1. Original Article
Motion-Based Angiogenesis Analysis: A Simple Method
to Quantify Blood Vessel Growth
Edmund Y. Tong,1
Geoffrey C. Collins,2
April E. Greene-Colozzi,1
Julia L. Chen,1
Philip D. Manos,1
Kyle M. Judkins,1
Joseph A. Lee,1
Michael J. Ophir,1
Farrah M. Laliberte,1
and Timothy J. Levesque1
Abstract
Existing methods to quantify angiogenesis range from image analysis of photographs to fluorescent microscopy.
These methods are often time consuming and costly; they also may not detect capillaries if they are indistinct
from the background of the image. We have developed a simple method based on the motion of blood to create
an image that reveals the entire angiogenic vasculature. Two image analysis software programs were used
separately to demonstrate the method. Using either ImageJ or Environment for Visualizing Images, we analyzed
a video clip of regenerated tissue from the partially amputated caudal fin of a zebrafish (Danio rerio). The
deviations among the frames in the video stack were calculated to reveal pixels where motion has occurred. The
resulting image highlighted all vessels through which blood flowed and allowed for automatic quantification of
the newly developed vasculature. Using this method, we quantified the angiogenic action of basic fibroblast
growth factor and vascular endothelial growth factor, as well as suppression of angiogenesis by an inhibitor. In a
preliminary study, we also found that it could be used to trace the developing vasculature in zebrafish embryos.
Thus, motion-based angiogenesis analysis may provide an easy and accurate quantification of angiogenesis.
Introduction
There are a wide variety of techniques used in the as-
sessment of angiogenic growth, which can be grouped
into in vitro and in vivo methods. In vitro studies are commonly
used to study cell migration, proliferation, and tube formation
in isolated endothelial cells. Moreover, in vitro studies can also
be conducted using the aortic ring and chick aortic arch as-
says. While in vitro tests are efficient and fairly consistent, they
do not account for biological interactions normally present in
a living organism.1
Consequently, in vitro results usually re-
quire confirmation from an in vivo test.
In vivo methods can be further categorized into a number of
assays; three of the most common include the chick chorio-
allantoic membrane (CAM), the corneal angiogenesis assay,
and, more recently, the zebrafish model. The CAM assay has
been the primary approach for studying the in vivo angiogenic
response because it is relatively simple, materials are easily
accessible, and individual CAMs can be used for multiple
tests.
Zebrafish are able to regenerate their caudal fin, and unlike
other vertebrate models, the stimulation or inhibition of an-
giogenesis during wound healing in zebrafish can be easily
investigated. In addition, the tissues and vessels of the tail fin
are particularly well defined, and the organization of the
vasculature is very similar to that of other vertebrates.2
In-
itially, the zebrafish embryo was used in angiogenic studies
because of its highly regular vasculature pattern and the rel-
ative ease in observing blood vessel growth.3
However, the
adult zebrafish has become an alternative model particularly
for studying caudal fin regeneration, as the physiological re-
sponse in an embryo may be quite different than that of the
fully developed adult organism.2
In recent years, image analysis has become a fairly new and
powerful tool for extracting quantitative measurements from
images in biomedical research. In this report we describe a
motion-based angiogenesis analysis (MBAA), which uses
blood movement to map out blood vessels and quantify an-
giogenesis in zebrafish. We applied basic fibroblast growth
factor (bFGF) and vascular endothelial growth factor (VEGF)
to simulate angiogenesis, as well as a down-stream VEGF in-
hibitor, 2-(2-Amino-3-methoxyphenyl)-4H-1-benzopyran-4-one
(PD 098,059), to inhibit angiogenesis.4–6
Post-acquisition anal-
ysis was first performed using Environment for Visualizing
Images 3.6 (ENVI 3.6), a software normally used for analyzing
satellite remote sensing data. We then used a public domain
image software, ImageJ, since it has become a more commonly
supported image processing software within the scientific
Departments of 1
Biology and 2
Physics, Wheaton College, Norton, Massachusetts.
ZEBRAFISH
Volume 6, Number 3, 2009
ª Mary Ann Liebert, Inc.
DOI: 10.1089=zeb.2008.0554
239

2. community.7
As shown in the Results and Discussion sections,
similar results were obtained using ImageJ and ENVI.
The MBAA method offers certain advantages over existing
methods. In most in vivo angiogenesis assays, an angiograph
is taken after the injection of a fluorescent agent or dye into the
bloodstream of the experimental specimen before image
analysis. More recently, investigators started to use transgenic
zebrafish, which express enhanced green fluorescence protein
in endothelial cells during vascular development to study an-
giogenesis. This method is successful, but special transgenic
zebrafish, such as fli:enhanced green fluorescence protein have
to be used.8
Using MBAA, an image revealing any blood cell
motion is produced without the administration of any agents to
illuminate the vessels; in other words, a few red blood cells
moving through a capillary allow the vessel to appear in the
final image. The likelihood of obtaining an image of the entire
regenerated vasculature in a given region for automatic quan-
tification is much greater. Specific to investigations using zeb-
rafish, the presence of pigments would not interfere with the
analysis using MBAA. Consequently, there is no need to use
the genetically altered transparent zebrafish or pretreatment
with chemicals to prevent pigment development.
Applying the same basic principle but using custom-made
Analytical Language for Images programs, Schwerte and
Pelster (2000) measured erythrocyte velocity, vessel diameter,
blood distribution, and the development of peripheral vascular
system in zebrafish embryos.9
Particle Image Velocimetry
(PIV) is another technique that has been employed in analyzing
blood flow velocity by investigators.10
The technique involves
capturing sequential digital images that are taken from blood
flowing through vessels that have been traced with markers
on the timing scale of nanoseconds. A laser light source and a
special camera are fitted to a fluorescent microscope allowing
for computer analysis of the velocity of blood flow through the
use of a timing system. There is a similarity between PIV and
MBAA in utilizing sequential images of moving blood cells or
other particles as a means to measure blood flow velocity in the
former and blood vessel growth in the latter; however, PIV is
too expensive a system to be modified to quantify angiogenesis.
Materials and Methods
Video acquisition
Zebrafish (purchased from Aquatica Tropicals, Plant City,
FL) were anesthetized in 100 mL water containing 200 mg=mL
Tricaine (Sigma, St. Louis, MO). Using a Nikon C-W10xA=22
dissecting microscope, the three most ventral bones from the
caudal fin were amputated with a razor blade anterior to the
branching point. After recovering from anesthesia in fresh
water, control fish were placed in fresh water tanks, while the
experimental fish were placed into a solution of bFGF
(8 mg=mL). After an exposure time of 1 h, the experimental fish
were placed into fish tanks for 3 days. A second experiment
was carried out in the same manner, using VEGF (12.5 ng)
instead of bFGF. In this experiment, another group of zebra-
fish was pretreated postamputation with the inhibitor PD
098,059 for 1 h before the application of VEGF. Using a Nikon
C-W10xA=22 dissecting microscope, the three most ventral
bones from the caudal fin were amputated with a razor blade
anterior to the branching point. After recovering from anes-
thesia in fresh water, control fish were placed in fresh water
tanks, while the experimental fish were placed into a solution
FIG. 1. Representative images of the experimental and
control fish (oriented with anterior to the left) after each step
of MBAA using ImageJ software. Micrographs of lower
ventral region of zebrafish caudal fin of experimental (A)
exposed to bFGF (8 mg=mL) imaged 3 days postamputation
(3 dpa) and control (B) exposed to no angiogenic stimulator.
(C) and (D) Eight-bit grayscale images of (A) and (B) after
loading video clip into ImageJ software. (E) and (F) Com-
posite images of (C) and (D) that combined the slices of the
video clip revealing only the vascular bed in white due to the
movement of blood cells. (G) and (H) Images after threshold
was applied to eliminate detection noise. (I) and (J) Final
images of (G) and (H) with only area of new vascular growth
outlined. Area of the new vasculature was quantified and
reported in pixels2
(control: 39,072; experimental: 85,649).
Point of amputation has been indicated by orange arrows.
MBAA, motion-based angiogenesis analysis; bFGF, basic fi-
broblast growth factor. (To view color image, go to www.
liebertonline.com).
240 TONG ET AL.

3. of PD 098,059 (20 mM) for 1 h. The fish were then transferred to
a solution of VEGF (12.5 ng) for an additional hour, and
subsequently placed into fish tanks for 3 days.
To analyze the amount of vessel growth 3 days postam-
putation (dpa), each fish was first anesthetized in 100 mL
water containing 200 mg=mL Tricaine (Sigma). The angiogenic
response was viewed at 40Âusing a Nikon Eclipse E200 mi-
croscope. When blood flow was stabilized after the initial ef-
fect of anesthesia, a Sony DFW–x700 camera was used in
conjunction with BTV Pro software to record a 16 bit short
video (1024Â768 pixels) lasting 2–4 s.
Analysis using ImageJ
The video clip was exported into ImageJ as a ‘‘Plugin.’’ The
video clip was converted into image stacks (slices) resulting in
an 8-bit grayscale image. The slices were merged by means
of standard deviation (SD). The best merged image was
achieved with stack sizes ranging from 5 to 10 slices. This range
provided the most optimized images that had minimal back-
ground noise, without the loss of image detail. Consequently, a
composite image of the vascular bed in white was produced
after the blood vessels were illuminated due to the movement
of blood cells, while the background remained black. A
threshold was then applied to eliminate background noise
caused by either fish movements or debris. Threshold values
were determined using the following parameters obtained
from the histogram of each merged image: mean (XX), maxi-
mum, and SD of pixel densities within the area of regenera-
tion. The maximum setting was kept the same, while the
minimum was set to a formula (1.5ÂSD) þ XX, which excluded
any pixel value below this threshold from quantification. This
formula is specific for the settings and conditions within our
lab, and values may differ depending on experimental pro-
tocol and image acquisition=processing settings. The area,
which contained the new vascular growth from the point of
amputation, was then outlined with the polygon selection
tool. Finally, the total area of the new vasculature was quan-
tified and reported in pixels. Although ImageJ could be used
to convert pixels to micrometers, we present our data in this
report in pixels. Other possible improvements to the threshold
processing are addressed in the Results and Discussion sec-
tions.
Analysis using ENVI 3.6
Using ENVI 3.6 the video clip was exported at a rate of
29.97 frames per second as TIFF files. The TIFF files were
subsequently imported to ENVI. The number of TIFF files
depended on the length of the video. A sequence of 20
colored-frame TIFF files was saved as ENVI files. Only the
red spectrum (channel) was selected, creating 8-bit grayscale
images. These frames were then merged based on the SD of
the pixel densities between the frames. This resulted in a com-
posite image that illuminated the moving blood through all
blood vessels. Once a threshold was applied and standard-
ized for each image to eliminate background noise, the area of
new vascular growth from the amputation point was out-
lined. Measurements of mean (XX) and SD of pixel values were
taken within the area of regeneration. A value derived from
the addition of one SD to the mean (d þ XX) was used as the
minimum threshold value. After the threshold was applied,
FIG. 2. Representative images of the experimental and
control fish (oriented with anterior to the left) after each step
of MBAA using Environment for Visualizing Images (ENVI)
3.6. Lower ventral region of zebrafish caudal fin of experi-
mental (A) exposed to bFGF (8 mg=mL) imaged 3 days
postamputation (3 dpa) and control (B) exposed to no an-
giogenic stimulator. (C) and (D) Grayscale images of (A) and
(B), respectively, after saving as ENVI files with only red
spectrum. (E) and (F) Composite images, revealing only the
vascular bed in white due to movement of blood, summing
the standard deviation of pixel densities of (C) and (D). (G)
and (H) Images after threshold was applied to eliminate
detection noise (red). (I) and (J) Final images highlighting the
newly developed blood vessels (blue) after selecting area of
new vasculature from point of amputation. Area of the new
vasculature was quantified and reported in pixels2
(control:
5613; experimental: 12,209). Point of amputation has been
indicated by orange arrows. (To view color image, go to
www.liebertonline.com).
MOTION-BASED ANGIOGENIC ANALYSIS 241

4. only the pixel values above the threshold range were high-
lighted. Consequently, the area of new blood vessel growth
was selected and pseudocolored blue. This area of new vas-
culature was then quantified and reported in pixels2
.
Results
The new blood vessels in the regenerated caudal fin are
scarcely visible in the initial images (Figs. 1A, B and 2A, B)
before any image processing was applied. In contrast, one is
able to highlight the vasculature for better observation after
applying a threshold to eliminate background=detection
noise (Figs. 1E, F and 2E, F). All the new blood vessels, in-
cluding those that are not clearly visible in the initial im-
age, appear vividly after processing a video of the specimen
through ImageJ (Fig. 1I, J) or ENVI (Fig. 2I, J). All blood vessels
are illuminated in white and can easily be distinguished from
all other tissue in the caudal fin of the zebrafish, which ap-
pears black. Further, the area of angiogenesis from the point of
amputation was outlined using ImageJ (Fig. 1I, J) or high-
lighted in blue using ENVI (Fig. 2I, J).
In both ImageJ and ENVI, the area of new vasculature was
expressed in pixels2
for a more quantitative comparison. Using
ImageJ, the area in the control was 5613 pixels2
(Fig. 1I), while
that of the experimental exposed to bFGF was 12,209 pixels2
(Fig. 1J). Using ENVI, the area of new vasculature in the same
control was 39,072 pixels2
(Fig. 2I), and that of the same ex-
perimental was 85,649 pixels2
(Fig. 2J). Although the numerical
values between the two programs are different, the ratio of
the number of pixels is the same. In ImageJ, the pixel area of
the experimental is 2.19 times that of the control; in ENVI, the
number of pixels of the experimental is 2.17 times the control.
The ratio between these two programs is 1:1, suggesting that
the process is consistent between ImageJ and ENVI. The dif-
ference in the absolute pixel values is due to a discrepancy in
the number of pixels per unit area between the two programs.
Figure 3 shows our preliminary study using MBAA to trace
the vasculature in the zebrafish embryo. The blood vessels in
the initial images in the head region (A) and the tail region
(B) of a 72 hours postfertilization unanesthetized zebrafish
embryo were highlighted in red after MBAA (C and D, re-
spectively). The vasculature in the head region is not visible
because the tissue in this region is too thick.
Using ENVI as the software in MBAA to quantify angio-
genesis, the proangiogenic effects of bFGF were used as one
example to demonstrate the effectiveness of this new method.
As seen in Figure 4, there was a statistically significant in-
crease in average area of new vascular growth in the experi-
mental group compared to the control group. Using ImageJ as
the software in MBAA to quantify angiogenesis, the proan-
giogenic effects of VEGF were used as another example. In
addition, inhibition of angiogenesis by PD 098,059 pretreat-
ment before VEGF application was also demonstrated. As
shown in Figure 5, there is a statistically significant difference
FIG. 3. Micrographs of the head region (A) and the tail
region (B) of a 72 hours postfertilization zebrafish embryo
(anterior to the left). (C, D) Final images of the same regions
with the vasculature highlighted in red after MBAA analysis
using ImageJ.
FIG. 4. After exposing zebrafish to a solution of bFGF
(8 mg=mL) for 1 h, a statistically significant increase in area of
new vascular growth during wound healing angiogenesis in
the caudal fin was observed ( p 0.05, two-sample T-test).
Image analysis was performed using ENVI. Statistical anal-
ysis was performed using two-sample T-test, and data are
presented as mean number of pixels2
with standard error
(SE). The experimental group demonstrated a significantly
larger area (39,417 pixels2
; n ¼ 16), compared to the aver-
age area of the control group (9259 pixels2
; n ¼ 12).
FIG. 5. After exposing zebrafish to a solution of vascular
endothelial growth factor (VEGF; 12.5 ng=mL) for 1 h, a sta-
tistically significant increase in area of new vascular growth
during wound healing angiogenesis in the caudal fin was
observed ( p 0.05, two-sample T-test). The angiogenic effect
of VEGF is inhibited by 1 h pretreatment with the inhibitor
PD098,059 (10 mM). Analysis was performed using ImageJ,
and data are presented as number of pixels2
. The VEGF
group demonstrated a larger area (49,516 pixels2
) compared
to the average area of the water control group (25,924
pixels2
), while the average of the inhibitor pretreatment
group was lower than both groups (43,297 pixels2
).
242 TONG ET AL.

5. between the angiogenic effects of the control and VEGF-
treated zebrafish. Due to the small sample size of the PD
098,059 pretreated zebrafish, there is no statistically signifi-
cant difference between this group and the other two groups.
Discussion
Although ENVI was first used in the development of the
MBAA method, ImageJ was found to be a more efficient
video=image-processing software. In addition, ImageJ is free
and widely used. Thus, it will be the focus of the following
discussion. If the composite image does not provide enough
contrast, it is most likely due to an inadequate recording of the
original videos. This is possibly caused by fish movement or
stage disturbance during the video recording. It may also be
due to video processing or inconsistent microscope settings.
These disturbances do not necessarily produce an image with
false-positives (inaccurate blood vessel detection), but in-
stead, simply do not produce any composite image that can be
analyzed. The noise trouble can sometimes be managed by
altering the number of frames used to create the composite
image. By combining less stacks (5–10 frames) or using a
different sequence of stacks within the video that are of higher
quality (periods where there may be no disturbances), the
composite image could be made clearer and further proces-
sing becomes possible. However, a minimum of five frames
are necessary to show the difference among the frames, and to
gain a composite image with enough contrast to effectively
highlight the blood vessels.
To obtain accurate and consistent data from the videos, it is
important to standardize every aspect of the recording and to
use a constant method to establish the threshold. It is most
important to maintain a consistent light intensity while avoid-
ing any glare. Any moving water or fish movement may alter
the light intensity or cause glare, which interferes with the im-
age processing and affects quantification. In addition to what is
mentioned in the Materials and Methods section for this study,
there are other ways to better standardize the imaging process.
One possible alternative would be to first create a standard
histogram (e.g., designating the mean to be 128 and the SD to be
10 on the histogram). The greater the SD between pixel values
among frames, the more distinct the blood vessels will appear in
the composite image. All of the images would then be stretched
to fit the standard histogram by adjusting the average bright-
ness (mean) and contrast (SD). By normalizing the brightness
among frames, the threshold can be set at a constant value that
best highlights the blood vessels, such as SD ¼ 3 or SD ¼ 4.
Although measuring the total area of a particular vascular
bed was the only parameter quantified in this study, it is only
one aspect of the image analysis that may be performed. Many
other parameters can be calculated in both ImageJ and ENVI by
uploading any specific macro program to expand the versa-
tility of the MBAA method. For example, vessel length and
number of branching points could be measured. In our pre-
liminary study using the zebrafish embryo, we have demon-
strated that the method can be used to study the vasculature
during embryonic development. However, one limitation of
this method also became evident. When the sample is too thick,
MBAA cannot be used to estimate blood vessel growth.
Nevertheless, in two major angiogenesis models, the CAM
model and the cornea model, the thickness of the sample
should not be a problem. We plan to expand the application of
this simple method to the chick CAM model and the rabbit
cornea model to study angiogenesis. Further, although ImageJ
and ENVI software were used in this study, any image-
processing software would most likely be able to perform the
analysis. The precise protocols of the software may vary
slightly, but the basic approach remains the same. Finally, the
numerous steps described in image processing could be auto-
mated with an appropriate Plugin or Macro program.
Acknowledgments
We would like to thank the following persons for their
support and assistance throughout this project: Erin Post and
Beth Amaral for animal care, and Dr. Chris Kalberg for tech-
nical advising. The research was supported by Wheaton
Foundation Grant, Hood Faculty=Student Summer Research
Fellowship, Melon Faculty=Student Summer Research Fel-
lowship, Wheaton Trustee Scholarship, Wheaton Community
Scholarship, and Wheaton Research Partnership.
Disclosure Statement
No competing financial interests exist.
References
1. Auerbach R, Lewis R, Shinners B, Kubai L, Akhtar N. An-
giogenesis assays. Clin Chem 2003;49:32–40.
2. Bayliss P, Bellavance K, Whitehead G, Abrams J, Aegerter S,
Robbins H, et al. Chemical modulation of receptor signalling
inhibits regenerative angiogenesis in adult zebrafish. Nat
Chem Biol 2006;2:265–273.
3. Childs S, Chen J, Garrity D, Fishman M. Patterning of an-
giogenesis in the zebrafish embryo. Development 2002;129:
973–982.
4. Alessi D, Cuenda A, Cohen P, Dudley D, Saltiel A. PD
098,059 is a specific inhibitor of the activation of mitogen-
activated protein kinase in vitro and in vivo. J Biol Chem
1995;46:27489–27494.
5. Folkman J. Tumor Angiogenesis. In: Vande Woude G, Klein
G, Weinhouse S (eds) Advances in Cancer Research. Aca-
demic Press, New York, NY, 1985, pp. 175–199.
6. Ferrara N, Davis-Smyth T. The biology of vascular endo-
thelial growth factor. Endocr Rev 1997;18:4–25.
7. Rasband W. ImageJ. U.S. National Institutes of Health,
Bethesda, Maryland 1997–2008. http:==rsh.info.nih.gove=ij=.
8. Lawson N, Weinstein B. In vivo imaging of embryonic vas-
cular development using transgenic zebrafish. Dev Biol
2002;248:307–318.
9. Schwerte T, Pelster B. Digital motion analysis as a tool for
analyzing the shape and performance of the circulatory sys-
tem in transparent animals. J Exp Biol 2000;203:1659–1669.
10. Vennemann P, Kiger K, Lindken R, Groenendijk C,
Stekelenburg-de Vos S, ten Hagen T, et al. In vivo micro
particle image velocimetry measurements of blood-plasma
in the embryonic avian heart. J Biomech 2006;39:1191–1200.
Address correspondence to:
Edmund Y. Tong, Ph.D.
Department of Biology
Wheaton College
26 East Main St.
Norton, MA 02766
E-mail: etong@wheatoncollege.edu
MOTION-BASED ANGIOGENIC ANALYSIS 243