1. Assessing the stability of microbubbles over time
Katherine Kiang
Bioacoustics Research Lab, Electrical and Computer Engineering, University of Illinois Urbana
Champaign
Background
Microbubbles, which are small gas filled bubbles between one micrometer and one
millimeter in diameter, have many potential biomedical applications in the diagnosis and
treatments of diseases. More specifically, microbubbles can be used as ultrasonic contrast agents
in order to image of many areas of the body, such as the heart, prostate, or liver [1] [2]. The
pressure caused by the high frequency of ultrasound scanner cause the microbubbles to contract
and expand. This causes the bubbles to be more reflective and therefore easier to see.
Additionally, ultrasound scanners can tune into the “overtones” created by the resonance of the
microbubbles, allowing for extremely targeted and specific imaging [3]. For better use of
microbubbles in this context, it is of interest to understand how these microbubbles change over
time. To do this, microbubbles were made, and the size, microbubble number density, collapse
threshold, and attenuation were studied over a course of six weeks. The collapse threshold refers
to the pressure needed to burst the microbubbles while the attenuation refers to the reduced
strength of a transducer signal as microbubbles are added.
2. Procedure and methods
Production of microbubbles [4]
A large batch of albumin based one micron bubbles were made before the experiment
started. The dextrose stock solution was made with 15g dextrose [Fisher Scientific], 100mL
18mΩ water, 0.12 g Na2HPO4 [Sigma-Aldrich], 0.02g KH2PO4 [Sigma-Aldrich], and 0.1g
NaN3 [Sigma-Aldrich. The BSA stock solution was made with 5g BSA [Sigma-Aldrich], 100mL
18mΩ water, and 0.01g NaN3.
Each batch of microbubbles was fabricated using 12ml of 15% dextrose and 4ml of 5%
BSA in a 50ml centrifuge tube. N-decafluorobutane gas was then added to the tube for about six
seconds in order to purge the tube of the air. Immediately after the tube was vortexed for a
minute and transferred to a 40 mL ultracentrifuge tube. More n-decafluorobutane was added for
a three seconds and the tube was sonicated for 55 seconds at 90% power (450 Watts) (Figure 1).
A metal tube holder was used to hold the 40mL ultracentrifuge tube during sonication. Once the
tube was sonicated, 10 mL of DPBS was added to the tube and it was inverted to mix a few times
and let to sit 5-10 minutes on a rack.
Figure 1: Mock set-up of microbubble sonication technique
3. After fabrication, the bubbles had to be separated by size. The bubbles were brought up
into a 20 mL syringe. Air bubbles were removed and care was taken so as not to draw up any of
the foam that forms on top of the bubbles during sonication. Once the bubbles were in the
syringe the needle was taken off and the syringe capped and left to sit vertical in the fridge for
one to six hours.
More DPBS was added to the remaining bubbles in the ultracentrifuge tube and allowed
to sit for 5-10 minutes, after which they were put into another syringe by the same process
described above. The microbubbles were then left in the fridge for 1-6 hours. During this time, a
white band formed near the top of the syringe as the microbubbles separated by size. Size
separation was confirmed by suspending the white band in 5 mL of DPBS to view under the
microscope.
Once the bubbles finished rising, there were two parts of the bubbles left. The top white
band, and the lower liquid fraction. The lower liquid fraction was the 1 micron bubbles desired
(Figure 2). Therefore, that part was transferred to a new syringe to rise overnight in the same
manner as before. After the bubbles rose a second time and the lower fraction transferred into a
new syringe, they were ready for use. Seven 8mL syringes were used for each week of counting,
sizing, and DPCD. Three tubes of 30mL were used for three different attenuation experiments.
4. Figure 2: Rising Microbubbles [4]
Counting and Sizing The Microbubbles
Counting and sizing was done the day after the microbubbles were fabricated and then
every week thereafter for six weeks.
To count the microbubbles, a small amount of microbubbles were loaded into a
hemocytometer. Frist, a syringe of the previously prepared microbubbles was taken out and
gently rolled in someone’s hands to re-suspend the bubbles. Then, a few drops of the bubbles
were put into a coulter counter tube. If the microbubbles needed to be diluted (which only
happened the first week), another tube was prepared with a small about of DPBS. In a third tube,
a 1:9 ratio of microbubble to DPBS was mixed by using a pipette to draw the mixture up and
down. Once the microbubbles were prepared, a pipette was used to draw up about 100
microliters of the bubbles (either straight up or diluted). This was carefully loaded into one side
of the cleaned and prepared hemocytometer.
5. Once the hemocytometer was prepared a phase contrast microscope was used to image
the microbubbles inside the hemocytometer. For the counting images, 20x magnification was
used.
Images were taken of all 25 squares on the hemocytometer using the microscopes software. One
image was taken focused on the bubbles, another focused on the hemocytometer squares.
Counting was done by overlaying the images on Power Point (taking advantage of
transparency settings) and manually counting using a counter. The microbubble density equation
is as follows:
Number density = (N*D)/(Q)
N = number manually counted; D = dilution factor Q = number of squares
To size the microbubbles, 10 images were taken at random at 100x on the same
microscope. These images were focused solely on the microbubbles and converted for RGB to
Greyscale before saving. These images were analyzed in the Matlab software (DPCD suite
bubble size) written by Daniel King [5]. To do this the bubble sizing software was opened and
each image individually loaded (load new image button). Once the image was opened, the detect
circles function was able to be used. Manual size adjustment was also possible if needed. In the
Matlab workspace the x-axis would give the radius of the bubbles in microns and the count
would say how many bubbles were at each size. An average of the microbubble radius was taken
by adding up the amount of bubbles at each radius and multiplying it by the radius. This number
was then divided by the total amount of bubbles sized. Excel was used to keep track of the
microbubble size data.
6. Finding the collapse threshold of microbubbles
Assessing the collapse threshold of the microbubbles was done the day after the
microbubbles were fabricated and then every week thereafter for six weeks.
Finding the collapse threshold was done through a DPCD experiment designed by Dr.
Daniel King [5]. To set up this experiment, a RITEC RAM-5000 (machine to control the
pressure) and a 5MHz transducer were connected to a computer via an already inserted A/D card.
Pressure settings for the RITEC were calibrated weekly.
To connect the RITEC and transducer to the computer a number of materials were needed
– a small tank filled with degassed water (placed above a stir palate, a PCD detector, the
transducers, and cables. A PCD detector with a 5 MHz (transmit) transducer in the middle and a
receiver transducer on each side was placed inside the tank (Figure 3).
Cables were connected as follows:
5 MHz transducer & RF Burst No. 2 – L/R side of attenuation bar, respectively
R/L Receiver Transducers (2) - Receiver A No. 1 and Receiver B No 2, respectively (should be
diagonal)
Receiver A/B RF Monitor (2) – A/D card channel 1/3 (respectively)
Sync (back of RITEC) – Sync (A/D card)
7. Figure 3: Mock DPCD Setup
Once the setup was complete, the RITEC was powered on and 4Ch bubble snapshot was
opened on the computer.
The RITEC settings were as follows:
Receiver: Gain-22dB, High pass filter-1MHz, Low pass filter-20MHz
Triggers and Gates: Internal source, PRF – 10 Hz
GA-2: Frequencey-4.6 MHz, Number of cycles: 3
A directory was made and the 4Ch bubble snapshot settings were as follows (most default):
Sampling rate: 100 MHz, Channel Mode: Dual Mode, Voltage Range: 1V, No. of Points: 16k,
Trigger Edge: Rising, Trigger Level: 10, Coupling: DC, Impedance: 50 Ohms; Time between
repeats: 0
Baseline values were then taken. The RITEC was turned to a setting of 10 and using 4Ch
bubble snapshot, the transducer focus was found (look for a peak around 47 milliseconds) by
moving the PCD detector around the tank with the addition of the 50 micron wire with weight on
the end.
8. Another base value was needed with the detector positioned with its top just below the
highest point in the tanks water level. This was done using clamps. A magnetic stir bar was then
inserted into the tank and the stir plate turned onto a low setting (3-4). Once the PCD was
properly in place, 50 scans were taken with no RITEC output and bubbles to act as a control.
Once all the baselines were taken the actual DPCD experiment was able to be conducted.
To run the experiment, microbubbles (which were re-suspended using the same method
described in the sizing procedure) were added a few drops at a time and 500 snapshots for each
of ten different RITEC settings (4, 1, 20, 4, 8, 10, 6, 12, 17, 15 – see Table 1 for corresponding
pressure values) were taken. This process was repeated 10 times. Bubbles were added when
signals could not be seen. Over the course of the experiment, bubbles were added so as to make
sure signals were seen at smaller settings.
Table 1: Corresponding Pressure Setting For Each RITEC Setting
These results were analyzed in the same DPCD suite used to size the bubbles. This time
however, the Analyze/Auto classify software was used. Once this software was opened, the
dataset from the DPCD experiment was loaded using load dataset (frequency=4.6) and auto
classified. Care was taken to adhere to the correct naming convention to make sure the auto
classifier ran properly. After the data was classified the Plot Post excitation curve software was
9. used to plot the data in a graph which correlated the RITEC pressure settings to the percentage of
microbubbles that burst at that setting. The auto classified files were loaded into this software as
.mat files (put in 5MHz for transducer) and the pos-excitation curve was plotted. Because there
were few to no signals at many of the low (1,2,4) settings, the curves didn’t fit well every week.
To help analyze the curves, they were plotted with a -10 db minimum threshold. The area of
interest in analyzing this data was the 50% threshold, in other words the pressure at which 50%
of the microbubbles collapsed.
Attenuation of Microbubbles
Attenuation was tested after weeks 0, 3, and 6. This test used a 7.5 db transducer and a
UT340 Pulser Receiver System.
To setup the attenuation experiment a stir-plate was placed inside the largest tank in the
daedal room. Next a smaller tank was placed on top of the stir plate and filled with degassed
water. Then a hydrophone and the transducer were placed into the smaller, water filled tank
(Figure 4a, 4b). The hydrophone was toward the back and connected to its amplifier, which is
connected to Ch1-PDA 14 in the UTEX. The transducer in front is connected to Pulse Rec in the
UTEX. Finally, Sync Out from the computer is connected to Sync on the UTEX.
Figure 4: a) Mock Attenuation Setup: Eye Level b) Mock Attenuation Setup: Overhead
10. Similar to calibration, the focus of the transducer must be found. This is initially done by
hand and then fine-tuned using the Daedel Position Menu to move the Daedel and PDA 14 to see
the transducer signal. The UTEX settings when finding the focus are: Voltage- 200 V ; Pulse
Width- 40 ns; Rep Rate- 200 Hz; Internal; PE/Gain – 0; PC/Gain – 0 Mode – Pitch Catch.
Once the focus was found baseline snapshots were taken with the transducer at the focus,
12 micrometers in front of the focus, and 12 micrometers in back of the focus. These snapshots
were taken with the same UTEX settings as before except with 100 V and with varying widths.
The widths tested were: 2 ns, 4 ns, 6 ns, 8 ns. The transducer was then kept at 12 micrometers in
front of the focus and a plastic U shaped holder wrapped in Ceram Wrap (carefully, without
double layering and using rubber bands) and with a small stir bar inside was placed between the
transducer and hydrophone. To place the holder, one person lowed the holder while the other
poured in water using a beaker. The stir plate was then turned on. Once the holder was in place,
control snapshots of the images were taken at all the UTEX settings (the differing widths).
To begin the experiment, 6 mL of microbubbles were added by putting an 18G needle on
the microbubble syringe. Once microbubbles were added and allowed to disperse for about a
minute, images of the signal were taken at each UTEX setting. This was done for 3 time points.
Then another 6 mL was added and the process repeated. In the end, there was a 5x concentration.
Analyzation of the attenuation data was done by comparing the height of the signals at
each width as the microbubble concentration increased.
11. Results
Figure 5: Microbubble concentration for each week
Figure 6: The radius (in micrometers) of the microbubbles each week of testing
y = -4E+07x + 2E+08
R² = 0.754
1.00E+00
1.00E+01
1.00E+02
1.00E+03
1.00E+04
1.00E+05
1.00E+06
1.00E+07
1.00E+08
1.00E+09
0 2 4 6 8
MicrobubbleConcentration
Week
Weekly microbubble concentration
Count
Linear (Count)
y = 0.0274x + 0.5159
R² = 0.9109
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0 2 4 6 8
MirobubbleSize(Radius)
Week
Weekly MicrobubbleSize
Size
Linear (Size)
12. Figure 7: 50% Postecitation each week of testing
Figure 8: Attenuation amount each week for UTEX 2 (Series # is Concentration)
-1
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1 2 3
AttenuationAmount
Week
Attenuation Over Time UTEX 2 Setting
Series1
Series2
Series3
Series4
Series5
Linear (Series1)
Linear (Series2)
Linear (Series3)
Linear (Series4)
Linear (Series5)
y = 0.1429x + 2.7857
R² = 0.2597
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
0 2 4 6 8
RITECSettingWith50%Postexcitation
Week (**week 5 N/A becausedidn't hit50% postexcitation)
Weekly 50% Posexcitation
50% Posexcitation
Linear (50%
Posexcitation)
13. Figure 9: Attenuation amount each week for UTEX 4 (Series # is Concentration)
Figure 10: Attenuation amount each week for UTEX 6 (Series # is Concentration)
-1.6
-1.4
-1.2
-1
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
1 2 3
AttenuationAmount
Week
Attenuation Over Time UTEX 4 Setting
Series1
Series2
Series3
Series4
Series5
Linear (Series1)
Linear (Series2)
Linear (Series3)
Linear (Series4)
Linear (Series5)
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1 2 3
AttenuationAmount
Week
Attenuation Over Time UTEX 6 Setting
Series1
Series2
Series3
Series4
Series5
Linear (Series1)
Linear (Series2)
Linear (Series3)
Linear (Series4)
Linear (Series5)
14. Figure 11: Attenuation amount each week for UTEX 8 (Series # is Concentration)
Analysis and Discussion
After running an ANOVA on the microbubble count data, a p value of 0.016039 was
found, suggesting a statistically significant difference in the microbubble concentration over
time. The microbubble size data had a p value of 0.012359, suggesting a statistically significant
difference in the microbubble size over time. This implies that the microbubbles were not stable
in their concentration and size over the duration of the experiment. A possible explanation for
this instability would be that the bubbles burst during storage. Since the average microbubble
size increased, it would indicate that the smaller microbubbles burst first.
Looking at the post excitation curve data over time, there doesn’t seem to be a statistical
significance, as the p value was 0.597098. This implies stability of the microbubbles over time in
regards to their collapse threshold. However, this could be related to not seeing many at low
RITEC settings during DPCD, which could lead to a less accurate post excitation curve.
-1
-0.5
0
0.5
1
1.5
2
1 2 3
AttenuationAmount
Week
Attenuation Over Time UTEX 8 Setting
Series1
Series2
Series3
Series4
Series5
Linear (Series1)
Linear (Series2)
Linear (Series3)
Linear (Series4)
Linear (Series5)
15. When analyzing the attenuation data, the attenuation amounts had an inconsistent mixture
of negative and positive values for the same concentration and UTEX settings. A positive value
implies the signal shrunk and vice versa. Therefore, positive and negative numbers do not make
sense. What should be seen is increasingly positive numbers as the concentration of
microbubbles increased.
Conclusions
Overall, this experiment showed some significant differences in microbubble count and
concentration over time. This instability seems to contradict the hypothesis that the albumin
based microbubbles are stable over time. However, there seems to be stability in the collapse
threshold of the microbubbles over time. This inconsistency suggests the need more research to
be done, possibly with microbubbles at higher concentrations. Additionally, the attenuation
experiment must be assessed to see if there are problems with the experimental method causing
the illogical data.
Sources:
[1] "European Heart Journal - Cardiovascular Imaging." Microbubbles and Ultrasound: From
Diagnosis to Therapy. N.p., n.d. Web.
[2] F. J. Fry, N. T. Sanghvi, R. S. Foster, R. Bihrle and C. Hennige. Ultrasound and
Microbubbles: Their Generations, Detection and Potential Utilization in Tissue and Organ
Therapy - Experimental. Ultrasound in Medicine and Biology, 21 1227-1237, 1995.
[3] Blomley, Martin J K, Jennifer C. Cooke, Evan C. Unger, Mark J. Monaghan, and David O.
Cosgrove. "Microbubble Contrast Agents: A New Era in Ultrasound." BMJ : British Medical
Journal. BMJ, n.d. Web.
16. [4] Borrelli, Michael J., and William D. O'Brien. "Result Filters." National Center for
Biotechnology Information. U.S. National Library of Medicine, n.d. Web.
[5] D. A. King and W. D. O’Brien, Jr., “Quantitative Analysis of Ultrasound Contrast Agent
Postexcitation Collapse,” IEEE Trans UFFC 61:7, 1237-1240 (2014).