1. Vascular Locator LLC
Group 14: ​Daniel Foster, Angela, Nishant Parajuli, Glen Filo, Truong Mai
Final Report ­ Spring 2016 ­ NIH
1) Abstract
Our device uses doppler ultrasound to locate arteries for the medical procedure arterial line placement.
Arterial line placement is used to monitor real time blood gas and blood pressure. These are highly
crucial diagnostics for surgeries or patients in critical care. The gold standard for artery locating,
imaging ultrasound, is expensive so our solution is to use the cheaper doppler ultrasound and our
Android app to accurately locate arteries. The principal theory behind doppler ultrasound is the doppler
shift. When a sound wave reflects off of a moving object, its pitch is shifted. This allows the doppler
ultrasound to detect blood flow by then taking the difference between the emitted and the reflected
sound and outputting it as a normal audio signal. Our app analyzes this signal by first filtering out noise
and then calculating the volume of the filtered signal. The value of the loudest volume is stored and
automatically updated. A ratio of the current volume over the loudest is calculated and the color of the
app's display is a function of this ratio. The screen turns fully red when the ratio = 1 and blue when the
ratio = 0, with shades of purple in between. Our device also incorporates a marker attached to the
probe by a holder which allows the location directly underneath the probe to be marked. When two or
more locations on the same artery are marked a line is drawn between them to indicate the location of
the artery.
2) Description of the problem
Arterial line placement (which involves sticking a cannula into the radial artery) is a common procedure
used to monitor patients blood pressure and gas levels in real time which are useful diagnostics for
patients in surgery or critical care [2]. However, there are some significant obstacles in carrying out the
radial artery line placement due to the depth of the radial artery. The existing solutions are using
imaging ultrasound, which is expensive and requires high electrical power (since it needs to generate
many powerful ultrasonic pulses), or simply feeling the general area of the pulse and poking the patient
until you find the artery which is inaccurate and could cause additional injury if you poke a nerve [5].
Since many communities, especially in impoverished and remote locations, lack sufficient health care
services, electricity, and doctors, the people there always suffered from late treatment, which causes an
increase in mortality rates from natural disasters. In our project, we target those organizations which aid
those patients and put our designed app into practice to help them locate the radial arteries in the
patient’s forearm. Through testing the accuracy of our device on different people and improving the
results on our arm phantom, which is made by our team to mimic the physical properties of a real
person’s arm, we have created a user­friendly and highly accurate device that has the potential to come
on the biomedical business market and contribute to rescuing the injured. The feeling for the pulse
method or “Allen’s Test” is also used in many first world hospitals which allows for another potential
market for our device as a cheaper alternative to imaging ultrasound which is sometimes used to assist
arterial cannulation [13].
3) Project objective statement:
(1). How we address this problem​ :
A. We can target the radial artery through knowing the general location and the fact that
veins,which are a source of noise, do not create a pulsed doppler signal or as loud of a
signal as the artery. And the artery we’re looking for is the biggest in its area so it creates

2. the the loudest signal. We’re looking for the loudest signal of that loudest artery in order to
detect the largest part of the artery.
B. The probe is held in a holder with a marker. After an artery is located we use the lever on
the holder to rotate the probe out of the way and leave a mark where the probe was. The
user can make multiple marks along the artery and draw a line along those marks to
improve the information available to the user and increase the likelihood of a successful
cannulation
(2).The innovation of our approach:
A. Our Android App is developed to process audio signals from the doppler system into a color
changing display that helps the user accurately locate their radial artery. No other
application exists that locates arteries using a doppler ultrasound system. Standalone
doppler ultrasound systems are usually used to monitor fetal heart rate or as a stethoscope
with speaker. Doppler ultrasound technology is also sometimes incorporated with imaging
ultrasound to monitor blood velocity.
B. Our marker holder allows us to make a mark directly underneath the doppler probe. This
also does not exist.
(3).How’s the final design solves the problem
Our app changes colors when over the probe is over an artery and it changes colors more
dramatically when over the loudest part of the signal. This has been tested to be a fairly accurate
method (see Part 6) of locating a pulsating vessel. Our marker allows us to mark the location of
this loudest signal and with several marks our device can show the location of the radial artery.
(4).Specifications
● Table 1: ​Specifications for Doppler Flowmeter Model 810­A [7]
Manufacturer Parks Medical Electronics
Model Number 810A
Primary Function blood velocity detector; non­invasive
Power & output source D cell batteries with speaker & headphones
Nominal Operating Frequency 8­9 MHz
● Table 2: ​Specifications for Vascular Locator Android Application [8]
Architecture multithreaded audio processing
Library TarsosDSP
Platform Android Studio in Java
Audio Bit Depth 10
Sample Rate 22k

3. Frames per Second 20
Frequency (high­pass) 800 Hz
Probe Diameter 1 cm
Ultrasound Depth (9 MHz) 4 cm
4) Documentation of Design
The marker holder is made of wood. It is attached to the probe at 45 degree angle as shown in ​Figure
6​. The holder is easy to build and cost­effective. The main function of the marker holder is to mark the
location of artery. After Vascular Locator (Android Application) gives a confirmation that an artery has
been detected, the marker then marks a point on the skin over the artery. After multiple points along the
artery have been made it is lines can be drawn between them to reveal the location of the artery.
Fig. 1:​ Mark down two locations of artery with line drawn in between
The mobile application (Vascular Locator) is attached to the ultrasound doppler system (Parks Medical
810A) using an audio jack cable. The main function of Vascular Locator is to find the location of artery.
The application development was conducted in several phases. At first, the graphical interface
consisted of amplitude (y­axis) versus frequency (x­axis) graph. The occurrence of simultaneous peaks
indicated the location of artery was found. However, taking into consideration that a lay person might
find it difficult to interpret the data, the group realized that a more user­friendly interface was required.
We utilized a color coded design. The location of artery was indicated through a color change between
blue and red which represents the pulse.
Our custom made android application works by doing an analysis of the output signal of the doppler
ultrasound device, filtering out the noise(with a high pass filter at 800Hz) and then calculating the
volume from the filtered signal. A ratio of the current volume over the loudest is calculated and the color
of the app's display is a function of this ratio. Since the color display is that of a spectrum, there are
shades of purple in between the red color, when the ratio is equal to one, and blue, when the ratio is
equal to zero.
Parks ultrasound is a cheaper system than that of an imaging ultrasound because it has one transducer
only, while imaging ultrasounds have a multitude of transducers. With regards to light based vascular
detectors, they cannot be used to locate the radial artery as they do not have a great depth of
penetration (i.e. 3 mm). For example, from Veinviewer data, the accuracy falls to less than 50% if the

4. depth of the intended vessel of interest is greater than 3 mm [12]. The radial artery is, on average,
located about 3 to 4 mm from the surface of the forearm, with a diameter of 2.6 mm [9]; from literature,
an ultrasound frequency of 9 MHz could roughly penetrate up to a depth of 4 cm, with good spatial
resolution [10]. The diameter of the probe is 1 cm, about thirty­eight times larger than artery; therefore,
while proven to work in the detection of the radial artery, a probe with a much smaller diameter would
have a higher accuracy in locating the artery due to being near­identical in dimensions; an example of
one such probe would be a Parks Medical microvascular probe with a 2 mm diameter tip [11].
The doppler ultrasound system has a normal audio output due to taking the difference between the
emitted and reflected frequencies and outputting it through a 1/4" audio jack (which is why we placed a
1/4" to 3.5 mm adapter into the device audio port and then placed an auxiliary cable into that). The
adapter in the phone audio jack takes the signals from the TRS aux cable and sends them to the
microphone channel of the TRRS phone audio jack.
We also designed an arm phantom to test our device (see Part 6).
5) Prototype of the Final Design
Fig. 2: ​Final prototype design (doppler ultrasound unit, probe and marker holder, android application on
tablet) on the left and the probe with the marker holder on the right.
The image above is what our vascular locater device would look like should it be ordered by a customer
today and then shipped out. As previously mentioned, the entire package would include a doppler
ultrasound unit, probe cover, marker holder, and the android application; in theory, our device was
tested on two different doppler ultrasound systems, therefore should the customer have their own
doppler ultrasound device with an audio/headphone port, they can simply use their ultrasound device
and pay for the application, probe cover, and marker holder rather than purchase a complete package.
https://www.youtube.com/watch?v=106Bg_V7jHQ&feature=em­upload_owner
This video above summarizes the main features of this paper along with giving the viewer a live
demonstration of our android application connected to our Parks Medical doppler ultrasound device that
is being used to locate the radial artery on a patient. First, we described our product background along
with medical need for this system before moving on to our demonstration. We then showed a patient’s
forearm lying flat on a table with the operator then coming in and applying ultrasound gel below the
wrist on the same side as the thumb while then placing the probe onto the gel covered area and then
turning on the doppler ultrasound device. From there the operator then begins to scan over the area
until we see the app detect the radial pulse, the loudest noise, by changing the screen color to red and
then going back to blue after the pulse contraction; the colors begin to change quickly from red to blue
in accordance with the rhythm of the radial pulse. The location is then marked using the marker and
from there, the operator then moves downward until they detect another area where the app behaves
the same as it did in the previous area. After marking the second location with a mark, the operator
then draws a line between the two points in order to mark the pathway for insertion of the catheter into
the radial artery (see figure 3 in part 6 below).

5.
6) Proof that the design is functional and will solve the problem
Moving the probe quicker makes louder noise, so we try to move as slow as possible to reduce it. To
filter this out completely, we conducted an experiment on two team members to make sure that the
result was reasonable. We used Parks medical ultrasound model 810A and Audacity to collect data.
Volume of the ultrasound device is fixed. Our procedure is step 1: apply ultrasound gel. Step 2: press
the probe to the skin near the wrist. Step 3: turn on the doppler machine. Step 4: run Audacity to record
data. Step 5: plot and analyze data on Excel
Fig. 3:​ Signal in radial artery is stronger than noise at >800Hz in Glen and Truong’s right forearms
We observed that at low frequencies scanning noise is much louder than the artery signal so it is
necessary to filter it out. After figuring out that the high pass filter of 800Hz was an optimal one, we
moved on to design an Android app which first measures the loudest volume which is the pulse. When
the probe is scanning around, the app will collect lower volume signals to calculate ratios. The ratio is
current volume over loudest volume. If ratio is 1, the app will turn bright red. If the ratio is 0, the app will
turn bright blue. Ratio values between 0 and 1 are mixed colors of red and blue. When we see the app
turns bright red and bright blue alternatively, we use a marker to mark the location down. Then we need
to find and mark down another location. The mid­point is where to place the needle in. Tap the screen
to reset. Reset when the loudest volume is collected due to moving the probe too fast.
The app worked accurately in our accuracy test on a highly realistic arm phantom. This arm phantom
has muscle made of gelatin and metamucil which mimic the acoustic properties of the arm [3], radial
vein, radial nerve, radial artery, 3D­printed porous radius bone (missing link), and two tendons. Inner
diameters of the latex tubes mimicking vein, and artery are 3 and 4mm respectively. Outer diameters of
them are 6 mm. Nerve has diameter of 6mm. Depth of artery is roughly 4 mm. [15] Location of parts are
made as close as possible to a real human arm. Doppler fluid has particles having similar size to red
blood cells. Flow rates in vein and artery are 20 [4] and 40 ml/min [6] respectively which are operated
by two Harvard Apparatus PhD 2000s. The pulse maker, which is a servo, has a custom propeller to
pinch the tube 80 times/min. We pumped artificial blood through these vessels which mimics the
acoustic properties of the human blood. We purchased this from CAE Healthcare. [14] The doppler
fluid is stable, even it exposes to air.

6.
Fig. 4:​ Top and side views of the phantom. Ten drops of red food color makes the phantom red
Our full arm phantom testing apparatus consisted of two syringe pumps, one attached to the artery and
the other attached to the vein. A pulse generator attached to the artery, controlled by an arduino.
Artificial blood is pumped through the vessels. And finally our doppler system, probe, and app.
Fig. 5: ​Probe and marker holder. Pressing the bolt down will make the probe swing to the left while
moving the marker tip down. A point is marked when the tip touching the skin. This point is the
center of the probe when the probe is perpendicular to the skin surface. The image on the right is
testing the app
Accuracy was tested by having Person A slowly scanning the probe back and forth over the arm
phantom vessel while Person B viewed the app. Person B would then tell Person A where to stop over
the artery based only on the app’s display. The distance from the artery was then recorded from five
tests which ​yielded a mean distance of 1mm away from artery with a standard deviation of
0.6mm.
Fig. 6:​ Setting up accuracy test experiment.
From left to right(background): Syringe Pump with artificial blood; Pulse Generator; Arm
Phantom; Second Syringe Pump; Doppler system;
From left to right (Foreground): Usb battery powering the Arduino; Arduino which controls
the pulse generator; Doppler probe and marker in holder; Android tablet running our app
with audio jack connected to Doppler system

7.
To test user friendliness of the app, we conducted a false/positive test. The user operates the probe
and slowly scans it over the phantom. The subject is viewing the app’s display and cannot see the
probe’s location. The experimenter either scans over the vessel or not over the vessel and asks the
subject if they think the probe is over the vessel or not based on the app’s display. The subjects in this
experiment were all fellow students who did not have experience with our project.
Table 3:​ False/positive test results
Reference
[1] "Basic Physics of Ultrasound and the Doppler Phenomenon." Web. 27 May 2016.
[2] Timothy A. Morris, Andrew L. Ries, Richard A. Bordow.” Manual of Clinical Problems in
Pulmonary Medicine.” 6th edition. Print.
[3] Bude, RO, and RS Adler. "An Easily Made, Low­cost, Tissue­like Ultrasound Phantom Material."
National Center for Biotechnology Information. U.S. National Library of Medicine. Web. 23 May
2016.
[4] Zhang, Min, Xiu­Xin Fang, Ming­E Li, Chun­Hui Zheng, Xi­Huan Zhou, and Xiao­Qin Lyu.
"Handgrip Exercise Elevates Basilic Venous Hemodynamic Parameters in Healthy Subjects." Web.
24 May 2016.
[5] Shima, H., K. Ohno, K. Michi, K. Eqawa, and R. Takiquchi. "An Anatomical Study on the
Forearm Vascular System." Pubmed. Web. 23 May 2016.
[6] "Blood Vessels." Web. 23 May 2016.
[7] "Parks Medical Electronics, Inc. Full Product Listing." Parks Medical Electronics, Inc. Full
Product Listing. 2013. Web. 24 May 2016.
[8] "JorenSix/TarsosDSP." ​GitHub. 2016. Web. 26 May 2016.
[9] Seto, Arnold. “Ultrasound Guidance for Radial Access: Getting in the First Time.” Web. 27 May
2016.
[10] “The principle of ultrasound.” Echopedia. Web. 27 May 2016.
[11] Doppler Probes. Parks Medical Electronics, Inc. Web. 27 May 2016.
[12] ​ ​"The of Value of Width Projection Accuracy Relative to Vein Depth Real Depth Accuracy Value
(RDAV)."Christiemed. Web. 27 May 2016.
[13] "Annex I Modified Allen Test." ​NCBI. Web. 28 May 2016.
[14] Sarkar, Suman. "Artificial Blood." ​Indian Journal of Critical Care Medicine : Peer­reviewed,
Official Publication of Indian Society of Critical Care Medicine. Medknow Publications. Web. 29 May
2016.

8. [15] Kang, In Gu, Won Joon Jeong, and Kyoung Ming Moon. "The Relationship Between Radial
Artery Depth and Wrist Extension Angle Measured by Ultrasonography." Journal of The Korean
Society of Emergency Medicine, n.d. Web. 27 May 2016.
Appendix
App Source Code
import ​android.app.Activity;
import ​android.graphics.Color;
import ​android.os.Bundle;
import ​android.util.Log;
import ​android.view.View;
import ​android.view.WindowManager;
import ​android.widget.TextView;
import ​be.tarsos.dsp.AudioDispatcher;
import ​be.tarsos.dsp.AudioEvent;
import ​be.tarsos.dsp.filters.HighPass;
import ​be.tarsos.dsp.io.android.AudioDispatcherFactory;
public class ​MainActivity ​extends ​Activity {
​private int ​sampleRate ​= ​22050​;
​float ​filterFreq ​= ​800​;
​final ​String ​TAG ​= ​"DAN"​; ​//Log.d(TAG,""+volume);
private int ​bufferSize ​= ​1024​;
​double ​volume ​= ​0​;
​double ​HIGH​=​0​;
​double ​highVolume ​= ​HIGH​;
​double ​ratio1 ​= ​0​;
​double ​powerAdjust​=​.5​;​//below 1 will make it turn red more easily, above 1 will make it harder
private ​TextView ​real​;
​private ​TextView ​loudest​;
​private ​TextView ​ratio​;
​@Override
​public void ​onCreate(Bundle savedInstanceState) {
getWindow().setFlags(WindowManager.LayoutParams.​FLAG_FULLSCREEN,
WindowManager.LayoutParams.​FLAG_FULLSCREEN);
​super​.onCreate(savedInstanceState);
setContentView(R.layout.​activity_main);
​loudest ​= (TextView) findViewById(R.id.​loudest);
​real ​= (TextView) findViewById(R.id.​real);
​ratio ​= (TextView) findViewById(R.id.​ratio);

9. AudioDispatcher dispatcher = AudioDispatcherFactory.​fromDefaultMicrophone(​sampleRate​,
bufferSize​, ​0​);
dispatcher.addAudioProcessor(​filter​);
​new ​Thread(dispatcher,​"Audio Dispatcher"​).start();
}
HighPass ​filter ​= ​new ​HighPass(​filterFreq​, ​sampleRate​){
​@Override
​public void ​processingFinished() { ​//part of AudioProcessor object, doesn't do anything here
// TODO Auto­generated method stub
}
​@Override
​public boolean ​process(AudioEvent audioEvent) {
​float​[] audioFloatBuffer = audioEvent.getFloatBuffer();
​for ​(​int ​i = audioEvent.getOverlap(); i < audioFloatBuffer.​length​; i++) {
​//shift the in array
System.​arraycopy(​in​, ​0​, ​in​, ​1​, ​in​.​length ​­ ​1​);
​in​[​0​] = audioFloatBuffer[i];
​//calculate y based on a and b coefficients
//and in and out.
float ​y = ​0​;
​for​(​int ​j = ​0 ​; j < ​a​.​length ​; j++){
y += ​a​[j] * ​in​[j];
}
​for​(​int ​j = ​0 ​; j < ​b​.​length ​; j++){
y += ​b​[j] * ​out​[j];
}
​//shift the out array
System.​arraycopy(​out​, ​0​, ​out​, ​1​, ​out​.​length ​­ ​1​);
​out​[​0​] = y;
audioFloatBuffer[i] = y;
}
​volume​=audioEvent.​calculateRMS(audioFloatBuffer);
​if​(​volume​>​highVolume​)
​highVolume​=​volume​;
runOnUiThread(​new ​Runnable() {
​public void ​run() {
​ratio1 ​= Math.​pow((​volume ​/ ​highVolume​),​powerAdjust​);
​int ​ratio2 = (​int​)(​ratio1​*​255​);
​double ​color = Color.​rgb(ratio2,​0​,​255​­ratio2);
setActivityBackgroundColor((​int​)color);
}

10. });
​return true​;
}
};
​public void ​screenTapped(View view) {
​highVolume​=​HIGH​;
​volume​=​0​;
​ratio1​=​0​;
}
​public void ​setActivityBackgroundColor(​int ​color) {
View view = ​this​.getWindow().getDecorView();
view.setBackgroundColor(color);
}
​@Override
​public void ​onBackPressed() {
System.​exit(​0​);
​super​.onBackPressed();
}
}