/dl or the
2-hour postload plasma glucose level is higher than 200 mg/dl, the patient is diagnosed as
having diabetes.
In 1997, the ADA proposed the current clinical classification, which was adopted
by the World Health Organization (WHO) in 1999, based on the fact that the majority of
diabetes cases falls into two broad etiopathologic categories: Type 1 and Type 2 diabetes.
Given the increasing number of forms of diabetes recognized by a specific etiology,
diabetes is currently classified using both etiologic types and clinical stages etiologic
types. The terms Type 1 and Type 2 of the current classification system refer to the most
common forms of diabetes.
Type 1 diabetes is caused by the destruction of the insulin producing β-cell of the
islets of Langerhans in the pancreas and usually leads to the clinical stage that requires
insulin treatment for survival. Type 1 diabetes is characterized by the presence of
antibodies that reflect the autoimmune process leading the β-cell destruction. In
individuals who have Type 1 diabetes, their immune system recognizes the β-cell as an
3
antigen or pathogen and attacks the cells with antibodies (Beaser & Staff of Joslin
Diabetes Center 2007). The destruction of the β-cell results in the loss of insulin
secretion and absolute insulin deficiency (i.e., insulinopenia).
Type 2 diabetes is the most common form of diabetes, comprising approximately
90 percent of all diabetes cases. It is characterized by disorders of insulin action and
insulin secretion due to insulin resistance and reduced insulin sensitivity. Compared to
Type 1 diabetes, symptoms are not as severe in individuals having Type 2 diabetes since
the body produces enough or even too much insulin. Because Type 2 diabetes is
asymptomatic, it is often unrecognized and untreated during the early stages of
development. The differences between Type 1 and Type 2 diabetes are summarized in
Table 1.1.
Table 1.1: Comparison of Type 1 and Type 2 diabetes.
Type 1 diabetes Type 2 diabetes
Percentage of people with diabetes
10% 90%
Onset Usually below 30 years Usually over 30 years
Condition when discovered
Usually moderately to severely ill
Often not ill at all, or having mild symptoms
Cause of diabetes Reduced or absent insulin production
Insulin resistance and relative or absolute insulin secretory deficiency
Insulin level None to small amounts Markedly elevated early
Acute complications Ketoacidosis
Non-ketotic hyperosmolar hyperglycemic coma. Usually not prone to ketoacidosis
Usual treatment Insulin, meal plan, exercise Diet, exercise, antidiabetes medications and insulin if needed
4
1.1.2 Glucose monitoring and insulin delivery
In the management of diabetes, a strict control of blood glucose concentration can
reduce the risk of short- and long-term complications. Multiple measurements of glucose
concentration are necessary to maintain the blood glucose level within the normal range.
In particular, self-blood glucose monitoring (SBGM) is very important in diabetes
treatment as it provides information related to glucose patterns to help the healthcare
team make appropriate treatment decisions, provides data with which patients can
independently make daily treatment-related decisions, and provides feedback on the
efficacy of the treatment, especially in regard to physical activities and nutritional
therapy. Conventional SBGM techniques use a portable glucose monitor, which requires
a blood sample obtained by pricking fingers. Although this method is widely used, it
limits the number of measurements because it is painful and invasive.
Insulin treatment is essential for Type 1 diabetes; however, it is not required in the
early stages of other types of diabetes—although may be required, along with a
combination of other treatments (e.g., diet, oral agents, and exercise) when the glucose
level is not well controlled. The goals of insulin treatment are to eliminate symptoms of
hyperglycemia, prevent diabetic ketoacidosis, avoid developing severe catabolic state,
reduce frequent infections, decrease fetal and maternal morbidity in pregnancy, and
prevent and delay microvascular and macrovascular complications. The traditional
insulin delivery method involves using a syringe and needle for injection; this needle
injection system is flexible and allows dosages to be adjusted readily. However, the
number of injections and injection sites on a body are limited, and this method involves
5
pain that can affect patient compliance. To provide advantages over traditional injection
drug delivery methods, several noninvasive methods have been studied.
The challenge with the management of diabetes is to avoid the continuous process
of finger pricking for blood glucose samples and daily insulin injections, which can often
lead to noncompliance. Noncompliance typically occurs in the form of refusing
treatment, reducing or increasing insulin doses, and skipping medications. Such actions
are common problems that can prevent the efficacy of the treatment and negatively
impact the patient’s health. Ultimately, noncompliance can lead to complications such as
nerve damage, retinal damage, microvascular damage, chronic renal failure, kidney
failure, dental disease, and cardiovascular disease. In order to overcome noncompliance,
a great deal of effort has focused on finding alternative methods that can deliver insulin
noninvasively.
1.2 Specific Aims and Scope
Previous studies using ultrasound to facilitate noninvasive insulin delivery or
extraction of interstitial skin fluid for glucose monitoring have shown positive results.
However, these studies used a bulky commercial sonicator to generate the ultrasound
energy. Interest in a compact and lightweight ultrasound device indicates the need to
develop a portable ultrasound device for transdermal drug delivery.
Several possible low frequency transducer designs can be used in a drug delivery
application, such as the low frequency flextensional resonators, tonpilz transducers, or
“thickness”-type resonators. Another inexpensive candidate for a portable device is the
6
low-profile, lightweight cymbal transducer. The cymbal transducer is a flextensional
transducer that is less than 2 mm thick, weighs less than 3 grams, and resonates between
1 and 100 kHz depending on the geometry. Multiple cymbals can be patterned for
increased spatial intensity fields for drug transport. Indeed, the use of the low-profile,
lightweight cymbal ultrasound array has previously demonstrated transport enhancement
of insulin across in vitro human skin as well as in vivo rats and rabbits. The cymbal
design has also been used for noninvasive glucose sensing in rat models.
The goal of the current research is to develop a practical ultrasound system for
noninvasively monitoring blood glucose and transporting insulin across the skin, which is
operated by a closed-loop feedback controller. To overcome the drawbacks of ultrasound
system that used in previous studies (El-Kamel, Al-Fagih, & Alsarra 2008; Kost 2002;
Mitragotri, Blankschtein, & Langer 1995c; Tachibana & Tachibana 1991; Zhang, Shung,
& Edwards 1996) and to improve the mobility of ultrasound devices, a small-sized low-
frequency (1 – 100 kHz) cymbal transducer has been developed. Using the cymbal
transducer, the efficacy of transdermal insulin delivery has been evaluated within in vivo
experiments on rats, rabbits, and large pigs (Lee et al. 2004; Lee, Newnham, & Smith
2004; Park, Werner, & Smith 2007). This transducer has a compact design with a
thickness of less than 2 mm and a weight of less than 3 grams. Multiple cymbal
transducers can be patterned side by side into 2 × 2 and 3 × 3 arrays. Arranging the
transducer patterns in such a way increases the spatial acoustic intensity fields for drug
transport. As the specific aims of this study, in vivo experiments were designed to
demonstrate the feasibility of noninvasive transdermal insulin delivery and glucose
sensing in large animals as a milestone to a future clinical application, and feasibility of a
7
closed-loop feedback-controlled ultrasound system using cymbal transducers. In addition,
the study to answer the essential questions of transdermal insulin delivery was included.
1.3 Dissertation Outlines
In chapter 2, background information about alternative methods of glucose
monitoring, insulin delivery, and closed-loop between the monitoring and delivery are
presented based on literature reviews. Acoustic theories of the cymbal transducer and
potential mechanism of ultrasonic transdermal method are described in chapter 3.
Chapter 4 presents detailed descriptions of the construction procedures of cymbal
transducers, glucose sensors, and feedback controller for the ultrasound system.
Details of in vivo experiments on the transdermal insulin delivery, ultrasonic glucose
monitoring, and the regulation of glucose level by the closed-loop feedback
controlled ultrasound system are explained in chapter 5. In chapter 6, the results of
each in vivo experiment as well as the measured acoustic properties of the cymbal
array are presented. Chapter 7 provides the conclusions drawn from each task with in
vivo experiments. In addition, fundamental resonance frequencies of the cymbal
transducer, the step-by-step construction procedures of the cymbal arrays, comparison
of the efficacy of transdermal insulin delivery to intraperitonial injection, and a
program for the feedback controller used in this study are provided in the appendix.
Chapter 2
BACKGROUND - LITERATURE REVIEW
As an alternative to overcoming the compliance negatively impacted by the
current diabetes treatment (i.e., finger pricking and multiple injections of insulin), several
studies have presented positive results of ultrasonic transdermal insulin delivery and
extraction of glucose in the interstitial fluid (ISF). In this section, researches on
transdermal drug delivery and glucose sensing were revised. In addition, several studies
of a feedback controlled system for diabetes care were revised.
2.1 Transdermal Drug Delivery
One alternative to noninvasive drug administration is transdermal drug delivery,
which provides advantages over traditional injection drug delivery methods. Several
noninvasive methods exist for transdermal drug delivery, including chemical mediation
using liposomes and chemical enhancers or physical methods such as microneedles,
iontophoresis, electroporation, and ultrasound (Montorsi et al. 2000; Prausnitz 1997;
Sershen & West 2002; Tezel, Sens, & Mitragotri 2002; Whelan 2002). However,
currently, few drugs, proteins, or peptides have been successfully administered
transdermally for clinical applications due to the low skin permeability of these relatively
9
large molecules, a problem attributed to the stratum corneum (e.g., the outermost skin
layer). One hypothesis related to transdermal drug delivery is that, once the drug has
traversed the stratum corneum, the next layer is easier to cross meaning the drug can thus
reach the capillary vessels to be absorbed(Mitragotri et al. 1995; Tezel, Sens, &
Mitragotri 2002). However, at this time, even with these enhancing mechanisms,
transdermal administration of drugs for clinical applications remains limited.
Recent reviews have demonstrated that ultrasound-mediated transdermal drug
delivery offers promising potential for noninvasive drug administration(Mitragotri 2005;
Pitt, Husseini, & Staples 2004; Smith 2007; Tachibana & Tachibana 2001). Researchers
have suggested that the working principle of ultrasonic transdermal drug delivery stems
from cavitation (Guzman et al. 2003b; Mitragotri, Edwards, Blankschtein, & Langer
1995; Mitragotri, Blankschtein, & Langer 1997b; Schlicher et al. 2006). Low frequency
ultrasound is capable of generating microbubbles in water and tissue, which allow water
channels to be produced within lipid bilayers. The resulting disorder created in the
stratum corneum facilitates the crossing of a hydrophilic drug or molecule. To this end,
the number of drugs and compounds shown to transdermally cross skin via ultrasound
continues to increase (Smith 2007). In vivo studies using ultrasound in transdermal drug
delivery showed that the skin permeability was enhanced for up to 6 hours. The
enhanced skin permeability is listed in Table 2.1 for each ultrasound operating condition.
10
Table 2.1: Enhanced skin permeability by ultrasound.
Ultrasound Drug /
Animals fre-quency
mode intensity exposure
time Permeability
(a) Insulin / rabbit
105 kHz
50% / 10 sec
repetition time
5 kPa 90 min
Glucose level was recovered after 6 hours from the beginning of experiments.
(b) D-[3H]Mannitol / pigs (17-20 kg)
20 kHz
50% / 10 sec
repetition time
6.5 W/cm2 5 min Mannitol was delivered up to 4.5 hours from the beginning of experiments.
(c) Insulin / hairless
rats 20 kHz
20% / 8 sec
reputation time
5 W/cm2 15 min
Glucose level decreased and kept in hypoglycemic condition up to 3 hours of experiments.
(d) Insulin / hairless
rats 20 kHz
20% / 8 sec
reputation time
5 W/cm2 15 min
Glucose level decreased and kept in hypoglycemic condition up to 3 hours of experiments.
(e) Insulin / hairless
rats 20 kHz
40% / 8 sec
repetition time
2.5 W/cm2 15 min
Glucose level decreased and kept in hypoglycemic condition up to 2.5 hours of experiments.
(f) Insulin / hairless
rats 20 kHz
10% / 1 sec
repetition time
125 and 225
mW/cm2 60 min
Glucose level decreased and kept in hypoglycemic condition up to 4 hours from the beginning of ultrasound exposure.
(g) Salicylic
acid / hairless
rats
20 kHz
10% / 1 sec
repetition time
125 mW/cm2
60 min
The amount of salicylic acid recovered in urine was increased until the end of the 2.5 hours experiments.
(h) Insulin / hairless
rats 20 kHz 10%
225 mW/cm2
30 min
Glucose level decreased and kept in hypoglycemic condition up to 1.5 hours from the beginning of ultrasound exposure.
11
Noninvasive ultrasonic transdermal delivery of insulin has generated significant
public interest due to the increasing number of people affected by diabetes. In the 20 to
105 kHz frequency range, enhanced transport in the presence of ultrasound has been
shown in both in vitro and in vivo experiments (Boucaud et al. 2000; Boucaud et al. 2002;
Kost 2002; Mitragotri, Blankschtein, & Langer 1995c; Smith et al. 2003; Smith 2007;
Tachibana 1992b; Zhang, Shung, & Edwards 1996). Many such early experiments were
performed using an ultrasound sonicator, ultrasonic bath, or commercial transducer. For
example, investigators demonstrated effective in vivo transport of insulin at 48 kHz using
an ultrasonic bath and at 105 kHz using a commercially obtained transducer (Tachibana
1992a). However, the major drawback thus far in exploiting ultrasound for noninvasive
drug delivery is the large size and poor mobility of the ultrasound device. Commercial
sonicators are large, heavy, table-top devices specifically designed for lysis of cells,
catalyzation of reactions, creation of emulsions, or cleaning purposes.
In transdermal insulin delivery using mechanical (Mitragotri, Edwards,
Blankschtein, & Langer 1995; Nordquist et al. 2007), electrical (Murthy et al. 2006),
chemical enhancers (Sintov & Wormser 2007), or the combination of these enhancers
(Chen et al. 2009), researchers monitored insulin concentrations to determine how
effectively insulin was delivered through the skin by their methods. Researchers also
measured glucose level as an indicator of insulin delivery.
Using ultrasound at 20 kHz, in vitro experiments were performed to investigate
transdermal insulin delivery and insulin concentrations were measured either by
radioimmunoassay (RIA) or enzyme-linked immunosorbent assay (ELISA) (Mitragotri,
Blankschtein, & Langer 1995b). In the in vitro study of iontophoresis-driven insulin
12
transport through microneedle-induced skin microchannels, insulin concentrations were
measured by high performance liquid chromatography (HPLC) (Chen, Zhu, Zheng, Mou,
Wan, Zhang, Shi, Zhao, Xu, & Yang 2009). Insulin concentrations were also measured
by HPLC during the in vitro experiments of enhanced permeation flux of insulin using
trypsin as a biochemical enhancer (Li et al. 2008). In vitro experiments of
electroporation used fluorescein-isothiocyanate (FITC) insulin to investigate the
enhancement of transdermal transport by measuring fluorescence (Connolly, Cotton, &
Morin 2002). During the in vivo study of enhanced insulin delivery by co-administrating
short synthetic peptide, insulin concentrations were measured by immunoradioassay
(Chen et al. 2006). Using immunoradiometric assay (IRMA), insulin concentrations were
measured in transdermal insulin delivery using microneedles in in vivo condition
(Martanto et al. 2004).
While insulin concentrations were measured in both in vitro and in vivo studies,
glucose concentrations were widely used as the indicator of transdermal transport in in
vivo studies. For example, during the in vivo experiments using ultrasound, blood
glucose levels were monitored every 30 minutes (Boucaud, Garrigue, Machet, Vaillant, &
Patat 2002; Mitragotri, Blankschtein, & Langer 1995c). In transdermal insulin delivery
studies using needle-free liquid jet injectors (Arora et al. 2007), microneedles (Prausnitz
2004), and biochemical enhancers (Li, Quan, Zang, Jin, Kamiyama, Katsumi, Yamamoto,
& Tsutsumi 2008), blood glucose levels were measured periodically.
Monitoring insulin concentrations in serum directly answers the questions of
whether insulin is delivered through the skin and how much insulin is transported by
enhancers including ultrasound. It also provides insulin delivery rates at the given
13
conditions of enhancing methods. Ultimately, measuring insulin concentrations in serum
will supply the information for the insulin dose guideline of each method, and this
guideline will provide essential information for developing an insulin delivery system.
2.2 Glucose Monitoring
Measuring glucose in interstitial fluid has shown to be an alternative to blood
glucose measurements. One device used to measure the glucose concentration of
interstitial fluid is a subcutaneous sensor, such as the MiniMed Continuous Glucose
Monitoring System (Medtronic, Minneapolis, MN) (Mastrototaro 1999). Depending on
the model, the subcutaneous sensor can measure the glucose level every one to five
minutes. However, it has some drawbacks, such as the need to replace sensors every
three to seven days and the interaction of biological components, causing fibrosis (Garg
et al. 2006; Weinstein et al. 2007; Wong et al. 2006). Devices using microdialysis also
measure the glucose concentration of interstitial fluid (Heinemann 2003). Despite the
advantage of providing sufficiently precise measurements, such devices are either only
available to physicians or too bulky to wear during daily activities. Additional possible
methods of glucose sensing such as impedance spectroscopy (Caduff et al. 2003), reverse
iontophoresis (Sekkat et al. 2002), optical technique using near-infrared fluorescence
(Ballerstadt et al. 2004), and tear glucose monitoring have been underdeveloped.
14
2.3 Closed-loop Control for Diabetes Care
The importance of tightly controlled blood glucose concentrations has been well
recognized in the management of diabetes. The publication on the landmark Diabetes
Complications and Control Trial (DCCT) describes improved glycemic control as having
the potential to reduce the risk of long term complications such as chronic renal failure,
retinal damage, nerve damage, cardiovascular disease, and microvascular damage(The
Diabetes Control and Complications Trial Research Group (DCCT) 1993). As a
longstanding goal in diabetes care, a system that can automatically restore near-
physiologic glycemic control, a so-called ‘artificial pancreas’, has been studied for over
30 years(Bessman & Schultz 1972; Felts 1973; Kumareswaran, Evans, & Hovorka 2009).
Such a system requires three main elements: continuous glucose monitoring, insulin
delivery, and a controller that regulates the insulin delivery.
Continuous insulin delivery using an infusion device called an insulin pump was
introduced in the 1970s, and it is commercially available for Type 1 diabetes(Jeitler et al.
2008; Volpe et al. 2009; Weissberg-Benchell, ntisdel-Lomaglio, & Seshadri 2003). In a
recent comprehensive review involving a meta-analysis, it was concluded that severe
hypoglycemic rates were markedly lower with continuous subcutaneous insulin infusion
versus multiple daily injection therapy(Pickup & Renard 2008). Continuous
subcutaneous insulin infusion (CSII) is effective in managing brittle or unstable Type 1
diabetes. However, it has drawbacks such as infections that can occur at the site of
needle insertion. Also, after changing infusion sets every three days, flow problems tend
to occur(Cohen & Shaw 2007).
15
Research in glucose monitoring has resulted in the development of the
continuous glucose monitoring (CGM) system(Cox 2009; Hanaire et al. 2009; Ryan &
Germsheid 2009; Skyler 2009). The CGM system measures glucose concentrations in
the interstitial fluid every one to five minutes using sensors inserted subcutaneously.
Although the CGM system has shown a benefit in glucose control, it has some limitations
such as the need to replace sensors every three to seven days and the interaction of
biological components causing physiological clots(Garg, Zisser, Schwartz, Bailey,
Kaplan, Ellis, & Jovanovic 2006; Weinstein, Bugler, Schwartz, Peyser, Brazg, &
McGarraugh 2007; Wong, Buckingham, Kunselman, Istoc, Leach, & Purvis 2006). Also,
it is still an invasive technique, requiring the subcutaneous insertion of sensors.
Currently, CSII and CGM techniques are used as an open-loop system that
requires patients or caretakers to interpret the glucose information(Kowalski 2009). To
close the loop between the two systems so as to have the systems provide the
interpretation, several algorithms have been studied and evaluated either in in silico
conditions(Marchetti et al. 2008; Patek et al. 2007) or in hospital conditions(Renard
2008; Weinzimer et al. 2008). The evaluation results of clinical trials have demonstrated
the feasibility of glucose control in 24 – 48 hours. While these evaluations show positive
results in the development of a closed-loop system for diabetes care, the technique relies
on subcutaneous injection of needles and, though minimal, is still invasive. To address
the limitations of subcutaneous insertion of CSII and CGM systems, ultrasound has been
used to transdermally extract glucose in the interstitial fluid(Cantrell, McArthur, &
Pishko 2000; Lee et al. 2005; Mitragotri et al. 2000; Park et al. 2009) as well as to
16
transport insulin(Lee, Snyder, Newnham, & Smith 2004; Mitragotri 2005; Park, Werner,
& Smith 2007; Pitt, Husseini, & Staples 2004).
Chapter 3
BACKGROUND - ACOUSTIC THEORY
3.1 Design of cymbal transducer
A cymbal transducer, defined as a piezoelectric electroacoustic transducer, is
classified as a type V flextentional transducer. A cymbal transducer capable of producing
a very low frequency has a compact, lightweight structure. Based on the dimensions and
materials of the transducer, the cymbal transducer has an adjustable frequency ranging
between 1 and 100 kHz (Maione et al. 2002; Newnham, Xu, & Yoshikawa 1991;
Newnham, Xu, & Yoshikawa 1994).
In the design of a cymbal transducer, the lead zirconate titanate (PZT) ceramic
disk is sandwiched between metal endcaps, maintaining a shallow cavity beneath the
surface. The radial motion of the ceramic is converted into the axial displacement of the
endcaps, which is the fundamental vibration mode of the transducer. A cross-sectional
diagram of the cymbal transducer is provided in Figure 3.1. The metal endcap has a
cavity radius ( cR ), a cavity depth ( cd ), dimple diameter ( dR ), and thickness ( et ). As a
standard design with a 1 mm thick PZT disk, two brass endcaps of cR = 3 mm, cd = 0.32
mm, dR = 9 mm, and et =0.25 mm were used to construct a circular cymbal transducer.
18
3.2 Acoustic Characteristics of Cymbal Transducer
In two-dimensional finite element models of the cymbal transducers, the metal
caps and the piezoelectric element were considered. In addition, an acoustic fluid as a
medium was considered in the models. Based on the finite element models, the
resonance frequencies of the cymbal transducer in air and water were determined.
The fundamental vibration mode of the endcap shapes corresponding to the (0, 1)
vibration mode of the endcap and the equation of motion of the endcap is very similar to
that of a thin plate:
Figure 3.1: Diagram of cymbal transducer in cross sectional view.
)()( 00 KrBIKrAJr , (Equation 3.1)
Cavity
Cavity
PZTBondinglayers
Endcaps
Cavity
Cavity
PZTBondinglayers
Endcaps
19
where A and B are unknown constants, is the displacement of the endcap across its
radius ( r ), K is elastic constant of the plate, 0J is the zero order Bessel function and 0I
is the modified zero order Bessel function (Kinsler et al. 1982). The solution of the
equation of motion is given by Equation 3.2:
where rw is the endcap displacement across the radius, the C ’s are unknown constants
and is defined by the endcap material density ( e ), the endcap thickness ( et ), flexural
modulus of elasticity ( D ), the Young’s modulus ( E ) and the endcap radius ( eR ) as
Equation 3.2 is re-written by applying a rigid clamping boundary condition around the
periphery of the endcap ( ar ):
Since 1C is the unknown constant, proportionality can be derived from this equation
instead of exact solutions. Based on the proportionality and some algebra, the
fundamental resonance frequency ( arf , ) in the air is derived from Equation 3.3 and
described by Equation 3.5:
where, d is the dimple diameter ( dR2 ) and is Poisson’s ratio (Tressler 1997).
30201 )()( CrICrJCrw , (Equation 3.2)
2
24
e
eee
DR
Et
D
t
. (Equation 3.3)
)(
)()()()()(
1
100001 aI
aJaIrIaJrJCrw
. (Equation 3.4)
222,
1
1
1~
ede
arR
Ef
, (Equation 3.5)
20
If the transducer is operated in water or another medium, the resonance frequency
changes due to the extra mass, which dampens the amplitude of resonance and decreases
the frequency at resonance. In water or another medium, the resonance frequency is
given as Eq. 3.6:
where w,rf is the in-water resonance frequency, a is the radius of the cymbal transducer,
and w is the water density (kg/m3).
The fundamental resonance frequency is affected by the type and thickness of
PZT disk, the material and thickness of the endcap, the depth and radius of cavity, and
the bottom radius of the cavity. For different PZT types and endcap dimensions, the
resonance frequency in air or water is presented in the Appendix A.
3.3 Acoustic Cavitation
Ultrasound-mediated transdermal method is based on the hypothesis that, once the
drug has been transported through the outermost layer of skin (i.e., stratum corneum), the
next layer is easier to cross, enabling the drug to reach the capillary vessel. Such an
assumption is based on the facts that ultrasound can permeate the skin and that the
glucose level of interstitial fluid is highly correlated to the blood glucose level. Although
ultrasound is known to enhance transdermal drug or protein delivery, the working
mechanisms have yet to be completely understood. The cavitation effect, which is one of
ee
w
arwr
t
a
ff
7885.0
3
81
,,
, 3.6
21
the bioeffects of ultrasound, has been suggested as the primary mechanism for enhancing
the efficacy of ultrasonic transdermal drug delivery. (Guzman, McNamara, Nguyen, &
Prausnitz 2003b; Mitragotri, Edwards, Blankschtein, & Langer 1995; Mitragotri,
Blankschtein, & Langer 1997a; Schlicher, Radhakrishna, Tolentino, Apkarian, Zarnitsyn,
& Prausnitz 2006).
3.3.1 Definition
Acoustic cavitation is defined as the generation and activities of gas- or vapor-
filled bubbles in response to large rarefractional pressure in an aqueous solution
(Leighton 1994). In addition to the excitation of gaseous bubbles already existing in the
solution, acoustic waves with large rarefractional pressure can create bubbles when the
pressure is great enough to overcome the tensile strength of the solution. Common
terminology distinguishes two types of cavitations: stable and transient cavitation
(Carstensen & Flynn 1982).
In stable cavitation, which represents the oscillation of bubbles, the bubble radius
varies about an equilibrium value and violent collapse does not occur as the acoustic
intensity and frequency are not high enough to cause such a collapse. However,
microstreaming can occur near the boundaries as a result of the conversion of absorbed
acoustic energy into the flow (Leighton 1994). Such microstreaming can produce sheer
forces in tissue that have a rigid boundary. If the sheer forces are great enough to disrupt
cells, they can enhance skin permeability. Meanwhile, transient cavitation, which
oscillates in an unstable manner of the equilibrium radius, occurs at greater acoustic
22
pressure. Transient cavitation includes the violent collapsing of bubbles after a few
cycles of oscillation. As the result of the violent collapsing of transient cavitation,
significant energy is transformed to create a high energy jet stream, which can disrupt the
cells (Mitragotri & Kost 2004).
3.3.2 Cavitation Threshold
Although several thresholds may be involved in acoustic cavitations, Neppiras
defined two thresholds, the stable cavitation threshold and transient cavitation threshold.
Thresholds are described in terms of acoustic pressure amplitude ( AP ), ambient pressure
( 0P ), initial bubble radius ( 0R ), and angular frequency of applied ultrasound ( ).
During refraction, the gas flow rate into a bubble ( dtdm / ) that includes surface
tension is given by
where D is the diffusion constant, C is the gas concentration in the liquid in the
absence of the bubble, and is surface tension (Neppiras 1980a). Similarly, the gas
flow rate out of the bubble during compression is given by
A
A
PRP
PRDC
dt
dm
000
21
3
8 , (Equation 3.7)
000
0
214
C
C
PRRDC
dt
dm , (Equation 3.8)
23
where 0C is the saturation gas constant in the liquid. When the gas flow during both
phases of pressure cycles is equal, a threshold condition is reached and it is described by
Equation 3.9
The threshold is more exactly described by considering damping factor ( ), imposed
angular frequency ( ) and resonant angular frequency ( 0 ) (Neppiras 1980a; Neppiras
1980b):
where 0
220
2
0
2
3 P
R
. While the resonance of a stable bubble for frequencies up
to 100 kHz occurs at (Minnaert 1933)
where is the specific heat ratio of the gas inside the bubble, the resonance above 100
kHz is defined by taking into account the surface tension:
Known as the ‘Blake’ threshold (Blake 1949), the threshold of transient cavitation
for small bubbles (well below resonance size) is described by
1
000
2
0
211
2
3
PRC
C
P
PA . (Equation 3.9)
2222
1
000
2
0
)1(2
112
3
PRC
C
P
PA (Equation 3.10)
2/1
0
00
31
P
R, (Equation 3.11)
2/1
00
0
00 3
)13(21
31
PR
P
R
(Equation 3.12)
24
For bubbles have greater size than resonance size, this threshold becomes
where rR is the resonant radius and 0/ PPp A .
1
00
30
30
232
3
4
RPRPPT
. (Equation 3.13)
,)1(3
21)1(46.0
3/12/10
ppp
R
R
r
(Equation 3.14)
Chapter 4
MATERIALS AND METHODS - APPRATUS
4.1 Construction of Cymbal Transducer Arrays
Based on the design and basic theory of the cymbal transducers described in
sections 3.1 and 3.2, three types of transducer arrays were constructed using the circular
and rectangular cymbal transducers. The design and fabrication procedures of the
cymbal transducer are protected by US Patents 5729007 and 6232702 (Newnham &
Dogan 1995; Newnham & Zhang 1999). This section gives the detail descriptions of the
assembly and acoustical evaluation of the cymbal transducers and arrays and presents an
acoustical evaluation of them.
4.1.1 Fabrication of cymbal arrays
The piezoelectric disk having a diameter of 12.7 mm and a thickness of 1 mm was
made from PZT-4 (Piezokinetics, Inc., Bellefonte, PA). To allow electricity to be
conducted uniformly through the piezoelectric disk, both axial surface of the disk were
covered with silver electrodes (Figure 4.1). Before bonding the disk to metal endcaps,
the electrodes were slightly scraped with 500 grit sandpaper (3M, St. Paul, MN) to
remove an accumulated oxidation layer and expose a clean silver surface. After scraping
the oxidation layer, residual dust on the surface of the disk was removed using acetone.
26
The metal endcaps with the shape of a musical cymbal were made of 0.25 mm
thick brass. To obtain flat and circular disk with a diameter of 12.7 mm, purchased brass
sheet (0.25 mm thick) was punched using a manual punching set (Precision Brand,
Downners Grove, IL). The punched brass disk was molded to have a cymbal shape by
pressing it at 60 kpsi in a hydraulic press. The diagram in Figure 4.2 shows the
dimensions of the endcaps having the standard cavity depth, 0.32 mm (Tressler 1997).
For symmetry, the cavity depth of each pressed endcap was measured at the center of the
cap by using a Starrett dial gauge. Then, caps with the closest cavity depth were paired
together.
Figure 4.1: PZT-4 disk and a brass endcap
27
Two cymbal shape endcaps were bonded to the piezoelectric disk using
Eccobond® 45 LV epoxy base and 15 LV (Emerson & Cuming, Billerica, MA) catalyst
with a 3:1 ratio. To increase the electrical conductivity between the caps and the
piezoelectric disk, silver powder (<250 μm, Sigma-Aldrich Corp., St. Louis, MO) was
added to the epoxy mixture in a 1:4 ratio. The epoxy mixture was carefully spread on the
edge of the disk, and then the caps were placed on the disk. To ensure a better seal,
binder clips were placed along the edges of the transducer. After binding caps on the
disk, the cymbal transducer was placed at room temperature for 24 hours.
While the cymbal transducer was cured, a 1 mm thin polyurethane sheet was
constructed for a cymbal array template. A two part polyurethane (Por-A-Mold 2160
A/B, PathwayPolymers, Chattanooga, TN) was thoroughly mixed in 1:1 volume ratio.
Figure 4.2: Dimensions of endcaps and PZT-4 disk
28
To minimize air bubbles, the polyurethane mixture was degassed and carefully poured
onto a glass plate coated with a mold release (ease release 200, Mann Formulated
Products, Easton, PA). The mixture was pressed by another glass plate with 1 mm thick
spacers at each corner and left for 24 hours for curing. To construct the cymbal
transducer array in 2 × 2 and 3 × 3 patterns, the polyurethane sheet was cut in a size of 30
× 30 mm2 and 50 ×50 mm2. Four holes of 12.7 mm diameter were punched on the sheet
for the 2 × 2 array and nine holes were punched for the 3 × 3 array. The polyurethane
sheet was also used as a spacer when potting transducer arrays.
After placing the cymbal transducers into the template, a conductive adhesive
(silver epoxy, E-Solder No. 3021, VonRoll ISOLA, New Haven, CT) and flexible,
nonmagnetic, stranded wires (26 AWG 66/44, Pegasus Cable, Inc., Wilmington, DE)
were used to connect the circular cymbal transducers. To ensure a proper electrical
connection, the transducers were left to dry overnight. Once the transducers and wires
were connected, they were encased in the same polyurethane used for the cymbal
template. After mixing and degassing to avoid capturing air bubbles, the polyurethane
mixture was carefully poured in a silicon compound mold for each array shape. Once the
transducers are submerged in the polyurethane mixture, the array was left for 24 hours to
cure. The 2 × 2 and 3 × 3 cymbal arrays are shown in Figure 4.3. The step-by-step
instruction of the cymbal array construction is described in appendix B.
29
4.1.2 Construction of matching circuits
In order to obtain maximum power transfer from the operating system to the
transducer arrays that have input electrical impedance, it is necessary to match the
impedance of the transducer to the output impedance of the operating system. In this
study, a π-type network circuit was used for the impedance matching to a 50 Ω output
impedance of amplifiers. Theoretical values of C1, C2, and L1 were obtained using a
MATLAB code for measured impedance value of each array. After getting theoretical
values, actual circuits shown in Figure 4.4 were constructed to obtain the maximum
acoustic output by adjusting the value of each component (C1, C2, and L1).
Figure 4.3: Lightweight cymbal arrays were constructed using four transducers in a two-by-two pattern, (left) or nine cymbal transducers in a three-by-three pattern (right). The two-by-two array was used for ultrasonic glucose sensing, while the three-by-three array was used in the transdermal insulin delivery.
30
4.2 Exposimetry: Acoustic Characteristics of Cymbal Arrays
According to the exposimetry guidelines established by the American Institute of
Ultrasound in Medicine (AIUM 1993), the acoustic pressure field of the array was
determined with a calibrated miniature omnidirectional reference hydrophone (Model
TC4013, S/N: 5199093, RESON, Inc., Goleta, CA). The cymbal array was submerged in
a water tank (51 x 54 x 122 cm3) which was made almost anechoic by placing 1.27 cm
thick rubber sound absorbing material around its wall (Figure 4.5). To avoid cavitation
effects, the tank was filled with distilled water degassed by a custom made degasser to
reduce the dissolved oxygen content to 1-2 ppm. The position of the hydrophone was
precisely controlled by a Velmex Positioning System (Velmex Inc., East Bloomfield,
NY). In order to obtain the acoustic pressure profile over a plane 2 mm away from the
array surface, automatic scannings with 1 mm step size were performed by the computer
Figure 4.4: Electric impedance matching network using a π-type circuit. After constructing the circuit based on the theoretical values, actual value of each componentwas adjusted to have maximum acoustic output.
31
controlled exposimetry positioning system. The automatic scanning was performed over
the area of 50 × 50 mm2. Based on three scannings, spatial peak-temporal peak (Isptp)
intensities of each array in a mean and standard deviation were determined by the
Equation 4.1 .
Maximum voltage amplitude of signal ( opV ) received by hydrophone were converted
into maximum pressure amplitude ( opP ) by the sensitivity of the hydrophone ( A ). With
the density ( ) of 998 kg/m3 and the speed of sound ( c ) of 1482 m/s, Isptp was
determined for each array.
cA
V
c
PI opop
sptp 2
22
22 . (Equation 4.1)
Figure 4.5: Equipment setup of the exposimetry to characterize the cymbal arrays. Acoustic pressure field of each cymbal array was measured in the water tank filled withdegassed water.
Cymbal Array Agilent
33250A
AR
25A250
Ch1 Ch2
Oscilloscope Agilent 54622A
Matching circuit
Hydrophone Reson TC4013
Preamp Reson VP1000
Water tank
32
To drive the array, a radio frequency (RF) signal was generated by a
function/arbitrary waveform generator (33250A, Agilent, Santa Clara, CA) and amplified
by an RF amplifier (Model 25A250, Amplifier Research, Souderton, PA). Using an
external tuning network, the electrical impedance of the array was tuned to the output
impedance of the amplifier. The pulse duration and repetition period of the RF signal
were monitored using a digital oscilloscope (54622A, Agilent, Santa Clara, CA). In
order to determine electrical input condition to the cymbal array with the matching circuit,
input voltage to the array was measured for each array by using an electrical test probe
(100×, Tek P5100, Tektronix, Inc., Beaverton, OR) with the exposimetry setup.
4.3 Fabrication and Calibration of Electrochemical Glucose Sensors
As part of the noninvasive glucose monitoring system, disposable biosensors are
used in the current study. Biosensors using electrochemical reaction have been used to
monitor changes in physiological parameters. Based on the electrochemical properties,
enzyme-coated biosensors can be used as glucose sensors.
In the presence of a glucose oxidase enzyme (GOx), glucose reacts with oxygen
(O2) and water (H2O) to produce gluconic acid and hydrogen peroxide (H2O2) (Wilkins &
Atanasov 1996):
Glucose + O2 + H2O (GOx) Gluconic acid + H2O2.
When a positive electric potential is applied in the presence of platinum as a catalyst, two
hydrogen peroxide molecules are oxidized into oxygen, water, and four electrons:
2H2O2 (Pt) 2O2 + 2H2O + 4 e-.
33
The current produced in this reaction is proportional to the concentration of hydrogen
peroxide, which is proportional to the glucose concentration.
4.3.1 Fabrication of glucose sensors
In this study, the biosensors for the transdermal glucose sensing were fabricated
based on the electrochemical reaction of enzyme coated electrodes. For the biosensors,
polymer thick film electrodes were acquired from Conductive Technologies Inc. (York,
PA). The biosensor consists of three electrodes: working, counter and reference
electrodes (Figure 4.6). In order to get the current from the electrochemical reaction, the
working electrode should be deposited with platinum which worked as a catalyst. Once
platinic acid of 25 mM (Sigma-Aldrich Corp., St. Louis, MO) in 0.1 M NaCl solution
was placed over three electrodes, a voltage of -0.7 V between working and counter
electrodes was applied for 60 minutes using a potentiostat (Model 283, Princeton Applied
Research, Oak Ridge, TN). After the washing the platinum deposited electrodes with
deionized distilled water, glucose oxidase (1,000,000 unit/g, Sigma-Aldrich Corp., St.
Louis, MO) was immobilized on the three electrodes by polyethylene glycol (PEG)
hydrogel. After PEG diacrylate (MW = 575, Aldrich Corp., St. Louis, MO) was diluted
with distilled water to produce a 20% solution, glucose oxidase (GOx) was mixed. After
adding liquid photoinitiator (2-hydroxy–2-methyl-1-phenyl-1-propanone, Ciba Specialty
Chemicals, Tarrytown, NY) the PEG-GOx solution, three electrodes were covered with
the solution and exposed to 6 W, 365 nm UV light for 20 seconds in air.
34
4.3.2 Calibration of glucose sensors
To determine the unknown glucose concentration of interstitial fluid, each
biosensor was calibrated using a concentration curve created by a series of glucose
solutions, 0, 50, 100, 150, 200 and 300 mg/dl. The glucose solutions were prepared by
dissolving glucose (Dextrose Anhydrous, VWR, West Chester, PA) in 0.9 % phosphate
buffered (PBS) solution. Once the three electrode part of the biosensor was submerged in
the calibration solution and connected to the potentiostat, a voltage of 0.7 V was applied
to the working electrode against the counter electrode. The voltage applied to the
working electrode was increased from 0 V to 0.7 V with 10 mV/sec increasing ratio. To
minimize the residual effect of the glucose solution, the biosensors was gently washed
Figure 4.6: Photograph of an enzyme based electrochemical biosensor which consists of three electrodes: working/sensing, counter, and reference electrodes
Counter electrode (Carbon)
Working electrode
(Platinized Carbon)
Reference electrode (Silver/ Chloride)
35
with deionized distilled water and the calibration was performed from the lower glucose
concentration to the higher concentration.
4.4 Closed-Loop Feedback Controlled Ultrasound System for Insulin Delivery and Glucose Sensing
A simple proportional-plus-integrator-plus-derivative (PID) controller has been
used in this study. There are three parts to the PID controller: the proportional gain PK ,
the integral gain IK , and the derivative gain DK . These three values determine how the
controller responds to error in the current state of the system ( G ) with respect to the
desired state ( refG ). In a closed-loop system, the tracking error ( E ) is the difference
between the current output and the desired output. After the tracking error is sent to the
PID controller as its input (Figure 4.7), the output signal of the controller (U ) will be
obtained as the summation of the proportional gain multiplied by the error, the integral
gain multiplied by the integral of the error, and the derivative gain multiplied by the
derivative of the error:
This signal will be sent to the plant and the new output will be obtained. The new output
will be sent back again to find the new error. For a PID controlled system, increasing the
proportional gain factor PK , decreases the rise time but oscillations increase. By
increasing the integral gain factor IK , the noise tolerance improves but the settling time
dt
dEKdt EKKU DIP . (Equation 4.2)
36
lengthens. The damping is increased and any overshoot is reduced by increasing the
derivative gain factor DK .
The ultrasound systems for the transdermal glucose sensing and insulin delivery
were combined through a proportional closed-loop feedback controller. In the controller,
the reference glucose concentration, reference glucose concentration was set as 115
mg/dl. The difference between the reference and measured values was defined as an
error and used as the input of the controller. Using the constant of proportionality of the
system, PK set as 1, the input on/off signal U was sent to the ultrasound systems. Using
a control program written in MATLAB® (Mathwork, Inc., Natick, MA), the ultrasound
systems were operated automatically. After the reference glucose concentration and the
exposure time of the ultrasonic glucose sensing system were set, the repetition times of
the measurements and the total number of measurements were set. Once the control
program was started, the glucose sensing system was operated and the first glucose level
measurement was taken by converting measured current value based on the calibration
curve of each biosensor. Depending on the glucose levels, the insulin delivery system
Figure 4.7: Block diagram of a proportional-plus-integral-plus-derivative (PID) controller. The variables are defined in table 1. The plant is a system controlled by thecontroller. In this study, it is the ultrasound system for insulin delivery and glucosemonitoring.
Controller
KP, KI, KD
Plant U
+ E
G
–
Gref
37
either operated or stopped. In this study, the ultrasound exposure time for glucose
sensing was set at 5 minutes, and the measurements were performed every 20 minutes for
120 minutes. The definitions and values of the variables in Figure 4.7 are listed in Table
4.1.
Table 4.1: To control the ultrasound system using cymbal arrays, a simple proportionalcontroller was used.
Symbol Definition
refG Desired blood glucose concentration, 15 mg/dl
E Error between the measured and reference values,
refGGE
PK Proportional gain factor, 1
IK Integral gain factor, 0
DK Derivative gain factor, 0
U Input to the system, EKU P
G Measured blood glucose concentration
Chapter 5
MATERIALS AND METODS – IN VIVO EXPERIMENTS
To evaluate the cymbal transducers in transdermal insulin delivery and glucose
sensing, several in vivo experiments were performed in large pigs. This section decribes
each in vivo experiments for (1) the feasibility of transdermal insulin delivery, (2) the
feasibility of using the cymbal array in ultrasonic glucose sensing , and (3) the feasibility
of glucose control by a feedback controlled ultrasound system using the cymbal arrays.
As a milestone to a preclinical application, in vivo experiments were performed in large
pigs possessing a weight and blood volume similar to humans.
All procedures described in this report involving live animals were approved by
the Institutional Animal Care and Use Committee (IACUC) at the Pennsylvania State
University. Each animal was pre-anesthetized for intubation with a combination of
ketamine hydrochloride (10 - 12 mg/kg intramuscularly, Ketaject®, Phoenix, St. Joseph,
MO) and sodium xylazine (1 - 2 mg/kg intramuscularly, Xyla-Ject®, Phoenix, St. Joseph,
MO). Pigs were fitted with an intravenous (IV) catheter in the auricular vein and an
endotracial tube (size 6-7) was inserted into the airway. Anesthesia throughout the
remaining experiment was maintained to surgical depth via inhalant isoflurane
(IsothesiaTM, Abbott Laboratories, North Chicago, IL) using an inhalational anesthesia
unit (Narkovet Deluxe, North American Drager, Telford, PA).
39
5.1 Transdermal Insulin Delivery
Six Yorkshire pigs (100 – 140 lbs) obtained from the Penn State Swine Center
were divided into two experimental groups. As the control (n = 3), the first group did not
receive any ultrasound exposure with the insulin while second group (n = 3) was treated
with ultrasound and insulin. Under the anesthesia, the axillary area of the pigs were
shaved using an electric shaver, and a depilatory agent was applied to the skin of both
groups, control and exposure, to eliminate any remaining hair. With the pig in the lateral
recumbency (Figure 5.1), a 1 mm thick, water-tight standoff was attached between the
skin and the 3 × 3 array using tissue glue (Vetbond®, 3M, St. Paul, MN). The reservoir
within standoff was filled with insulin (Humulin® R, rDNA U-100, Eli Lilly and Co.,
Indianapolis, IN) through a small hole in the back of the array for both the control and
exposure experiments. To prevent disruption of ultrasound transmission, care was taken
to remove any bubbles from the solution in the reservoir (standoff) between the axillary
area and the array.
40
At the beginning of the experiment, blood sample (0.3 ml) was collected from the
ear vein of each pig for a baseline glucose level analysis. The glucose level (mg/dl) in
the blood was determined using ACCU-CHEKTM blood glucose monitoring system
(Roche Diagnostics Co., Indianapolis, IN, USA). Multiple blood samples (2-4 each time)
were taken every 15 minutes for 90 minutes. Elapsed time from the initial injection of
the ketamine-xylazine until the first glucose measurement was no greater than 15
minutes. For comparison between the pigs, the change in the blood glucose level was
normalized to a baseline with respect to each animal's initial blood glucose recording at 0
minutes.
For comparison between the results of each pig, the change of blood glucose level
was normalized with respect to the initial blood glucose level of each pig. The control
group (three pigs) used insulin inside the reservoir without ultrasound exposure while the
Figure 5.1: Photograph of a transdermal insulin delivery experiment with pig placed in alateral recumbent position with the array attached.
ArrayArrayArray
41
second group (three pigs) was treated with ultrasound and insulin at 30 kHz with an Isptp
= 100 mW/cm2 for 60 minutes. For both groups, the standoff reservoir with the insulin or
saline was removed at 60 minutes although glucose determination continued until 90
minutes from the start. At the end of the experiments, the pig was euthanized
(Pentobarbitol, Fatal Plus, 130 mg/kg IV, Vortech Pharmaceuticals, Ltd., Dearborn MI)
under anesthesia.
Statistical analysis was performed using Microsoft Excel® (Microsoft Corp.,
Redmond, WA). The blood glucose verses time data were pooled for each group and
analyzed as its mean and standard error (x ± s.e.). An ANOVA was used to analyze the
statistical significance of the differences among the means of groups. The p-value was
used to determine if the between-group differences are significantly greater than chance.
For all the data, a single or double asterisk was used if the p-value is less than the 0.05 or
0.01 level of significance, respectively.
To answer how much insulin is transported through the skin, additional
transdermal insulin delivery experiments were performed at 20 kHz. Six pigs were
divided into a control and ultrasound exposure groups. Blood samples of 5 ml were
collected every 15 minutes for 90 minutes to analyze insulin concentrations as well as
glucose levels. At the completion of the experiments, each pig was euthanized
(Pentobarbitol, Fatal Plus, 130 mg/kg IV, Vortech Pharmaceuticals, Ltd., Dearborn MI)
under anesthesia.
The concentrations of insulin in the blood samples were analyzed by enzyme
linked immunoassay (ELISA) tests. In the blood samples, both human insulin and
porcine insulin were found and there is 99% similarity between the two types of insulin.
42
Therefore, the blood samples were analyzed for both human insulin and porcine insulin
using Mercodia ELISA kits (Insulin ELISA, 10-1113-01 and Porcine Insulin ELISA, 10-
1200-01, Mercodia, Uppsala, Sweden). The insulin ELISA is a solid-phase two-site
enzyme immunoassay and is based on the direct sandwich technique in which two
monoclonal antibodies are directed against separate antigenic determinants on the insulin
molecule. During incubation, insulin in the sample reacts with peroxidase-conjugated
anti-insulin antibodies and anti-insulin antibodies bound to the microtitration well. A
simple washing step removes the unbound enzyme-labeled antibody. The bound
conjugate is detected by reaction with 3, 3', 5, 5'-tetramethylbenzidine (TMB). The
reaction is halted by adding acid to give a colorimetric endpoint that is read
spectrophorometrically. After preparing the enzyme conjugate solution, wash buffer,
and sufficient microplate wells that can accommodate calibrators and samples in
duplicate, 25 µl of calibrators and samples were pipetted into the microplate wells. Then,
100 µl of enzyme conjugate solution was added to each well and the plate was incubated
on a plate shaker (700 - 900 rpm) for one hour at room temperature (18 - 25 °C). The
incubated plate was washed six times with an automatic plate washer. After the final
wash, the plate was inverted and firmly tapped against absorbent paper to remove the
remaining washing solution. During the next step, 200 µl of substrate TMB was added
into each well and the plate was incubated for 15 minutes at room temperature. After the
incubation, 50 µl of stop solution was added into the well and the plate was placed on a
shaker for 5 minutes to ensure thorough mixing. Within 30 minutes after the final step,
the plate was placed in an ELISA reader and the optical density of each well was read at
450 nm and the insulin concentration of each well was calculated. Since there was a
43
maximum of 28% cross-reaction between human and porcine insulin for the ELISA tests,
additional calculation of the insulin concentration was performed to compensate for the
cross-reaction.
5.2 Ultrasonic Transdermal Glucose Sensing
For perspective ultrasonic noninvasive glucose monitoring, eight Yorkshire pigs
(~ 200 lbs) obtained from The Pennsylvania State University Swine Center were used in
five groups according to the ultrasound exposure conditions. After anesthesia, hair on the
axillary area of pigs in lateral recumbency was removed and the 2 × 2 cymbal array with
double layers of a 1 mm thick and water-tight standoff made of plastic was place on the
skin. Figure 5.2 shows the placement of array and the biosensors on the pig. Four
biosensors were placed between two layers of the standoff and the reservoir within the
standoff was filled with 0.9% PBS solution. The volume of PBS solution was 2.25 ml.
Since it was assumed that the concentration of glucose in the collection chamber depends
on the volume of PBS in the reservoir, the volume was kept as 2.25 ml for all
experiments. The experiments were performed on the same location of each pig.
44
To permeabilize the skin for extraction of glucose in interstitial fluid, the cymbal
array was operated at a frequency of 20 kHz with an intensity of Isptp = 100 or 50
mW/cm2 for 5, 10 or 20 minutes (Table 5.1). After connecting the biosensors to the
potentiostat at the end of the ultrasound exposure, the current from the electrochemical
reaction on the biosensors was measured by applying a voltage of 0.7 V to the working
electrode against the counter electrode. Using the potentiostat (Model 283, Princeton
Figure 5.2: (a) Layout of the glucose sensor incorporated with the cymbal transducerarray placed on the skin surface. (b) Photograph of the ultrasonic glucose sensing on a pig with the array attached to the skin and the reservoir filled with PBS solution.
Pig
Cymbal Array
Saline
Biosensors
2 mm
Pig
Cymbal Array
Saline
Biosensors
2 mm
Cymbal Array
Biosensors
Cymbal Array
Biosensors
Cymbal Array
Biosensors
(b)
(a)
45
Applied Research, Oak Ridge, TN), the voltage applied to the working electrode was
increased from 0 V to 0.7 V with 10 mV/sec increasing ratio. Any false currents caused
by interference were observed. Using the potentiostat control program (Electrochemistry
PowerSuite™ v. 2.1.1, Princeton Applied Research, Oak Ridge, TN), the current was
recorded on a computer. Based on the biosensor calibration results and the linear
relationship between the current and the glucose concentration, the glucose levels in
interstitial fluid were determined. For the comparison to the glucose levels determined
using biosensors, blood samples were collected at the same time when the currents from
the biosensors were measured. The glucose level of blood sample was measured by
ACCU-CHEKTM blood glucose meter (Roche Diagnostics Co., Indianapolis, IN) and i-
STAT portable clinical analyzer (MODEL 200, Abbott Laboratories, Abbott Park, IL).
Table 5.1: The detail information of ultrasound exposure conditions.
Acoustic intensity, Isptp Exposure time (minutes)
20
10 100 mW/cm2
5
10 50 mW/cm2
5
46
5.3 Closed-Loop Feedback Controlled Ultrasound System for Glucose Control
Four Yorkshire pigs (~ 200 lbs) were obtained from the evaluation of the closed-
loop feedback controlled ultrasound system. After anesthesia, hairs on the axillary area
and medial area of the front leg of the pig in lateral recumbency were removed.
Figure 5.3 shows the layout of the glucose control system and the experimental setup.
For extracting glucose in interstitial fluid, the 2 × 2 array with double layers of a 1 mm
thick and water-tight standoff was attached to the skin on the medial area of the front leg
using tissue glue (Vetbond®, 3M, St. Paul, MN). Two biosensors were placed between
two layers of the standoff and connected to the potentiostat to obtain current
measurements. For extracting glucose in interstitial fluid, the 2 × 2 array with double
layers of a 1 mm thick and water-tight standoff was attached to the skin on the medial
area of the front leg using tissue glue (Vetbond®, 3M, St. Paul, MN). Two biosensors
were placed between two layers of the standoff and connected to the potentiostat to
obtain current measurements. The interstitial fluid reservoir within the standoff was filled
with 0.9 % PBS solution through a small hole in the back of the array. To deliver insulin
transdermally, the 3 × 3 array with a single layer of the standoff was attached to the skin
on the axillary area, and the reservoir was filled with insulin (Humulin® R, rDNA U-100,
Eli Lilly and Co., Indianapolis, IN).
47
Figure 5.3: (a) The layout of the glucose sensing system incorporated with the insulin delivery system as placed on the skin surface. After five minutes of ultrasound exposure, the biosensor determined the glucose concentration of the interstitial fluid. Based on theglucose level, the insulin delivery system was automatically operated. (b) Photograph of the ultrasound systems placed on the pig for insulin delivery and glucose sensing. Theultrasound system was controlled by a closed-loop controller.
(a)
(b)
48
Figure 5.4 shows the experimental setup with a closed-loop feedback controlled
ultrasound system using cymbal arrays. To permeabilize the skin for glucose extraction,
a RF signal of 20 kHz was generated, amplified, and sent to the 2 × 2 cymbal array in the
glucose sensing system (Figure 5.4 solid line). The array was operated at a 20% duty
cycle (a pulse duration of 200 ms and a pulse repetition period of 1 second) with an
intensity (Isptp) of 100 mW/cm2 for 5 minutes. At the end of the ultrasound exposure, the
current from the electrochemical reaction on the biosensors was measured by applying a
voltage of 0.7 V to the working electrode against the counter electrode. Using the
potentiostat control program (Electrochemistry PowerSuite™ v. 2.1.1, Princeton Applied
Research, Oak Ridge, TN), the current was recorded in a computer. According to the
biosensor calibration results and the linear relationship between the current and glucose
concentration, the glucose levels in the interstitial fluid were determined. Based on the
glucose level determined by the biosensor, the error between the reference and measured
values was calculated. According to the calculated error, the insulin delivery system
using the 3 × 3 cymbal array (Figure 5.4 dashed line) was either turned on or off. If the
glucose level was determined to be higher than the reference value, 115 mg/dl, the
control program turned on the arbitrary waveform generator to generate a RF signal at 30
kHz. The amplified signal was sent to the 3 × 3 cymbal array through an impedance
matching network. For the insulin delivery, the array was operated at a 20% duty cycle
with Isptp = 100 mW/cm2. After 15 minutes of ultrasound exposure for the insulin
delivery, the glucose sensing with the biosensor was performed again. Using the same
procedure, the glucose level was determined, and the ultrasound system for the insulin
delivery was operated. The closed-loop control of the glucose level was performed for
49
120 minutes. Every 20 minutes for the 120 minutes period, the biosensors determined the
glucose level in the interstitial fluid. Additionally, for comparison with the glucose levels
that were determined using the biosensors, the glucose levels of blood samples were also
measured by the ACCU-CHEKTM blood glucose meter (Roche Diagnostics Co.,
Indianapolis, IN) and the i-STAT portable clinical analyzer (MODEL 200, Abbott
Laboratories, Abbott Park, IL). The results of the glucose meter tests were presented in a
mean and standard error.
Figure 5.4: Schematic diagram shows the experimental setup with the closed-loop feedback controller for noninvasive glucose regulation. The ultrasound operation forboth the glucose sensing and insulin delivery systems was automatically operated.
Chapter 6
RESULTS
6.1 Acoustic Pressure Distribution of Cymbal Transducer Arrays
From multiple exposimetry scans, the intensity of the 2 × 2 array was obtained as
Isptp = 100.8 ± 0.9 mW/cm2 while the intensity of the 3×3 array was Isptp = 100.0 ± 0.5
mW/cm2. Spatial distributions of temporal peak intensities of the arrays were plotted in
dB with respect to the highest intensity (Isptp) to show spatial distribution of the intensity
field from the maximum radiation point (Figure 6.1). Using similar Isptp driving
conditions, the intensity field was determined in a plane 1 mm from the array face for
both the rectangular and the circular designs. Figure 6.1 shows typical two dimensional
scanning plots of the temporal peak intensity for the 2 × 2 and 3 × 3 cymbal arrays over a
50 × 50 mm2 area with 1 mm steps. The color bar on the side of the graph illustrates the
normalized temporal peak intensity change in dB. In order to obtain the intensity of Isptp
= 100.0 mW/cm2, the 2 × 2 cymbal array was operated at the input voltage of 613 ± 12.5
V (peak-to-peak) while the 3 × 3 cymbal array was operated at 1.03 ± 0.05 kV. Both
arrays had the spatial peak intensity near the center of each array. In order to estimate the
power consumption of each array, small resister (1 Ω) was added between the amplifier
and the matching circuit. Currents to the 2 × 2 and 3 × 3 cymbal arrays, which were
determined by measuring the voltages across the small resister, were 525 ± 11 mA and
1.25 ± 0.01 A, respectively.
51
6.2 Transdermal Insulin Delivery
6.2.1 In Vivo Experiments in Large Pigs
Results of the ultrasonic transdermal insulin delivery in large pigs for the two
groups are graphed in Figure 6.2 as change in the blood glucose level during the 90
minute experiment in terms of the mean and standard error. After the pigs were
anesthetized, the average initial glucose level of the six pigs was 146 ± 13 mg/dl.
Generally for pigs, blood glucose level is approximately 100-110 mg/dl (Danfaer 1999;
Pond & Houpt 1978). As mentioned, for a comparison between the pigs, the change in
Figure 6.1: Acoustic intensity profiles of the two-by-two (left) and three-by-three (right) arrays were determined by measuring acoustic pressure and plotted in dB. The spatial-peak-temporal-peak intensities were 100.8 ± 0.9 mW/cm2 for the two-by-two array and 100.0 ± 0.5 mW/cm2 for the three-by-three array.
(a) (b)
52
the blood glucose level was normalized to a baseline which was the initial glucose level
for each pig.
For the control group (insulin without ultrasound), the glucose level increased to
31 ± 21 mg/dl compared to the initial baseline over the 90 minute experiment. The slope
of this increase was +20 mg/dl/hour (r2 =0.9). In contrast with the ultrasound exposure
group (insulin with ultrasound), the glucose level decreased to –74 ± 5 mg/dl at 60
minutes and continued to decrease to –91 ± 9 mg/dl at 90 minutes after the standoff was
removed. Additionally, a gross examination of the pig's skin was performed after
exposure to look for visible lesions on the skin surface. Visual examination of the post
Figure 6-2: Change of blood glucose level in the ultrasound mediated transdermal insulin delivery on pigs.
146 ±13
mg/dl
***
* * *
Control, insulin-no ultrasound (n = 3)Ultrasound with insulin (n = 3)
Ch
an
ge
of
blo
od
glu
cose
lev
el (
mg
/dl)
50
0
-50
-100
-150150 45 7530 60 90
Time (minutes) Standoff removed
146 ±13
mg/dl
***
* * *
Control, insulin-no ultrasound (n = 3)Ultrasound with insulin (n = 3)
Ch
an
ge
of
blo
od
glu
cose
lev
el (
mg
/dl)
50
0
-50
-100
-150
146 ±13
mg/dl
146 ±13
mg/dl
***
* * *
***
* * *
Control, insulin-no ultrasound (n = 3)Ultrasound with insulin (n = 3)Control, insulin-no ultrasound (n = 3)Ultrasound with insulin (n = 3)Ultrasound with insulin (n = 3)
Ch
an
ge
of
blo
od
glu
cose
lev
el (
mg
/dl)
50
0
-50
-100
-150
Ch
an
ge
of
blo
od
glu
cose
lev
el (
mg
/dl)
50
0
-50
-100
-150
50
0
-50
-100
-150150 45 7530 60 90
Time (minutes)
150 45 7530 60 90150 45 7530 60 90
Time (minutes) Standoff removed
53
ultrasound exposed skin did not indicate any noticeable damage or significant change to
the skin.
To determine the statistical significance between the results in Figure 6-2 of the
two groups at the 15 minute increment time points of the experiment, an ANOVA was
used to analyze the data. Asterisks above a glucose-time data point were used if a
statistical difference existed between the control and exposure. At 15 minutes the
analysis showed no statistical difference between the groups. However, a comparison of
the control to the ultrasound exposure groups at 30 minutes and greater indicated that the
results were statistically significant at a p-value of 0.05 or better.
6.2.2 Analysis of Insulin Concentration in Transdermal Insulin Delivery
Uncompensated human insulin concentrations (µU/l) of serum samples in a mean
and standard deviation are shown in Figure 6.2a with porcine insulin concentrations
(ng/l). After 15 minutes of ultrasound exposure, the human insulin concentration
increased only by 0.5 µU/l from the initial value, 40.3 µU/l. However, at 30 minutes, the
concentration increased by 23.7 µU/l for 15 minutes. After 30 minutes, human insulin
concentrations were obtained in the range of 41.8 - 48.6 µU/l. For 90-minute periods,
porcine insulin concentrations stayed in the range of 12.23-12.38 ng/l. In order to
compare the change of porcine concentrations, the human insulin concentrations were
determined in ng/l by unit conversion (MW = 5808 Da, 1 IU = 6 nmol) (Newnham, Xu,
& Yoshikawa 1991). Due to the small value, the standard deviation of human insulin
concentration is not clearly presented in the plot. Only one different amino acid is found
54
between human and porcine insulin (Berson & Yalow 1963). Therefore, there is a
maximum of 28% cross-reaction for ELISA tests between the two types of insulin.
Based on this finding that initial insulin concentration of serum samples is solely due to
the porcine insulin, the cross-reaction ratio of 13.3% was obtained. With the cross-
reaction ratio, the compensated human insulin concentrations were determined. Figure
6.2b shows the changes in porcine insulin concentrations with the changes of human
insulin concentrations for both with and without compensating for the cross-reaction.
The results of comparison between two insulin concentrations indicate that human insulin
was transported through the skin due to the ultrasonically enhanced skin permeability.
To compare the change of glucose levels to the insulin concentrations, the total
insulin concentrations (human and porcine insulin) in ng/l were presented in Figure 6.3
with the change of blood glucose level in mg/dl over 90-minute experiments. For the
rapid increase of insulin from 15 minutes to 30 minutes, the glucose level decreased at
higher rate. As the insulin concentration increased and stayed higher than the initial
value, the glucose level continuously decreased.
55
Figure 6.2: (a) The concentrations of human (, µU/l) and porcine (◊, ng/l) insulin were determined for 90-minutes periods. While porcine insulin concentrations stayed in therage of 12.23 - 12.38 ng/l, human insulin concentration increased by 0.5 and 23.7 µU/l after 15 30 minutes ultrasound exposure, respectively. (b) Human insulin concentrationswere determined in ng/l by unit conversion and compensated for the cross-reaction between porcine and human insulin for ELISA tests. The cross-reaction ratio was obtained at 13.3%.
0
10
20
30
40
50
60
70
0 15 30 45 60 75 90
Time (minutes)
12.0
12.2
12.4
12.6
12.8
13.0
Human insulinPorcine insulin
Hu
man
in
sulin
co
nce
ntr
atio
n (
µU
/l) P
orcin
e insu
lin co
ncen
tration
(ng
/l)
-2.0
0.0
2.0
4.0
6.0
8.0
10.0
12.0
14.0
16.0
1 2 3 4 5 6 7
Time (minutes)
Porcine
Human (uncompensated)
Human (compensated)
Insu
lin c
on
cen
trat
ion
(n
g/l)
(a)
(b)
56
6.2.3 Ultrasonic Glucose Sensing
Based on the concentration curve previously created for each biosensor, the
glucose concentrations of the interstitial fluid were determined using the current
measured by the biosensors. For each exposure condition, the results of ultrasonic
glucose sensing using the cymbal array and biosensor system were graphed and
compared to the results of a conventional glucose meter in Figure 6.4.
Figure 6.3: The total insulin concentrations in ng/L were presented with the change of blood glucose level in mg/dl over 90-minute experiments. As insulin concentrations increased faster, the glucose level decreased at higher rate.
-120
-100
-80
-60
-40
-20
0
0 15 30 45 60 75 90
Time (minutes)
0
2
4
6
8
10
12
14
16
18
GlucoseInsulinC
han
ge
of
glu
cose
(m
g/d
l) In
sulin
con
centratio
n (µ
U/l)
57
With an acoustic intensity of Isptp = 100 mW/cm2, the glucose level determined by
the biosensors after 20 minutes of ultrasound exposure was 175 ± 7 mg/dl while the level
measured by the conventional glucose meter was 166 ± 5 mg/dl. After 10 minutes of
ultrasound exposure, the glucose levels determined by the biosensor and glucose meter
were 152 ± 9 mg/dl and 130 ± 3 mg/dl, respectively. In a comparison to the glucose
level, 131 ± 5 mg/dl, measured by the conventional glucose meter, the glucose
concentration of the interstitial fluid was 127 ± 16 mg/dl which was determined by the
biosensors after 5 minutes of ultrasound exposure. The glucose levels of the interstitial
fluid after the ultrasound exposure with the intensity of Isptp = 50 mW/cm2 for 10 and 5
minutes were determined by the biosensors as 140 ± 10 mg/dl and 82 ± 15 mg/dl,
Figure 6.4: The glucose concentrations of the interstitial fluid determined using theultrasound system and a conventional glucose meter. Asterisks above the data bar wereused if there was no statistical difference between the ultrasonic glucose sensing method and the conventional method based on the ANOVA with the 0.05 or 0.01 level ofsignificance.
0
20
40
60
80
100
120
140
160
180
200
20 kHz, 20 min,100 mW/cm2
20 kHz, 10 min,100 mW/cm2
20 kHz, 5 min,100 mW/cm2
20 kHz,10 min,50mW/cm2
20 kHz, 5 min, 50mW/cm2
Glu
co
se
le
ve
l [m
g/d
l]Biosensor
Glucose meter
frequency: exposure time:
Isptp:
20 kHz10 min 100 mW/cm2
20 kHz5 min 100 mW/cm2
20 kHz 20 min 100 mW/cm2
20 kHz5 min 50 mW/cm2
20 kHz 10 min 50 mW/cm2
** ** ** **
*
58
respectively. Compared to the results of the biosensors, the glucose levels measured by
the conventional glucose meter after 10 and 5 minutes exposure were 124 ± 4 mg/dl and
125 ± 4 mg/dl, respectively.
In Figure 6.4, asterisks above the data bar were used if there was no statistical
difference between the ultrasonic glucose sensing method and the conventional method
based on the ANOVA with the 0.05 or 0.01 level of significance. For the intensity of Isptp
= 100 mW/cm2, the glucose concentrations determined by the ultrasound system were
statistically the same as the value determined by the glucose meter. In addition, the
results of the 10 minute ultrasound exposure at the intensity of Isptp = 50 mW/cm2 showed
no difference between two measurements. At the end of each experiment, the skin area
that was exposed to the ultrasound was examined for the skin reaction and no skin
irritation was found from all experiments.
6.2.4 Glucose control in large pigs using a closed-loop feedback controlled ultrasound system
For all four pigs designated as pig #1 – #4, the currents from the biosensors were
measured every 20 minutes for a 120-minute period. Based on the concentration curve
previously created for each biosensor, the glucose concentration of interstitial fluid was
determined. From the comparison of the glucose levels, the glucose concentrations
determined by the biosensors were up to 10 mg/dl higher than those measured by the
glucose meter (Figure 6.5).
Using the ultrasound and the biosensors, the initial glucose level of pig #1,
determined as 117 mg/dl, was lower than the reference level, 115 mg/dl (Figure 6.5a).
59
Therefore, the ultrasonic insulin delivery system was initially turned on for insulin
delivery. After 20 minutes of experiments, the glucose level was 106 mg/dl, and the
ultrasonic transdermal delivery system was turned off. Additional 20 minutes after the
experiments, the insulin delivery system was turned on again since the glucose level was
116 mg/dl and turned off again 80 minutes after the experiments because of lower
glucose concentration than the reference value. For pig #2, the insulin delivery system
was turned on at the second glucose measurement; 40 minutes after the experiment had
started (Figure 6.5b). By operating the insulin delivery system for 40 minutes, the
glucose level decreased to 104 mg/dl. During the period from 60 to 120 minutes
following the experiments, the glucose level was controlled within 104 – 116 mg/dl by
the automatically operated insulin delivery system. The initial glucose level of pig #3
was 137 mg/dl, and the insulin delivery system was automatically operated (Figure 6.5c).
After 80 minutes of transdermal insulin delivery from the beginning of the experiments,
the glucose level decreased to 107 mg/dl and, after the delivery system was turned off,
increased to 127 mg/dl. The glucose level of pig #4 was regulated in the range of 109 –
130 mg/dl by the ultrasonic glucose sensing and the insulin delivery system that was
automatically controlled by the closed-loop controller (Figure 6.5d). Without the
operation of the insulin delivery system, the glucose level increased by 9-10 mg/dl for 20
minutes(Park, Werner, & Smith 2007).
60
Figure 6.5: (a) – (d) Glucose concentrations of four pigs, which were regulated by the ultrasonic transdermal glucose sensing and insulin delivery systems combined by a feedback controller.
Case #1
80
90
100
110
120
130
140
0 20 40 60 80 100 120
Time (min)
Glu
cose
Lev
el (
mg
/dl)
BiosensorGlucose meter
on
Case #1
80
90
100
110
120
130
140
0 20 40 60 80 100 120
Time (min)
Glu
cose
Lev
el (
mg
/dl)
BiosensorGlucose meter
onon
Case #2
80
90
100
110
120
130
140
0 20 40 60 80 100 120
Time (min)
Glu
cose
Lev
el (
mg
/dl)
BiosensorGlucose meter
on
offoff
on
Case #2
80
90
100
110
120
130
140
0 20 40 60 80 100 120
Time (min)
Glu
cose
Lev
el (
mg
/dl)
BiosensorGlucose meter
onon
offoffoffoff
onon
(b)
(a)
61
Figure 6.5: (continued)
Case #3
80
90
100
110
120
130
140
0 20 40 60 80 100 120
Time (min)
Glu
cose
Lev
el (
mg
/dl)
BiosensorGlucose meter
on
offon
Case #3
80
90
100
110
120
130
140
0 20 40 60 80 100 120
Time (min)
Glu
cose
Lev
el (
mg
/dl)
BiosensorGlucose meter
onon
offoffon
Case #4
80
90
100
110
120
130
140
0 20 40 60 80 100 120
Time (min)
Glu
cose
Lev
el (
mg
/dl)
BiosensorGlucose meter
onon
off
off
on
Case #4
80
90
100
110
120
130
140
0 20 40 60 80 100 120
Time (min)
Glu
cose
Lev
el (
mg
/dl)
BiosensorGlucose meter
onononon
offoff
offoff
onon
(d)
(c)
Chapter 7
CONCLUSION
7.1 Transdermal Insulin Delivery
The goal of this task was to determine if a light-weight, low-profile ultrasound
device based cymbal transducer could be used for in vivo transdermal insulin in animals
which approximate the size and weight of a human. Although commercial sonicators
have been admirable devices for demonstrating drug delivery, the ultrasonic probe or
converter from a sonicator can weigh a kilogram or more; the cymbal array weighs less
than 38 grams. Individual cymbals can be arranged into multi-element array designs
since this can increase the effective aperture of ultrasound area with respect to skin area
and some research indicates that the delivery dose increases with ultrasound exposure
area (Smith, Lee, Maione, Roy, McElligott, & Shung 2003). Interestingly, the cymbal
design originates from underwater research for naval applications and current research is
underway to incorporate existing battery technology in the miniaturization of portable
power.
Generally, xylazine causes hyperglycemia in rats and rabbits (Harkness &
Wagner 1995; Hillyer & Quesenberry 1997; Kawai, Keep, & Betz 1997; Pavlovic et al.
1996). In pigs, xylazine suppresses insulin release which results in the higher initial
average blood glucose level of 146 mg/dl pigs (Heim et al. 2002). The mean value for
normal blood glucose levels for pigs is approximately 108.5 mg/dl yet (Danfaer 1999) the
63
use of the xylazine appears to have the continual effect of increasing the glucose level at
a rate of 20 mg/dl/hour even after the experiment started. This may indicate either the
lack of passive permeability of the pig's skin to the insulin or the sustained consequence
of the xylazine. Yet the useful effect of using these insulin-suppressed pigs was to
demonstrate the feasibility of reducing a high glucose level (≈150 mg/dl) to a normal
glucose level (below 100) albeit for a pig model. In contrast to the control group, the
ultrasound with insulin group showed a blood glucose level decrease of -91 mg/dl at 90
minutes compared to the baseline. Moreover, compared to the control at 90 minutes, the
difference between the two groups was approximately 120 mg/dl and the ANOVA
analysis indicated a statistical difference (p > 0.05) between the two groups at the time
points of 30 minutes and greater.
When fasting, glucose of 110-126 mg/dl is classified as impaired fasting glucose,
140-200 mg/dl is impaired glucose tolerance and greater than 200 mg/dl is considered
diabetic (Carnevale Schianca et al. 2003). In this case a diabetic person would need to
inject enough insulin to reduce his or her blood glucose by about 100 mg/dl.
Summarizing our previous ultrasound exposure results using hyperglycemic animals at
90 minutes, the glucose level continued to decrease to -296.7 ± 52.8 mg/dl in rats
and -208.1 ± 29 mg/dl in rabbits (Lee, Snyder, Newnham, & Smith 2004; Smith, Lee,
Maione, Roy, McElligott, & Shung 2003). Although the decrease in blood glucose in this
pig research was not as large as it was in the rats and rabbits, this may be due to the fact
that pigs have a far greater blood volume and body mass. Nevertheless the results
indicate that the array was capable of safely bringing the diabetic glucose level within the
normal range. In conclusion, the results herein demonstrate a promising pre-clinical
64
outcome for the low profile cymbal array to be used for ultrasound enhanced in vivo
insulin transport using an animal model which mimics human use.
Ultrasound has been used as an enhancer for transdermal drug delivery, especially
for transdermal insulin transport (Boucaud, Garrigue, Machet, Vaillant, & Patat 2002;
Pitt, Husseini, & Staples 2004; Prausnitz & Langer 2008; Smith 2007). However,
concerns have been expressed that the insulin secreted by the subject animals might
solely control the change of glucose. In order to address this essential issue in ultrasonic
transdermal insulin delivery, an other task was designed to investigate how much insulin
was transported through the skin and the effects of transdermally delivered insulin on the
glucose control especially using the cymbal transducer array as a practically portable
ultrasound device. The cross-reaction compensated results of insulin concentrations from
90-minute experiments demonstrated that insulin is transdermally delivered due to the
ultrasound exposure and indicates that the glucose levels are controlled by the effects of
both delivered (human insulin) and secreted insulin (porcine insulin). In addition, these
results confirm transdermal insulin delivery by ultrasound and supply essential
information for developing a wearable ultrasound device for transdermal drug transport.
7.2 Ultrasonic Glucose Sensing
In several studies, researchers have demonstrated that ultrasound can permeabilize
skin to enhance the transdermal delivery of drugs (Mitragotri 2005; Park, Werner, &
Smith 2007; Pitt, Husseini, & Staples 2004; Tachibana & Tachibana 2001). The results of
cavitation have been suggested as the main mechanism of ultrasound mediated
65
transdermal drug delivery although it has not been completely understood (Guzman,
McNamara, Nguyen, & Prausnitz 2003b; Guzman et al. 2003a; Mitragotri, Edwards,
Blankschtein, & Langer 1995; Schlicher, Radhakrishna, Tolentino, Apkarian, Zarnitsyn,
& Prausnitz 2006). In addition to the drug delivery, transdermal extraction of glucose in
the interstitial fluid by ultrasound has shown potential in becoming an alternative
technique of glucose measurement (Cantrell, McArthur, & Pishko 2000; Lee, Nayak,
Dodds, Pishko, & Smith 2005).
Based on the effects of ultrasound that enhances the skin permeability and the
mechanism of glucose diffusion due to the concentration differences, this research was
designed. Previously, in vivo experiments of transdermal glucose sensing on rats were
conducted using the light-weight cymbal transducer array and electrochemical biosensors
and the results have demonstrated the feasibility of using the cymbal array in ultrasonic
glucose measurement. Additionally, the results have shown a reliable accuracy of this
technique compared to the conventional method (Lee, Nayak, Dodds, Pishko, & Smith
2005).
In perspective to human application, the purpose of this research was to
demonstrate the feasibility of the cymbal array in ultrasonic glucose monitoring on large
pigs which have a similar size and weight to humans. As the initial study of in vivo pig
experiment, the results of the ultrasound exposure at the same exposure condition (Isptp =
100 mW/cm2, 20 minutes) used in a previous study, have shown that the ultrasound
system consisting of the cymbal array and biosensors was able to noninvasively
determine the glucose concentration on large animals having a similar size to humans. To
explore the reliability of the ultrasound system for glucose sensing, further experiments
66
have been performed with different ultrasound exposure conditions. From the statistical
analysis, the results of ultrasound exposure with Isptp = 100 mW/cm2 for 10 and 5 minutes
have shown a reliable accuracy compared with the results of conventional glucose meter
measurements.
The results presented herein have demonstrated the feasibility of using the cymbal
transducer array for ultrasonic glucose sensing on large animals for preclinical
applications. Furthermore, the results of ultrasonic glucose measurements with 100
mW/cm2 (Isptp) have indicated that the ultrasound system using the cymbal array has been
able to determine the glucose concentration in short exposure time.
7.3 Closed-Loop Feedback Controlled Ultrasound System for Glucose Regulation
The challenge with diabetes care is frequent patient non-compliance due to the
unpleasant continuous process of finger pricking for blood glucose samples and daily
insulin injections. Noncompliance typically involves refusing treatment, reducing or
increasing insulin doses, or skipping medications. These actions are common problems
that can prevent the efficacy of treatment and reduce patient health. Eventually, non-
compliance leads to complications such as nerve damage, retinal damage, microvascular
damage, chronic renal failure, kidney failure, dental disease, and cardiovascular
disease(The Diabetes Control and Complications Trial Research Group (DCCT) 1993;
Wickramasinghe, Yang, & Spencer 2004). Overcoming non-compliance requires a
feedback-controlled active insulin delivery system that is noninvasive and easy to use.
67
Current drug delivery technologies allow controlled dosing of a drug, but are
limited as they do not respond to actual biological status, meaning they lack a feedback
loop. To address this need, a paradigm shift is happening in diabetes care from the
current passive (one drug at a single dose over time) controlled- release to an active
delivery system that includes sensing and biofeedback. With it known that ultrasound
enhances transdermal drug delivery by permeabilizing skin to drugs(Mitragotri 2005;
Smith 2007; Smith 2008), the research is presented using ultrasound within an initial
study of a feedback-controlled active insulin delivery system with cymbal arrays that can
continuously monitor biological status and control insulin delivery in response to
changing glucose levels.
For the glucose sensing to occur, size was a consideration; a smaller sized array
was needed to minimize the dilution of glucose into the PBS solution, and a large enough
sized one was needed for ultrasound exposure to permeabilize the skin. To meet these
two requirements, the 2 × 2 cymbal array was used for the glucose sensing. Glucose
levels in previous in vivo experiments with rats and rabbits were determined by using 2 ×
2 arrays. The errors between the measurements by conventional glucose meters and the
ultrasonic method were -16% and -3% for rats and rabbits, respectively. In the studies of
using the cymbal array for the transdermal insulin delivery, a broader acoustic intensity
field without increasing the spatial-peak temporal-peak intensity (Isptp) was demonstrated
that performed glucose reduction better(Luis et al. 2007). Based on this result, the 3 × 3
cymbal array was used in the transdermal insulin delivery to expose a larger skin area to
ultrasound. Using the closed-loop feedback algorithm, the systems were combined and
68
evaluated in in vivo experiments on four pigs. The glucose levels of each pig were
controlled for 120 minutes with a maximum error of 9%.
To develop an active ultrasound system that tightly controls glucose levels and
one that is practically portable so that patients can wear it on a daily basis, tone must
evaluate the portable design of transducers and the closed-loop feedback control system.
In addition to the design of cymbal transducer arrays that have previously shown
potential as portable ultrasound systems, the results presented herein demonstrated the
feasibility of the ultrasound systems using cymbal arrays for controlling glucose levels.
7.4 Future Work
The future work of this research includes (1) modifying the feedback controller
with a complex control algorithm that counts the changing rate of glucose concentrations,
(2) integrating the relationship between the ultrasound exposure time and the duration of
enhanced skin permeability into the controller, and (3) improving the biosensor design to
extend its lifetime. A simple proportional feedback controller was used in this study to
evaluate the feasibility of the cymbal array system to control glucose concentrations.
While the simple controller showed potential to close the ultrasonic glucose sensing and
insulin delivery systems, the controller responded only to the difference of glucose levels
from a reference value. In order to control rapid increment of glucose concentrations
such as post-meal glucose levels, a modification of the simple controller that counts the
changing ratio of glucose concentrations is necessary.
69
Another consideration to improve the controller is integrating the relationship
between the insulin delivery dose and the ultrasound exposure time. In previous studies
of transdermal drug delivery, 15 minutes ultrasound exposure at 20 kHz with an intensity
of 2.5 - 5 W/cm2 showed enhanced skin permeability for 2.5 – 3 hours. Also, 5 minutes
ultrasound exposure at 6.5 W/cm2 resulted the skin permeable to insulin for 4.5
hours(Mitragotri, Blankschtein, & Langer 1995a). At lower intensities, 30 – 60 minutes
ultrasound exposure with an intensity of 125 – 225 mW/cm2 permeabilized the skin for
1.5 – 4 hours(Mitragotri, Blankschtein, & Langer 1995b; Prausnitz 1997). While these
studies demonstrated that ultrasound enhanced the skin permeability for several hours,
comprehensive studies of the relationship between ultrasound exposure time, the duration
of enhanced permeability, and the dose of delivered insulin have not been explored.
Therefore, quantitative studies of the insulin delivery dose as a function of ultrasound
exposure time is necessary for the modification of the controller to automatically
determine the operating time of the arrays in the system.
For more accurate and convenient glucose sensing, the design of biosensors
should be modified to stabilize the hydrogel as well as to extend the sensors lifetime of
sensors. The biosensors used in this study have several advantages that they are easy to
prepare, economical, and disposable. Three days after preparing the biosensors with
hydrogel, the biosensor could sense glucose concentration with a maximum error of 3%.
Although the biosensors had stable sensitivities, the lifetime of the biosensors was limited
due to the lack of hydrogel adhesion. The hydrogel capturing GOx can be easily
detached from the screen printed electrodes by bending motions. Therefore, technical
improvements on the sensor design are required to over come the limited lifetime. As the
70
next step to an active ultrasound system, integration of the modifications in the
ultrasound system will promote the capability of a closed-loop feedback controlled
ultrasound system in favor of an active glucose control in diabetes care.
BIBLIOGRAPHY
AIUM 1993, Bioeffects and safety of diagnostic ultrasound American Institute of Ultrasound in Medicine, Rockville, MD.
American Diabetes Association 2007, 2007 Annual Report: Individual in Impact, Broad in Reach.
Arora, A., Hakim, I., Baxter, J., Rathnasingham, R., Srinivasan, R., Fletcher, D. A., & Mitragotri, S. 2007, "Needle-free delivery of macromolecules across the skin by nanoliter-volume pulsed microjets", Proc.Natl.Acad.Sci.U.S.A, vol. 104, no. 11, pp. 4255-4260.
Ballerstadt, R., Polak, A., Beuhler, A., & Frye, J. 2004, "In vitro long-term performance study of a near-infrared fluorescence affinity sensor for glucose monitoring", Biosensors & Bioelectronics, vol. 19, no. 8, pp. 905-914.
Beaser, R. S. & Staff of Joslin Diabetes Center 2007, Joslin's Diabetes Deskbook, 2nd edn, Joslin Diabetes Center, Boston, MA.
Berson, S. A. & Yalow, R. S. 1963, "Antigens in Insulin Determinants of Specificity of Porcine Insulin in Man", Science, vol. 139, no. 3557, pp. 844-845.
Bessman, S. P. & Schultz, R. D. 1972, "Sugar electrode sensor for the "artificial pancreas"", Horm.Metab Res., vol. 4, no. 6, pp. 413-417.
Blake, F. G. 1949, The onset of cavitation in liquids. I. Cavitation threshold sound pressures in water as a function of temperature and hydrostatic pressure, Dept. of Engineering, Sciences and Applied Physics. Harvard University, Cambridge, MA, Tech. Mem. 12.
Boucaud, A., Garrigue, M. A., Machet, L., Vaillant, L., & Patat, F. 2002, "Effect of sonication parameters on transdermal delivery of insulin to hairless rats", J.Control Release, vol. 81, no. 1-2, pp. 113-119.
Boucaud, A., Tessier, L., Machet, L., Vaillant, L., & Patat, F. "Transdermal Delivery of Insulin using Low Frequency Ultrasound", in IEEE Ultrasonics Symposium, pp. 1453-1456.
Caduff, A., Hirt, E., Feldman, Y., Ali, Z., & Heinemann, L. 2003, "First human experiments with a novel non-invasive, non-optical continuous glucose monitoring system", Biosensors & Bioelectronics, vol. 19, no. 3, pp. 209-217.
Cantrell, J. T., McArthur, M. J., & Pishko, M. V. 2000, "Transdermal Extraction of Interstitial Fluid by Low-Frequency Ultrasound Quantified with 3H2O as a Tracer Molecule", Journal of Pharmaceutical Sciences, vol. 89, no. 9, pp. 1170-1179.
72
Carnevale Schianca, G. P., Rossi, A., Sainaghi, P. P., Maduli, E., & Bartoli, E. 2003, "The significance of impaired fasting glucose versus impaired glucose tolerance: importance of insulin secretion and resistance", Diabetes Care, vol. 26, no. 5, pp. 1333-1337.
Carstensen, E. L. & Flynn, H. G. 1982, "The potential for transient cavitation with microsecond pulses of ultrasound", Ultrasound Med.Biol., vol. 8, no. 6, p. L720-L724.
Chen, H., Zhu, H., Zheng, J., Mou, D., Wan, J., Zhang, J., Shi, T., Zhao, Y., Xu, H., & Yang, X. 2009, "Iontophoresis-driven penetration of nanovesicles through microneedle-induced skin microchannels for enhancing transdermal delivery of insulin", J.Control Release, vol. 139, no. 1, pp. 63-72.
Chen, Y., Shen, Y., Guo, X., Zhang, C., Yang, W., Ma, M., Liu, S., Zhang, M., & Wen, L. P. 2006, "Transdermal protein delivery by a coadministered peptide identified via phage display", Nat.Biotechnol., vol. 24, no. 4, pp. 455-460.
Cohen, N. D. & Shaw, J. E. 2007, "Diabetes: advances in treatment", Intern.Med.J., vol. 37, no. 6, pp. 383-388.
Connolly, P., Cotton, C., & Morin, F. 2002, "Opportunities at the skin interface for continuous patient monitoring: a reverse iontophoresis model tested on lactate and glucose", IEEE Trans.Nanobioscience., vol. 1, no. 1, pp. 37-41.
Cox, M. 2009, "An overview of continuous glucose monitoring systems", J.Pediatr.Health Care, vol. 23, no. 5, pp. 344-347.
Danfaer, A. 1999, "Carbohydrate and Lipid Metabolism," in A Quantitative Biology of the Pig, I. Kyriazakis, ed., CABI Publishing, New York, NY, pp. 333-362.
El-Kamel, A. H., Al-Fagih, I. M., & Alsarra, I. A. 2008, "Effect of sonophoresis and chemical enhancers on testosterone transdermal delivery from solid lipid microparticles: an in vitro study", Curr.Drug Deliv., vol. 5, no. 1, pp. 20-26.
Felts, P. W. 1973, "Pancreas transplantation and the artificial pancreas: status of current experimental approaches", South.Med.J., vol. 66, no. 1, pp. 66-73.
Garg, S., Zisser, H., Schwartz, S., Bailey, T., Kaplan, R., Ellis, S., & Jovanovic, L. 2006, "Improvement in glycemic excursions with a transcutaneous, real-time continuous glucose sensor", Diabetes Care, vol. 29, no. 1, pp. 44-50.
Guzman, H. R., McNamara, A. J., Nguyen, D. X., & Prausnitz, M. R. 2003b, "Bioeffects caused by changes in acoustic cavitation bubble density and cell concentration: a unified explanation based on cell-to-bubble ratio and blast radius", Ultrasound Med.Biol., vol. 29, no. 8, pp. 1211-1222.
73
Guzman, H. R., McNamara, A. J., Nguyen, D. X., & Prausnitz, M. R. 2003a, "Bioeffects caused by changes in acoustic cavitation bubble density and cell concentration: a unified explanation based on cell-to-bubble ratio and blast radius", Ultrasound Med.Biol., vol. 29, no. 8, pp. 1211-1222.
Hanaire, H., Dubet, A., Chauveau, M. E., Anduze, Y., Fernandes, M., Melki, V., & Ritz, P. 2009, "Usefulness of Continuous Glucose Monitoring for the Diagnosis of Hypoglycemia after a Gastric Bypass in a Patient Previously Treated for Type 2 Diabetes", Obes.Surg.
Harkness, J. E. & Wagner, D. J. 1995, The Biology and Medicine of Rabbits and Rodents Williams and Willkins, Baltimore MD.
Heim, K. E., Morrell, J. S., Ronan, A. M., & Tagliaferro, A. R. 2002, "Effects of ketamine-xylazine and isoflurane on insulin sensitivity in dehydroepiandrosterone sulfate-treated minipigs (Sus scrofa domestica)", Comp Med., vol. 52, no. 3, pp. 233-237.
Heinemann, L. 2003, "Continuous glucose monitoring by means of the microdialysis technique: underlying fundamental aspects", Diabetes Technol.Ther., vol. 5, no. 4, pp. 545-561.
Hillyer, E. & Quesenberry, K. E. 1997, Ferrets, Rabbits, and Rodents: Clinical Medicine and Surgery W.B. Saunders Co, Philadelphia PA.
Hodgson, B. B. & Robert J. 2006, Saunders Nursing Drug Handbook 2006 ELSEVIER SAUNDERS, St. Louis, Missouri.
Jeitler, K., Horvath, K., Berghold, A., Gratzer, T. W., Neeser, K., Pieber, T. R., & Siebenhofer, A. 2008, "Continuous subcutaneous insulin infusion versus multiple daily insulin injections in patients with diabetes mellitus: systematic review and meta-analysis", Diabetologia, vol. 51, no. 6, pp. 941-951.
Kawai, N., Keep, R. F., & Betz, A. L. 1997, "Hyperglycemia and the Vascular Effects of Cerebral Ischemia", Stroke, vol. 28, pp. 149-154.
Kinsler, L. E., Frey, A. R., Coppens, A. B., & Sanders, J. V. 1982, Fundamentals of Acoustics, 3rd edn, John Wiley & Sons, New York.
Kost, J. 2002, "Ultrasound-Assisted Insulin Delivery and Noninvasive Glucose Sensing", Diabetes Technology & Therapeutics, vol. 4, no. 4, pp. 489-497.
Kowalski, A. J. 2009, "Can we really close the loop and how soon? Accelerating the availability of an artificial pancreas: a roadmap to better diabetes outcomes", Diabetes Technol.Ther., vol. 11 Suppl 1, p. S113-S119.
74
Kumareswaran, K., Evans, M. L., & Hovorka, R. 2009, "Artificial pancreas: an emerging approach to treat Type 1 diabetes", Expert.Rev.Med.Devices, vol. 6, no. 4, pp. 401-410.
Lance, L. L., Lacy, C. F., Goldman, M. P., & Armstrong, L. L. 2002, Quick Look Drug Book Lippincott Williams & Wilkins, Philadelphia, PA.
Lee, S., Nayak, V., Dodds, J., Pishko, M., & Smith, N. B. 2005, "Glucose measurements with sensors and ultrasound", Ultrasound in Medicine and Biology, vol. 31, no. 7, pp. 971-977.
Lee, S., Newnham, R. E., & Smith, N. B. 2004, "Short ultrasound exposure times for noninvasive insulin delivery in rats using the lightweight cymbal array", IEEE Trans.Ultrason.Ferroelectr.Freq.Control, vol. 51, no. 2, pp. 176-180.
Lee, S., Snyder, B., Newnham, R. E., & Smith, N. B. 2004, "Noninvasive ultrasonic transdermal insulin delivery in rabbits using the light-weight cymbal array", Diabetes Technol.Ther., vol. 6, no. 6, pp. 808-815.
Leighton, T. G. 1994, The Acoustic Bubble Academic Press, San Diego, CA.
Li, Y. Z., Quan, Y. S., Zang, L., Jin, M. N., Kamiyama, F., Katsumi, H., Yamamoto, A., & Tsutsumi, S. 2008, "Transdermal delivery of insulin using trypsin as a biochemical enhancer", Biol.Pharm.Bull., vol. 31, no. 8, pp. 1574-1579.
Luis, J., Park, E. J., Meyer, R. J., & Smith, N. B. 2007, "Rectangular cymbal arrays for improved ultrasonic transdermal insulin delivery", J.Acoust.Soc.Am., vol. 122, no. 4, pp. 2022-2030.
Maione, E., Shung, K. K., Meyer, R. J., Hughes, J. W., Newnham, R. E., & Smith, N. B. 2002, "Transducer design for a portable ultrasound enhanced transdermal drug delivery system", IEEE Transactions on Ultrasonics, Ferroelectrics and Frequency Control, vol. 49, pp. 1430-1436.
Marchetti, G., Barolo, M., Jovanovic, L., Zisser, H., & Seborg, D. E. 2008, "An improved PID switching control strategy for type 1 diabetes", IEEE Trans.Biomed.Eng, vol. 55, no. 3, pp. 857-865.
Martanto, W., Davis, S. P., Holiday, N. R., Wang, J., Gill, H. S., & Prausnitz, M. R. 2004, "Transdermal delivery of insulin using microneedles in vivo", Pharm.Res., vol. 21, no. 6, pp. 947-952.
Mastrototaro, J. 1999, "The MiniMed Continuous Glucose Monitoring System (CGMS)", J.Pediatr.Endocrinol.Metab, vol. 12 Suppl 3, pp. 751-758.
75
Meeuwisse-Pasterkamp, S. H., van der Klauw, M. M., & Wolffenbuttel, B. H. 2008, "Type 2 diabetes mellitus: prevention of macrovascular complications", Expert Rev.Cardiovasc.Ther., vol. 6, no. 3, pp. 323-341.
Minnaert, M. 1933, "On musical air-bubbles and the sounds of running water", Phil.Mag., vol. 16, pp. 235-248.
Mitragotri, S. 2005, "Healing sound: the use of ultrasound in drug delivery and other therapeutic applications", Nat.Rev.Drug Discov., vol. 4, no. 3, pp. 255-260.
Mitragotri, S., Blankschtein, D., & Langer, R. 1995a, "Ultrasound-Mediated Transdermal Protein Delivery", Science, vol. 269, no. 5225, pp. 850-853.
Mitragotri, S., Blankschtein, D., & Langer, R. 1995b, "Ultrasound-Mediated Transdermal Protein Delivery", Science, vol. 269, no. 5225, pp. 850-853.
Mitragotri, S., Blankschtein, D., & Langer, R. 1995c, "Ultrasound-Mediated Transdermal Protein Delivery", Science, vol. 269, no. 5225, pp. 850-853.
Mitragotri, S., Blankschtein, D., & Langer, R. 1997a, "An explanation for the variation of the sonophoretic transdermal transport enhancement from drug to drug", J.Pharm.Sci., vol. 86, no. 10, pp. 1190-1192.
Mitragotri, S., Blankschtein, D., & Langer, R. 1997b, "An explanation for the variation of the sonophoretic transdermal transport enhancement from drug to drug", J.Pharm.Sci., vol. 86, no. 10, pp. 1190-1192.
Mitragotri, S., Coleman, M., Kost, J., & Langer, R. 2000, "Analysis of ultrasonically extracted interstitial fluid as a predictor of blood glucose levels", Journal of Applied Physiology, vol. 89, pp. 961-966.
Mitragotri, S., Edwards, D. A., Blankschtein, D., & Langer, R. 1995, "A mechanistic study of ultrasonically-enhanced transdermal drug delivery", J.Pharm.Sci., vol. 84, no. 6, pp. 697-706.
Mitragotri, S. & Kost, J. 2004, "Low-frequency sonophoresis: a review", Adv.Drug Deliv.Rev., vol. 56, no. 5, pp. 589-601.
Montorsi, F., Salonia, A., Guazzoni, G., Barbieri, L., Colombo, R., Brausi, M., Scattoni, V., Rigatti, P., & Pizzini, G. 2000, "Transdermal electromotive multi-drug administration for Peyronie's disease: preliminary results", J.Androl, vol. 21, no. 1, pp. 85-90.
Murthy, S. N., Zhao, Y. L., Marlan, K., Hui, S. W., Kazim, A. L., & Sen, A. 2006, "Lipid and electroosmosis enhanced transdermal delivery of insulin by electroporation", J.Pharm.Sci., vol. 95, no. 9, pp. 2041-2050.
76
Neppiras, E. A. 1980a, "Acoustic Cavitation", Physics Reports-Review Section of Physics Letters, vol. 61, no. 3, pp. 159-251.
Neppiras, E. A. 1980b, "Acoustic Cavitation Thresholds and Cyclic Processes", Ultrasonics, vol. 18, no. 5, pp. 201-209.
Newnham, R. E. & Dogan, A. 1995, Metal-electroactive ceramic composite transducer, US Patent 5729077 (patent).
Newnham, R. E., Xu, Q. C., & Yoshikawa, S. 1991, Transformed stress direction acoustic transducer, 4,999,819 (patent).
Newnham, R. E., Xu, Q. C., & Yoshikawa, S. 1994, Metal-electroactive ceramic composite actuators, 5,276,657 (patent).
Newnham, R. E. & Zhang, J. 1999, Flextensitional metal-ceramic composite transducer, US Patent 6232702 (patent).
Nordquist, L., Roxhed, N., Griss, P., & Stemme, G. 2007, "Novel microneedle patches for active insulin delivery are efficient in maintaining glycaemic control: an initial comparison with subcutaneous administration", Pharm.Res., vol. 24, no. 7, pp. 1381-1388.
Park, E. J., Werner, J., Beebe, J., Chan, S., & Smith, N. B. 2009, "Noninvasive Ultrasonic Glucose Sensing with Large Pigs (~200 Pounds) Using a Lightweight Cymbal Transducer Array and Biosensors", Journal of Diabetes Science and Technology, vol. 3, no. 3, pp. 517-523.
Park, E. J., Werner, J., & Smith, N. B. 2007, "Ultrasound mediated transdermal insulin delivery in pigs using a lightweight transducer", Pharmaceutical Research, vol. 24, no. 7, pp. 1396-1401.
Patek, S. D., Breton, M. D., Chen, Y., Solomon, C., & Kovatchev, B. 2007, "Linear Quadratic Gaussian-Based Closed-Loop Control of Type 1 Diabetes", J.Diabetes Sci.Technol., vol. 1, no. 6, pp. 834-841.
Pavlovic, M., Wroblewski, K., Manevich, Y., Kim, S., & Biaglow, J. E. 1996, "The importance of choice of anaesthetics in studying radiation effects in the 9L rat glioma", Br.J.Cancer Suppl, vol. 27, p. S222-S225.
Pickup, J. C. & Renard, E. 2008, "Long-acting insulin analogs versus insulin pump therapy for the treatment of type 1 and type 2 diabetes", Diabetes Care, vol. 31 Suppl 2, p. S140-S145.
Pitt, W. G., Husseini, G. A., & Staples, B. J. 2004, "Ultrasonic drug delivery--a general review", Expert.Opin.Drug Deliv., vol. 1, no. 1, pp. 37-56.
77
Pond, W. G. & Houpt, K. A. 1978, The Biology of the Pig Cornell University Press, London, UK.
Prausnitz, M. R. 1997, "Reversible skin permeabilization for transdermal delivery of macromolecules", Critical Reviews in Therapeutic Drug Carrier Systems, vol. 14, no. 4, pp. 455-483.
Prausnitz, M. R. 2004, "Microneedles for transdermal drug delivery", Adv.Drug Deliv.Rev., vol. 56, no. 5, pp. 581-587.
Prausnitz, M. R. & Langer, R. 2008, "Transdermal drug delivery", Nat.Biotechnol., vol. 26, no. 11, pp. 1261-1268.
Renard, E. 2008, "Clinical experience with an implanted closed-loop insulin delivery system", Arq Bras.Endocrinol.Metabol., vol. 52, no. 2, pp. 349-354.
Rifkin, H. & Porte, D. 1990, Ellenberg and Rifkin's Diabetes, 4th edn, Elsevier Science, New York, NY.
Ryan, E. A. & Germsheid, J. 2009, "Use of continuous glucose monitoring system in the management of severe hypoglycemia", Diabetes Technol.Ther., vol. 11, no. 10, pp. 635-639.
Schlicher, R. K., Radhakrishna, H., Tolentino, T. P., Apkarian, R. P., Zarnitsyn, V., & Prausnitz, M. R. 2006, "Mechanism of intracellular delivery by acoustic cavitation", Ultrasound Med.Biol., vol. 32, no. 6, pp. 915-924.
Sekkat, N., Naik, A., Kalia, Y. N., Glikfeld, P., & Guy, R. H. 2002, "Reverse iontophoretic monitoring in premature neonates: feasibility and potential", Journal of Controlled Release, vol. 81, no. 1-2, pp. 83-89.
Sershen, S. & West, J. 2002, "Implantable, polymeric systems for modulated drug delivery", Adv.Drug Deliv.Rev., vol. 54, no. 9, pp. 1225-1235.
Shaw, J. E., Zimmet, P. Z., de Court, M., Dowse, G. K., Chitson, P., Gareeboo, H., Hemraj, F., Fareed, D., Tuomilehto, J., & Alberti, K. G. 1999, "Impaired fasting glucose or impaired glucose tolerance. What best predicts future diabetes in Mauritius?", Diabetes Care, vol. 22, no. 3, pp. 399-402.
Sintov, A. C. & Wormser, U. 2007, "Topical iodine facilitates transdermal delivery of insulin", J.Control Release, vol. 118, no. 2, pp. 185-188.
Skyler, J. S. 2009, "Continuous glucose monitoring: an overview of its development", Diabetes Technol.Ther., vol. 11 Suppl 1, pp. S5-10.
Smith, N. B. 2007, "Perspectives on transdermal ultrasound mediated drug delivery", Int.J.Nanomedicine, vol. 2, no. 4, pp. 585-594.
78
Smith, N. B. 2008, "Applications of ultrasonic skin permeation in transdermal drug delivery", Expert Opin.Drug Deliv., vol. 5, no. 10, pp. 1107-1120.
Smith, N. B., Lee, S., Maione, E., Roy, R. B., McElligott, S., & Shung, K. K. 2003, "Ultrasound mediated transdermal transport of insulin through in vitro human skin using novel transducer designs", Ultrasound in Medicine and Biology, vol. 29, no. 2, pp. 311-317.
Suetsugu, M., Takebayashi, K., & Aso, Y. 2007, "Association between diabetic microangiopathy and vascular endothelial function evaluated by flow-mediated vasodilatation in patients with type 2 diabetes", Int.J.Clin.Pract., vol. 61, no. 6, pp. 920-926.
Tachibana, K. 1992b, "Transdermal delivery of insulin to alloxan-diabetic rabbits by ultrasound exposure", Pharm.Res., vol. 9, no. 7, pp. 952-954.
Tachibana, K. 1992a, "Transdermal delivery of insulin to alloxan-diabetic rabbits by ultrasound exposure", Pharm.Res., vol. 9, no. 7, pp. 952-954.
Tachibana, K. & Tachibana, S. 1991, "Transdermal delivery of insulin by ultrasonic vibration", J.Pharm.Pharmacol., vol. 43, no. 4, pp. 270-271.
Tachibana, K. & Tachibana, S. 2001, "The use of ultrasound for drug delivery", Echocardiography., vol. 18, no. 4, pp. 323-328.
Tezel, A., Sens, A., & Mitragotri, S. 2002, "Incorporation of lipophilic pathways into the porous pathway model for describing skin permeabilization during low-frequency sonophoresis", Journal of Controlled Release, vol. 83, no. 1, pp. 183-188.
The Diabetes Control and Complications Trial Research Group (DCCT) 1993, "The Effect of Intensive Treatment of Diabetes on the Development and Progression of Long-Term Complications in Insulin-Dependent Diabetes-Mellitus", N Engl J Med, vol. 329, no. 14, pp. 977-986.
Tressler, J. F. 1997, Capped Ceramic Underwater Sound Projector The "Cymbal", Ph. D., The Pennsylvania State University.
Volpe, L., Pancani, F., Aragona, M., Lencioni, C., Battini, L., Ghio, A., Resi, V., Bertolotto, A., Del, P. S., & Di, C. G. 2009, "Continuous subcutaneous insulin infusion and multiple dose insulin injections in Type 1 diabetic pregnant women: a case-control study", Gynecol.Endocrinol. pp. 1-4.
Weinstein, R. L., Bugler, J. R., Schwartz, S. L., Peyser, T. A., Brazg, R. L., & McGarraugh, G. V. 2007, "Accuracy of the 5-day freestyle navigator continuous glucose monitoring system - Comparison with frequent laboratory reference measurements", Diabetes Care, vol. 30, no. 5, pp. 1125-1130.
79
Weinzimer, S. A., Steil, G. M., Swan, K. L., Dziura, J., Kurtz, N., & Tamborlane, W. V. 2008, "Fully automated closed-loop insulin delivery versus semiautomated hybrid control in pediatric patients with type 1 diabetes using an artificial pancreas", Diabetes Care, vol. 31, no. 5, pp. 934-939.
Weissberg-Benchell, J., ntisdel-Lomaglio, J., & Seshadri, R. 2003, "Insulin pump therapy: a meta-analysis", Diabetes Care, vol. 26, no. 4, pp. 1079-1087.
Whelan, J. 2002, "Electroporation and ultrasound for gene and drug delivery", Drug Discovery Today, vol. 7, no. 11, pp. 585-586.
Wickramasinghe, Y., Yang, Y., & Spencer, S. A. 2004, "Current Problems and Potential Techniques in In Vivo Glucose Monitoring", Journal of Fluorescence, vol. 14, no. 5, pp. 513-520.
Wilkins, E. & Atanasov, P. 1996, "Glucose monitoring: State of the art and future possibilities", Medical Engineering & Physics, vol. 18, no. 4, pp. 273-288.
Wong, L. J., Buckingham, B. A., Kunselman, B., Istoc, E., Leach, J., & Purvis, R. 2006, "Extended use of a new continuous glucose monitoring system with wireless data transmission in children with type 1 diabetes mellitus", Diabetes Technol.Ther., vol. 8, no. 2, pp. 139-145.
Zhang, I., Shung, K. K., & Edwards, D. A. 1996, "Hydrogels with enhanced mass transfer for transdermal drug delivery", J.Pharm.Sci., vol. 85, no. 12, pp. 1312-1316.
Appendix A
FUNDAMENTAL RESONANCE FREQUENCY OF CYMBAL TRANSDUCER – FEA
A.1 In Air
Table A.1: Comparison of the in-air resonance frequency of standard size brass-capped cymbals for various PZT types.
PZT type rf (kHz)
PZT-552 23.5 PZT-5A 23.6 PZT-4 24.0 PZT-8 24.1
Table A.2: Comparison of the in-air resonance frequency of standard size cymbals forvarious endcap materials.
Cap material rf (kHz)
Brass 23.5 Titanium 34.8
Kovar 26.7 Steel 32.3
Molybdenum 34.0
81
A.2 In Water
Table A.3: Comparison of the in-air resonance frequency of standard size brass-capped cymbals for different endcap thickness.
Cap thickness (mm) rf (kHz)
0.12 19.7 0.20 22.1 0.25 23.5 0.3 25.0 0.38 27.7 0.5 31.9
Table A.4: Comparison of the in-air resonance frequency of standard size brass-capped cymbals for different cavity depths.
Cavity depth (mm) rf (kHz)
0.12 16.7 0.18 18.4 0.25 20.8 0.32 23.4 0.47 29.2
Table A.5: Comparison of the in-water resonance frequency of standard size brass-capped cymbals for various PZT types.
PZT type rf (kHz)
PZT-552 16.5 PZT-5A 16.1 PZT-4 16.3 PZT-8 16.5
82
Table A.6: Comparison of the in-water resonance frequency of standard size cymbals forvarious endcap materials.
Cap material rf (kHz)
Brass 16.5 Titanium 20.0
Kovar 18.6 Steel 22.0
Molybdenum 24.7
Table A.7: Comparison of the in-water resonance frequency of standard size brass-capped cymbals for different endcap thickness.
Cap thickness (mm) rf (kHz)
0.12 10.9 0.20 14.5 0.25 16.5 0.3 18.3 0.38 21.3 0.5 25.8
Table A.8: Comparison of the in-water resonance frequency of standard size brass-capped cymbals for different cavity depths.
Cavity depth (mm) rf (kHz)
0.12 11.5 0.18 12.8 0.25 14.5 0.32 16.5 0.47 20.6
Appendix B
DETAIL INSTRUCTION OF CYMBAL ARRAY CONSTRUCTION
B.1 Metal Endcaps
1. Punch 0.5" diameter caps from brass sheet (see table below for thickness; generally 0.2-0.7mm) with Precision Brand Punch and Die Set.
2. Set cavity depth with shaping die. Use specific pressure (60kpsi) because excess pressure will flatten the die.
Array dimensions
Thickness (mm) Depth (mm)
1x1 0.12 0.25 2x2 0.25 0.25 3x3 0.25 0.32
84
85
B.2 Cymbals
1. Rub off the white powder from the 1mm thick, 0.5" diameter piezoelectric material (PZT) with ≥ 400 grain sand paper. The dot marks positive polarity and must be redrawn after powder is removed. Gently rub in a circular motion; do not “shine” the PZT, it will damage the electrode.
2. Weigh 3:1:1 2741LV epoxy to 15LV epoxy to silver powder. Mix thoroughly. Apply a thin strip to the edges of the PZT disc with a toothpick.
3. Carefully place the endcap onto the PZT disk and align by touch. Redraw the dot to mark polarity. Repeat with other side and align again. Wipe off excess epoxy with silver powder from the sides with a Kim-wipe. Clip sides and allow curing for 24hrs at room temperature. Leave excess epoxy as indicator. (Clipping the center will change the cavity depth.)
B.3 Polyurethane Base
1. Make a sheet of polyurethane after degassing. Spread out and remove air bubbles. Place spacers and top plate. For larger arrays, prepare thicker sheets. Cut squares (depends on molding; distances between are arbitrary but symmetry is preferred; ~37mm
86
x 37mm for 2x2) for the arrays. (Leaving too large of a gap between Cymbal transducers will cause the 3d Gaussian output to be distorted. After using polyurethane, flush with nitrogen gas before sealing.) 2. Punch the appropriate amount of 0.5" diameter holes in the squares. (The polyurethane bases are necessary for alignment and spacing, and to ensure that cymbals will be parallel to transducer surface.)
B.4 Connections
1. For 1 × 1 pattern, bare portion of red wires (arbitrary, but ~4.5ft) are connected directly to positive side of the cymbal and black, to the negative side. For 2 × 2 arrays, single-strand wires are used to connect the cymbals together, and the red/black wires are soldered onto the intersection.
2. Mix the silver epoxy, a conducting adhesive. The mix ratio is 1:0.08 solid to liquid. The mixture may seem thick but is acceptable.
3. Use a toothpick and the least amount possible to connect wires to cymbals. Use as little tape as possible. Again, keep excess epoxy as an indicator. Allow 8hrs curing time for silver epoxy.
87
B.5 Casting
1. Before casting, test connections with a multimeter.
2. Mix the two-part polyurethane in a 10:1 Part A to Part B ratio. Mix and degas in the vacuum. Polyurethane will rise when bubbles are drawn out, therefore, use a large container.
88
3. Spray the mold with Ease Release. Slowly pour polyurethane into mold to avoid capturing air bubbles, filling only up to 1mm. Remove air bubbles with a toothpick. 4. Slowly place the base with the cymbals into the mold (red/positive side down), one side first to prevent air bubble formation.
5. Use a toothpick to ensure that the base with the cymbals is parallel to the bottom surface.
89
6. Fill the rest of the mold with polyurethane.
(For 1 × 1 pattern, fill the entire mold with polyurethane before adding the base with the cymbal transducer. Then, slowly submerge the base. Also, when connecting the wires to the cymbals, make sure there is symmetry so that the weight is not all on one side. This will cause the base to rotate in the polyurethane. Shaping the base to fit tightly to the mold will also help prevent rotation.)
7. Tape the top and side of the groove where the wires come out for support. Allow polyurethane to cure for 24hrs. Leave a small amount of polyurethane as an indicator. (If the atmosphere is especially dry, the polyurethane has a tendency to dry with curve. To avoid this, cover with a towel.)
90
8. To remove transducer from molding, gently bend the mold. Excessive pressure may damage transducer or disrupt connections.
Appendix C
A COMPARISON BETWEEN INTRAPERITONEAL INJECTION AND TRANSDERMAL DELIVERY
One of the many questions with ultrasound mediated transdermal insulin delivery
is the relationship between the levels of glucose decrease from ultrasonic method versus a
direct subcutaneous injection of insulin. Therefore the blood glucose response from
direct injections of insulin against the transdermal delivery using the cymbal array was
examined.
A total of 16 experiments using eight Sprague-Dawley rats (350 - 550 g) were
performed in four experimental groups with four rats in each group: one ultrasonic
transdermal delivery and three subcutaneous injection groups. Each rat was used for
multiple experiments. Rats were anesthetized with a combination of ketamine
hydrochloride (60 mg/kg intramuscularly, Ketaject®, Phoenix, St. Joseph, MO) and
xylazine hydrochloride (10 mg/kg intramuscularly, Xyla-Ject®, Phoenix, St. Joseph,
MO). In addition to its role in general anesthesia, xylazine was used to induce a
temporary, but sustained (up to 12 hrs), hyperglycemia in rats.
For the ultrasonic transdermal delivery, the abdominal area of the rat was shaved
using an electric shaver and a depilatory agent was applied to the skin to eliminate any
remaining hair. After shaving, a 1 mm thick, water-tight standoff was attached between
the skin and the array (Figure C.1). With the rat in the dorsal decubitus position, a
reservoir within the standoff was filled with insulin (Humulin® R, rDNA U-100, Eli Lilly
and Co., Indianapolis, IN) through a small hole in the array. Care was taken to remove
92
all bubbles from the solution in the reservoir to prevent disruption of ultrasound
transmission. The elapsed time from the initial injection of the anesthetic until the start
of ultrasound exposure was no longer than 25 minutes. For the ultrasound exposure, the
cymbal array was operated at 20 kHz with an Isptp = 102.2 ± 2.3 mW/cm2 for 60 minutes.
After the 60 minutes ultrasound exposure, the array was removed and the skin examined
for visible lesions.
The dose of insulin selected for the injection groups was based on published
insulin doses used to control diabetes mellitus in other species and on pilot experiments
(not reported). Humulin® R Insulin (100 U/ml) was diluted with a 0.9% saline solution
(Phoenix Pharmaceutical, Inc., St. Joseph, MO) to 0.5 U/ml and delivered subcutaneously
(under the skin of the ventral abdomen) at 0.15, 0.20, or 0.25 U/kg.
Figure C.1: Photograph of a transdermal insulin delivery experiment on a rat place in a dorsal decubitus position with the array attached.
93
Blood was collected from the tail vein of each rat to obtain a baseline glucose
level and, following insulin administration, additional samples were collected every 15
minutes for 90 minutes. For the ultrasonic transdermal delivery, the base line glucose
level was measured at the beginning of the ultrasound exposure. The blood glucose level
(mg/dl) for each sample was determined using the ACCU-CHEKTM blood glucose
monitoring system (Roche Diagnostics Co., Indianapolis, IN). Each sample was tested at
least two times to confirm the accuracy of the reading
The data was corrected by subtracting the baseline glucose for each animal from
each data point such that only changes in blood glucose were compared. Statistical
analysis was performed using Microsoft Excel® (Microsoft Corp., Redmond, WA) and
the data of blood glucose versus time were pooled for each group and analyzed as the
mean and standard deviation. A t-test was used to analyze the statistical significance of
the differences among the means of groups. The p-value was used to determine if the
between-group differences are significantly greater than chance.
Results of the ultrasound delivery compared to injections doses for the four
groups are graphed as a decrease in the blood glucose level during the 90- minute
experiment (Figure C.2). The data were graphed and reported as the mean ± standard
deviation (x ± sd) of each group. Due to the anesthesia, the average initial glucose level
at the beginning of the experiment was 340 ± 69 mg/dl for the 16 experiments. Rats not
anesthetized with xylazine would have a blood glucose concentration closer to a normal
level of ≈100 mg/dl.
94
For direct subcutaneous injections with a dose of 0.15 U/kg and 0.20 U/kg, the
blood glucose level deviated little from the baseline value. Overall the glucose level
varied no greater than 32 mg/dl from the initial value over a 90 minute experiment period
for both doses. Yet for a subcutaneous insulin injection of 0.25 U/kg, the blood glucose
decreased by 190 ± 96 mg/dl after 90 minutes. In comparison, the ultrasound produced a
blood glucose decrease of 263 ± 40 mg/dl at 90 minutes. An t-test analysis at 90 minutes
indicated that all the groups were statistically different from each other at a p-level less
than 0.01. Visual examination of the skin exposed to ultrasound did not indicate any
damage or significant changes.
Figure C.2: Changes of blood glucose by the ultrasonic transdermal insulin delivery andsubcutaneous injections of insulin.
-350
-300
-250
-200
-150
-100
-50
0
50
100
0 15 30 45 60 75 90
Time (min)
Ch
ang
e o
f b
loo
d g
luco
se l
evel
(m
g/d
l)
340 ± 69 mg/dl
0.15 U/kg injection (n=4)
0.20 U/kg injection (n=4)
0.25 U/kg injection (n=4)
60 minutes ultrasound exposure (n=4)
-350
-300
-250
-200
-150
-100
-50
0
50
100
-350
-300
-250
-200
-150
-100
-50
0
50
100
0 15 30 45 60 75 900 15 30 45 60 75 90
Time (min)
Ch
ang
e o
f b
loo
d g
luco
se l
evel
(m
g/d
l)
340 ± 69 mg/dl
0.15 U/kg injection (n=4)
0.20 U/kg injection (n=4)
0.25 U/kg injection (n=4)
60 minutes ultrasound exposure (n=4)
0.15 U/kg injection (n=4)
0.20 U/kg injection (n=4)
0.25 U/kg injection (n=4)
60 minutes ultrasound exposure (n=4)
95
For humans to regulate their blood glucose level, the required insulin injection
dose is 0.5 - 1 U/kg/day for adults and children and 0.8 - 1.2 U/kg for adolescents
experiencing growth spurts (Hodgson & Robert J. 2006; Lance et al. 2002). Direct
subcutaneous injection doses for animals range from 0.1 - 0.4 U/kg subcutaneous (SC)
for dogs and 0.1 - 0.5 U/kg (SC) for ferrets (Plumb 2005). Ranges are given since
physiological variables have a direct effect on the specific blood glucose decrease from
injections. As many clinicians and diabetes patients know, the body's glucose response to
direct injections varies according to a host of variables such as body weight, fat
percentage, exercise level, and composition of the most recent meal. Consistency of diet
and exercise along with routine dose-glucose recording assists in the control of blood
glucose. Without proper glucose control, diabetic complications may result, including
renal failure, peripheral vascular disease, and limb amputation (Meeuwisse-Pasterkamp,
van der Klauw, & Wolffenbuttel 2008; Suetsugu, Takebayashi, & Aso 2007).
To facilitate the ability of a diabetic patient to avoid repeated and painful daily
injections of insulin, a safe, light-weight, low-profile, inexpensive and potentially
portable ultrasonic device is proposed. The goal of this task was to develop an
approximate relationship between dose levels from direct subcutaneous injections and
noninvasive ultrasound at a Isptp ≈100 mW/cm2 for 60 minutes. While an exact
mathematical relationship was not determined, the results in Figure C.2 indicate that the
ultrasound dose appears to be greater than an injection dose of 0.25 U/kg for rats. Not
included in the results was a single rat experiment which used a direct injection dose of
0.44 U/kg which resulted in a rapid decrease in blood glucose of 290.5 ± 8 mg/dl after
only 60 minutes. Given this rapid decrease, the animal was removed from the experiment
96
and this dose was determined to be too high for the rats. Yet the single point result can
indicate that the ultrasound dose would be somewhere between injection dose levels of
0.25 - 0.44 U/kg. Nevertheless the results are the first steps in determining a relationship
between ultrasound intensity levels and insulin dose responses. Further experiments
should explore the use of larger animals with a similar size and weight as humans or
animals which are truly diabetic such as pancreatectomized pig.
A person is considered diabetic if his or her blood sugar level is above 126 mg/dl
after eight hours of fasting. People without diabetes have fasting sugar levels that
generally run between 70 - 110 mg/dl. A glucose concentration of 110 - 126 mg/dl is
classified as impaired fasting glucose. In the oral glucose tolerance test, 140 - 200 mg/dl
is impaired glucose tolerance and greater than 200 mg/dl is considered diabetic
(Carnevale Schianca, Rossi, Sainaghi, Maduli, & Bartoli 2003; Rifkin & Porte 1990;
Shaw et al. 1999). In this last situation, a diabetic person would need to inject enough
insulin to reduce his or her blood glucose by about 100 mg/dl. Both the ultrasound and
direct injection of 0.25 U/kg achieve blood glucose level decreases of 190 mg/dl or
greater.
Use of transdermal drug delivery techniques has practical clinical application to
medications which need to be injected multiple times either daily or weekly. A recent
review on ultrasound drug delivery states that "small-sized low-frequency transducers
need to be developed so that patients can wear them" (Pitt, Husseini, & Staples 2004).
As with diagnostic ultrasound imaging, drug delivery using therapeutic ultrasound
requires a delicate balance between safety and efficacy and requires careful scientific
study. For a transdermal device to replace conventional needles, the bioeffects and safety
97
of each device needs to be carefully evaluated, since it will not matter how much of any
drug can be transported if the skin is damaged or the procedure is painful.
Appendix D
MATLAB CODES
D.1 Low pass π-circuit for impedance matching
% ============================================================= % Calculates a low pass pi matching circuit for any array. % Required values of array: % Resonance Frequency (fr), measured complex impedance at fr, % desired Q of the circuit. % Source impedance = 50 ohms % Required condition: % Q>0 % % Eun-Joo Park, modified on July 3, 2007, Tuesday % ============================================================= function [C1,C2,L]=ejpimc(f,RL,XL) %fr=Resonance frequency of array %RL=Transducer resistance %XL=Transducer reactance %Transducer array parameters w=2*pi*f; ZL=RL+j*XL; Q=sqrt(RL/50)+0.01 %Desired Q %Source output impedance parameters RS=50; %Source resistance XS=0; %Source reactance ZS=RS+j*XS; Rc=max(RS,RL)/(Q^2+1); QL=-XL/RL; %Node load Q RpL=RL*(1+QL^2); %Equivalent parallel load resistance CpL=QL/(RpL*w); %Equivalent parallel load capacitance q1=sqrt(RpL/Rc-1); %Match to source C2=(q1/(w*RpL))-CpL %C2 value
99
LL=(q1*Rc)/w; %Inductance due to load QS=-XS/RS; %Node source Q RpS=RS*(1+QS^2); %Equivalent parallel source resistance CpS=QS/(RpS*w); %Equivalent parallel source capacitance q2=sqrt(RpS/Rc-1); %Match to source C1=(q2/(w*RpS))-CpS %C1 value L_S=(q2*Rc)/w; %Inductance due to source L=L_S+LL %L value %Calculate the matched and unmatched impedance over the bandwidth BW=f/Q; %Bandwidth given by specified Q f_axis=linspace(round(f-BW/2),round(f+BW/2),1000); w_axis=2*pi*f_axis; ZC1=1./(j*w_axis*C1); ZC2=1./(j*w_axis*C2); Z_L=j*w_axis*L; Z1=(ZL.*ZC2)./(ZL+ZC2); Z2=Z1+Z_L; Zin=(Z2.*ZC1)./(Z2+ZC1); Yin=1./Zin; %Calculate the unmatched impedance over the bandwidth Ct=1/(2*pi*f*abs(XL)); Zt=RL+1./(j*w_axis*Ct); %Plots figure(1) f_axis=f_axis./1000; subplot(2,1,1) plot(f_axis,abs(Zin)) grid on xlabel('f (Hz)') ylabel('|Zin|') title('Magnitude of Impedance'); subplot(2,1,2) plot(f_axis,(180/pi)*angle(Zin)) grid on xlabel('f (Hz)') ylabel('Phase (deg)') title('Phase of Impedance');
100
D.2 Feedback controller
% ============================================================= % operating program with steps of % 1. Run ultrasound system for glucose sensing (10 minutes) % 2. Wait for glucose sensing with potentiostat (8 minutes) % 3. Run ultrasound system for insulin delivery (2 minutes + 18 % minutes) % 4. Repeat step 1-3 % % Eun-Joo Park, modified on January 4th, 2009 Sunday % ============================================================= function controller1(current) for ii=1:3 turn=ii % Turn on US system of GS fg33120a; a=1 % Pause process for manual measurement of current from potentiostat t_GS=7; % time for glucose sensing by potentiostat with two biosensors pause(t_GS*60); % pause (seconds) b=2 % Read current at 700 mV load GS_FC_1.txt; load GS_FC_2.txt; load GS_FC_3.txt; load GS_FC_9.txt; i_gs1=min(GS_FC_1(:,2)) i_gs2=min(GS_FC_2(:,2)) i_gs3=min(GS_FC_3(:,2)) i_gs9=min(GS_FC_9(:,2)) % i_gs=mean([i_gs1, i_gs9]) i_gs=mean([i_gs1, i_gs2, i_gs3, i_gs9]) c=3
101
if i_gs>=current; fg33250a_on; else fg33250a_off; end d=4 t_TD=3; % time to wait for the next GS (minutes) pause(t_TD*60); % pause (seconds) e=5 end % Turn on US system of GS fg33120a; aa=1 % Pause process for manual measurement of current from potentiostat t_GS=7; % time for glucose sensing by potentiostat with two biosensors pause(t_GS*60); % pause (seconds) bb=2 % Read current at 700 mV load GS_FC_1.txt; load GS_FC_2.txt; load GS_FC_3.txt; load GS_FC_9.txt; i_gs1=abs(min(GS_FC_1(:,2))) i_gs2=abs(min(GS_FC_2(:,2))) i_gs3=min(GS_FC_3(:,2)) i_gs9=min(GS_FC_9(:,2)) % i_gs=mean([i_gs1, i_gs9]) i_gs=mean([i_gs1, i_gs2, i_gs3, i_gs9]) cc=3
VITA
EUN-JOO PARK
Eun-Joo Park was born in Choong-Nam in Korea (Republic of). She received her
BS and MS degrees in Physics (Physical Acoustics) from Sung Kyun Kwan University,
Seoul, Korea in 1996 and 1998 respectively. The thesis of her MS degree was about
“The thermal effect of ultrasound in transdermal insulin delivery.” She received another
MS degree in Mechanical Engineering (Underwater Acoustics) from Boston University,
Boston, MA in 2003. The thesis was about “Acoustic characterization of saturated
sediments using a thick-walled, water-filled impedance tube.” After her second MS
degree, she worked at the Focused Ultrasound Lab in Brigham and Women’s Hospital,
Boston, MA as a researcher. In order to study biomedical ultrasound for her PhD degree,
she joined the interdisciplinary program in bioengineering at The Pennsylvania State
University in 2004 and has been working as a research assistant in the Therapeutic
Ultrasound Laboratory. Her research deals with biomedical applications of the cymbal
transducer arrays on noninvasive transdermal drug deliveries. Eun-Joo’s research
interests are the diagnostic and therapeutic ultrasound in medical applications especially
for cancer treatment and diabetes care. She is also interested in the research of
developing ultrasound transducers for specific medical applications.