Monte Carlo derived absorbed fractions for a voxelized model of
Oncorhynchus mykiss, a rainbow trout
Ruedig, E., Caffrey, E., Hess, C., & Higley, K. (2014). Monte Carlo derived
absorbed fractions for a voxelized model of Oncorhynchus mykiss, a rainbow
trout. Radiation and Environmental Biophysics, 53(3), 581-587.
doi:10.1007/s00411-014-0546-5
10.1007/s00411-014-0546-5
Springer
Accepted Manuscript
http://cdss.library.oregonstate.edu/sa-termsofuse
Original Paper
Elizabeth Ruedig, Emily Caffrey, Catherine Hess, Kathryn Higley
Monte Carlo Derived Absorbed Fractions for a Voxelized Model of
Oncorhynchus Mykiss, a Rainbow Trout
Elizabeth Ruedig (!)
Colorado State University, Department of Environmental & Radiological Health
Sciences, Fort Collins, CO 80523, USA
Email: elizarue@rams.colostate.edu
Tel: +19709417575
FAX: +19704910623
Emily Caffrey, Catherine Hess, Kathryn Higley
Oregon State University, Department of Nuclear Engineering & Radiation Health
Physics, Corvallis, OR 97331, USA
1
Abstract
Simple, ellipsoidal geometries have long been the standard for estimating
radiation dose rates in non-human biota (NHB). With the introduction of a regulatory
protection standard that emphasizes protection of NHB as its own endpoint, there has
been interest in improved models for the calculation of dose rates in NHB. Here we
describe the creation of a voxelized model for a rainbow trout (Oncorhynchus mykiss), a
freshwater aquatic salmonid. Absorbed fractions (AFs) for both photon and electron
sources were tabulated at electron energies of 0.1, 0.2, 0.4, 0.5, 0.7, 1.0, 1.5, 2.0, and 4.0
MeV and photon energies of 0.01, 0.015, 0.02, 0.03, 0.05, 0.1, 0.2, 0.5, 1.0, 1.5, 2.0, and
4.0 MeV. A representative set of the data is made available in this publication; the entire
set of absorbed fractions is available as electronic supplementary materials. These results
are consistent with previous voxelized models, and reinforce the well-understood
relationship between the AF and the target’s mass and location, as well as the energy of
the incident radiation.
Keywords: voxel phantom, dosimetry, non-human biota, monte carlo
2
Introduction
In 2008, recommendations of the International Commission on Radiological Protection
(ICRP) shifted toward a new radioprotection paradigm. In contrast to the historical
paradigm, where wildlife was considered to be sufficiently protected provided that man
was adequately protected, the ICRP recommended in 2008 that the protection of wildlife
populations be considered as a separate protection endpoint (ICRP 2008). This decision
has raised numerous challenges for the scientific community as they seek to deepen their
understanding of the impacts of exposure to ionizing radiation on non-human biota
(NHB).
Evaluating radiation dose to NHB is of significant concern under this new protection
paradigm. Since dose-response relationships may not be well understood, particularly for
wild populations exposed chronically to low dose rates, it is imperative to make an
accurate determination of the dose rate to which NHB are exposed. If, during the course
of a dose-effects study, dose rates are improperly estimated, there are two possibilities:
effects may be attributed to dose rates that are too low, or to dose rates that are too high.
If the estimated dose rate is erroneously low, regulatory bodies may recommend costly
remediation or re-engineering due to the literature demonstrating dose-effects at an
inaccurately calculated, low dose rate. Conversely, if dose calculation methodologies
provide estimates of dose rates that are erroneously high, those bodies may consider such
dose rates as sufficiently protective limits – based on peer-reviewed evidence of no effect
– when in fact environmental impacts may be occurring.
The ICRP (ICRP, 2008) has provided dose conversion factors (DCFs) for a set of
reference animals and plants (RAPs) as a part of its recommendations for the protection
of NHB. These DCFs were derived using homogeneously distributed radionuclide
sources in ellipsoidal geometries, considered to be representative of an organism’s body.
However, there is significant discussion (ICRP 2008; Ruedig 2013; Petoussi-Henns et al.
2007; Zaidi and Xu 2007; Petoussi-Henss et al. 2007; Gómez Ros et al. 2008) as to how
well these simplistic models represent those cases where anatomy decidedly deviates
3
from an ellipsoidal shape, and/or radionuclides partition heterogeneously. In such cases,
voxelized modeling may provide a more accurate methodology by which to estimate
radiation dose rate. Voxelized models rely upon highly realistic phantoms that are
constructed with the aid of diagnostic imaging data (e.g. computed tomography (CT) or
magnetic resonance imaging (MRI)); these phantoms are subsequently used as the input
geometry for Monte Carlo particle transport simulations. Voxel phantoms have often
been employed to date in both human and veterinary dose estimations; less often, they
have been used to calculate dose rates in wildlife. This study will present novel
dosimetric data for a rainbow trout - a fish that is similar to the ICRP’s reference trout derived via Monte Carlo methods from a voxelized phantom.
Materials and methods
A male rainbow trout (Oncorhynchus mykiss) was obtained from Rainbow Trout Farm in
Sandy, Oregon, USA. The individual under study was 35.5 cm from the tip of the head to
the end of the tail fin. It measured approximately 40 mm at the thickest point, and 80 mm
from the bottom of the belly to the dorsal fin. Its mass was 658 g. The trout was frozen
immediately after catch, and thawed three days later, just prior to imaging.
Voxelized phantoms are created using anatomic data obtained by either computed
tomography (CT) or magnetic resonance imaging (MRI); the present study employed CT.
Computed tomography scanning was performed at the Oregon State University School of
Veterinary Medicine on a Toshiba Aquillion 64 slice machine. The images were acquired
helically, at 120 kVp and 50 mA, and with a pitch of 0.5. Slice thickness was 1 mm; in
total 362 slices were utilized in making the model. The resulting pixel matrix was 512
rows by 512 columns. Figure 1 depicts a three-dimensional rendering of the fish, as
reconstructed from the CT data using OxiriX1 medical imaging software (Rosset et al.
2004).
1
OsiriX Imaging Software. http://www.osirix-viewer.com/.
4
To create the realistic geometry used in a voxel phantom, individual organs are contoured
on each slice of the CT image in a process called segmentation. The segmentation
process is analogous to the organ contouring performed routinely for the purposes of
human and veterinary radiation therapy dosimetry. In this study, 3D Doctor2, a vectorbased imaging, modeling, and computation program, was used to perform organ
segmentation. Segments were hand-drawn after attempts at auto segmentation resulted in
highly irregular surfaces. Not all organs were individually contoured due to difficulties
distinguishing minute organs, or those organs located most proximally. The contents of
each segment are described in Table 1. Figure 2 illustrates segmentation for an axial slice
of the CT image, while Figure 3 depicts a complete rendering of the voxel phantom
created using 3D Doctor. Muscle and skin tissues have been suppressed in this rendering
to depict internal structures. In this figure the trout is shown without the muscle tissue so
that individual segments can be visualized.
The segment data are exported from 3D Doctor as a boundary file, which specifies start
and stop points of each segmented organ on a particular slice of the CT image. This file is
imported into Voxelizer3, which outputs a Monte Carlo N-Particle (MCNP) transport
code lattice geometry (Kramer et al. 2009). Once converted to MCNP format, each voxel
is filled with the appropriate material. Due to the dearth of information concerning
elemental composition of trout tissues, human tissue compositions were used where
appropriate, consistent with previous models (Larsson 2008; Ulanovasky and Pröhl 2008;
Ulanovasky and Pröhl 2006; ICRP 2008). Table 2 describes the tissue composition for
each segment, while Table 3 describes the physical properties of each segment. The
composition of the swim bladder was a special case due to the lack of an analogous organ
in the human body. Saunders (1953) investigated the fractional composition of gases in
the swim bladders of lake trout (Salvelinus namaycush). His data for swim bladder gas
composition in lake trout from 0-5m was used to describe the elemental composition and
density of the swim bladder segment.
2 Able Software Corp. 5 Appletree Lane, Lexington MA 02420. http://www. ablesw.com/3d-doctor/index.html.
3 Human
Monitoring Laboratory, National Internal Radiation Assessment Section, Radiation Protection Bureau, 775
Brookfield Road PL6302D, Ottawa, Ontario, K1A 1C1. 5
After completing the material specifications, particle transport simulations were
performed using MCNPX4. Several simulations were run, each with a different source
segment, until energy deposition data had been collected for all segments via the *f8
(energy deposition) tally. Particle transport was run for electron energies of 0.1, 0.2, 0.4,
0.5, 0.7, 1.0, 1.5, 2.0, and 4.0 MeV and photon energies of 0.01, 0.015, 0.02, 0.03, 0.05,
0.1, 0.2, 0.5, 1.0, 1.5, 2.0, and 4.0 MeV. The particle source was homogenously
distributed throughout each source segment. Due to the small size of some of the source
organs, source sampling efficiency was decreased from the default of 1% to 0.0001%, in
line with previously published voxel models (Caffrey and Higley 2013). Absorbed
fraction (AF) is the average fraction of a particle’s energy deposited in a target tissue.
AFs were tabulated for each source:target segment combination.
Results
Figures 4 and 5 graphically present the AF results for electron and photon sources,
respectively, in the liver segment. Extensive tabular data that details AF information as a
function of source, target, energy, and radiation type is available as electronic
supplementary material. Table 4 presents AFs for an electron source located in the liver
at seven electron energies, and is intended to illustrate the data available online. MCNP
presents error as the coefficient of variance (COV), a measure of the precision of the
particle transport calculation. When the COV exceeded 10%, the resultant AF was
excluded, and a dash entered into the table. For AFs between 5% and 10%, the AF is
presented underlined as a cautionary note.
Discussion
The trends in figures 4 and 5 are representative of the overall Monte Carlo data. Figure 4
is representative of electron data: it depicts the absorbed fraction of electron energy for a
4
Los Alamos National Laboratory, Los Alamos, NM https://mcnpx.lanl.gov/
6
monoenergetic source located in the voxel phantom’s liver segment. In human dosimetry,
it is commonly assumed that all electron energies are absorbed in the source organ (that
the AF=1). However, for many biota, organs are substantially smaller and it is therefore
important to consider the ‘escape’ of higher energy electrons from the source tissue. At
the lowest energies, very little or none of the particle’s energy is deposited outside of the
source segment. But as the electron’s energy increases, more and more of the electron’s
energy is deposited in distally located segments, asymptotically approaching a maximum
value. Electron energy deposition is relatively linear, and this maximum deposition is
heavily dependent upon the physical size of the target segment, provided that tissue
densities are similar, as is typically the case for soft tissues. It is of note that those curves
with the greatest deviation are the two segments whose density deviates greatly from that
of water: the skeleton and the swim bladder. Generally, the extent of escape will be a
function of segment size and geometry, as well as density and effective atomic charge
(Z).
Figure 5 is representative of photon data, and depicts the absorbed fraction of photon
energy for a monoenergetic source located within the voxel phantom’s liver segment.
Generally, source particles are attenuated in the source segment at the lowest energies,
although there is some deposition in adjacent segments even at low energies due to the
more penetrating nature of photons, as compared to electrons. The absorbed fraction
reaches a maximum deposition for each target segment, each at a slightly different
energy. Peak deposition is principally controlled by the distance from the source segment
to the target segment: charged particle build-up reaches a maximum for a particular
energy at a particular depth. Thus, deposition peaks at lower energies in those segments
closest to the liver compared to segments further from the liver. The total fraction of
energy deposited decreases past this maximum as the photons become more penetrating
and a smaller fraction of energy is deposited in those tissues closest to the source
segment.
Figure 6 depicts the self-absorbed fraction for photons in each segment in the voxelized
trout model (i.e., the target and source segment are the same). This plot more closely
follows typical patterns for absorbed fractions. The self-absorbed fraction decreases in
7
each segment up to about 0.1 MeV. At these low energies, photoelectric interactions
dominate, and the majority of the photon’s energy remains in the source segment. At
increasingly higher photon energies, Compton scattering becomes the dominant process,
and the AF increases slightly as Compton electrons are attenuated in the source segment.
At approximately 1 MeV, the AF begins to decrease again, as pair-production gammas
escape the target segment.
The voxel model presented in this paper is a significant improvement upon existing
dosimetric models for freshwater fish. While voxelized model-derived AFs are available
for other species such as a crab (Caffrey and Higley 2013), a frog (Mohammadi et al.
2011), and a mouse (Mohammadi and Kinase 2011; Mohammadi and Kinase 2011), this
paper presents the first such results for a salmonid fish. Given the ubiquity of exposure
scenarios for fish, these results may be of considerable use to those engaged in
environmental radioprotection.
It is difficult to directly compare the dosimetric data derived from the voxel model
presented here to the existing dosimetric data from ICRP (ICRP 2008). Voxelized models
produce segment-specific dose rates in lieu of whole body dose rates, and can account for
inter-organ crossfire. Despite this difference, we compared dose conversion factors
(DCFs) derived from the dosimetric data presented here to DCFs calculated using the
approach of Gómez-Ros (Gómez-Ros et al. 2008), which has been endorsed by the ICRP
(ICRP 2008). Generally, there was similarity between the results of the two approaches,
with DCFs broadly agreeing to within a factor of two to three.
Typically, regulatory structures that aim to protect wildlife from the effects of ionizing
radiation focus on deterministic impacts (e.g., population decline) rather than stochastic
impacts (e.g., individual risk of cancer) (ICRP 2008). Voxelized models by their very
nature relate tissue concentrations of radionuclides to individual organ dose rates. The
compromise or failure of one or more organ systems is the most likely pathway by which
a deterministic impact may arise. Therefore, utilizing voxelized models to calculate
radiation dose rates in wildlife may be of importance when studying dose-effects
8
relationships in wild populations; the ICRP has previously proposed (ICRP 1959)
monitoring organ dose rates in humans to ensure the prevention of deterministic effects.
Conclusion
There are several areas of work from which would be of benefit for the development of
future voxel models and related dosimetry. The first is an improvement in the materials
specification for the voxelized phantom. Elemental composition for NHB tissues is
seldom available in the open literature, and utilizing organism-specific tissue composition
data, as opposed to human, greatly improves voxel phantom dosimetry. Secondly, it
would be of particular interest to apply this voxelized fish model to a real world situation
such as the Sellafield site, the Pacific Proving Grounds, or the Fukushima release of
March 2011. Using real tissue concentration data from these sites as the basis for a
comparison of voxel-model derived dose rates with those calculated via an ellipsoidal,
homogeneous model would provide an intriguing analysis of current dosimetric models
while also evaluating the costs and benefits of employing voxelized models to improve
the accuracy of dose calculations in wildlife.
Finally, an in-depth investigation into the utility of voxel models is warranted. There is
ongoing debate concerning the value of voxel models, particularly given the increased
cost of deriving dosimetric data from a voxelized phantom. However, it is difficult to
compare voxelized models directly to the ellipsoidal geometries used by ICRP. The
choice to use a voxel model may depend upon organism geometry, size, tissue
heterogeneity, the user’s needs, or other factors. Voxel models provide organ dose rates
for heterogeneous distributions of radionuclides and account for organ crossfire, which
may be of use for studies requiring very accurate dosimetry for the determination of
ecological dose-effects relationships. In some cases, the realism of voxel models may
prove advantageous during the social licensing process. The ICRP’s ellipsoidal models
and approximations for calculating organ dose rates are also excellent approaches, and
are especially useful for compliance-related activities where a conservative solution,
9
obtained without excessive cost, is the principal concern. It is difficult to say definitively
which methodology is superior in the absence of a robust study detailing the increases in
accuracy afforded by voxel models, as well as the costs associated with the derivation of
voxelized dosimetric data. As such, further investigation into these models is certainly
called
for.
10
References
Caffrey EA, Higley KA (2013) Creation of a voxel phantom of the ICRP reference crab,
J Environ Radioactiv 120:14-18..
Gómez-Ros JM, Pröhl G, Ulanovsky A, Lis M, (2008) Uncertainties of internal dose
assessment for animals and plants due to non-homogeneously distributed radionuclides. J
Environ Radioact 99: 1449-1455.
ICRP (1959) Permissible dose for internal radiation, ICRP Publication 2.
ICRP (2002) Basic Anatomical and Physiological Data for Use in Radiological
Protection Reference Values, ICRP Publication 89, Ann ICRP 32(3-4).
ICRP (2008) Environmental Protection - the Concept and Use of Reference Animals and
Plants. ICRP Publication 108, Ann ICRP 38(4-6).
ICRU (1989) Tissue Substitutes in Radiation Dosimetry and Measurement, ICRU
Publication 44.
Kramer GH, Capello K, Chiang A, Cardenas-Mendez E, Sabourin T (2010) Tools for
Creating and Manipulating Voxel Phantoms. Health Phys 98(3): 542–548.
Larsson CM (2008) An overview of the ERICA Integrated Approach to the assessment
and management of environmental risks from ionising contaminants. J Environ Radioact
99(9): 1364–1370.
Mohammadi A, Kinase S, Saito K (2011) Comparison of photon and electron absorbed
fractions in voxel-based and simplified phantoms for small animals. Nucl Sci Tech 2:
365-368.
Mohammadi A, Kinase S (2011) Electron absorbed fractions and S values in a voxelbased mouse phantom. Radioisot 60: 505-512.
11
Mohammadi A, Kinase S (2011) Monte carlo simulations of photon specific absorbed
fractions in a mouse voxel phantom. J Nucl Sci Technol 1: 126-129.
Petoussi-Henss N, Zankl M, Hoeschen C, Nosske D (2007) Voxel or MIRD-type model:
a sensitivity study relevant to nuclear medicine. In R. M. and J. H. Nagel (ed) World
Congress on Medical Physics and Biomedical Engineering 2006. Springer, Berlin
Heidelberg, pp 2061–2064.
Rosset A, Spadola L, Ratib O (2004) OsiriX: An Open-Source Software for Navigating
in Multidimensional DICOM Images, J Digit Imaging 17(3): 205-216.
Ruedig, E (2013) Dose-effects relationships in non-human biota: development of field
sampling, dosimetric and analytic techniques through a case study of the aquatic snail
Campeloma decisum at Chalk River Laboratories, Dissertation, Oregon State University.
Saunders RL (1953) The swimbladder gas content of some freshwater fish with particular
reference to the physostomes. Can J Zool 31(6): 547-560.
Ulanovsky A, Pröhl G (2006) A practical method for assessment of dose conversion
coefficients for aquatic biota. Radiat Environ Biophys 45(3): 203–214.
Ulanovsky A, Pröhl G (2008) Tables of dose conversion coefficients for estimating
internal and external radiation exposures to terrestrial and aquatic biota. Radiat Environ
Biophys 47(2): 195–203.
Zaidi H, Xu XG (2007) Computational Anthropomorphic Models of the Human
Anatomy: The Path to Realistic Monte Carlo Modeling in Radiological Sciences. Annu
Rev Biomed Eng 9(1): 471-500.
12
Figure captions
Fig. 1 Three-dimensional rendering of trout created from CT data using OsiriX medical
imaging software
Fig 2 Segmentation of an axial slice in 3D Doctor. The white line at the bottom of the
image is the CT couch
Fig. 3 Segment rendering of voxelized trout model created using 3D Doctor
Fig.4 Electron absorbed fractions (AFs) for source located the in liver segment. Points
appearing at an AF value of 1x10-6 (black line) exceeded 10% COV and should not be
considered reliable
Fig. 5 Photon absorbed fractions (AFs) for source located in the liver segment. Points
appearing at an AF value of 1x10-6 (black line) exceeded 10% COV and should not be
considered reliable
Fig. 6 Photon self-absorbed fractions (i.e., target segment is also source segment), for all
segments
13
Fig. 1 Fig 2. 14
Fig. 3 15
Fig.4 1e+00
1e-01
1e-02
target
Absorbed Fraction
esophagus
heart
kidney
liver
1e-03
muscle
pyloric_caeca
rectum
skeleton
swim_bladder
1e-04
1e-05
1e-06
0.5
1.0
1.5
2.0
2.5
Energy (MeV)
3.0
3.5
4.0
16
Absorbed Fraction
Fig. 5 1e+00
1e-01
target
1e-02
brain
esophagus
eyes
heart
kidney
liver
1e-03
muscle
pyloric_caeca
rectum
skeleton
spleen
swim_bladder
1e-04
testes
1e-05
1e-06
0.01
0.05
0.10
0.50
Energy (MeV)
17
1.00
5.00
Self-absorbed fraction
0.1
1.0
0.01
0.10
Energy (MeV)
1.00
testes
spleen
skeleton
rectum
pyloric_caeca
muscle
liver
kidney
esophagus
segment
Fig. 6 18
Table 1 Description of anatomical features included in each segment
Segment Brain Esophagus Eyes Heart Kidney Liver Muscle Pyloric Caeca Rectum Skeleton Spleen Swim Bladder Organs Within Segment Brain Esophagus, Stomach Eyes Heart Kidney Liver, Gallbladder Skin, Scales, Fins, Gills, Pharynx, Thyroid, Spinal Cord, Nerves, Vasculature, Muscle, Fat, Connective Tissue Pyloric Caeca, Pancreas Rectum, Intestines Bone, Teeth Spleen Swim Bladder Table 2 Description of materials used to fill model segments.
Segment Reference Referenced Human organ Brain ICRU (1989) Brain, Grey/White Matter Esophagus ICRP (2002) Esophagus Eyes ICRP (2002) Eyes Heart ICRP (2002) Heart Kidneys ICRP (2002) Kidney Liver ICRP (2002) Liver Muscle & Soft Tissue ICRP (2002) Muscle Pyloric Caeca ICRP (2002) Stomach Rectum ICRP (2002) Large Intestine Skeleton ICRU (1989) Cortical Bone Spleen ICRP (2002) Spleen Testes ICRU (1989) Testes 19
Table 3 Trout tissue density, mass, and volume, as used in materials specification for
MCNP calculations
Trout Organ Volume or Tissue (cm3) Brain 0.566 0.5886 Density Used Reference (g/cm3) 1.04 ICRU (1989) Referenced Organ Brain Mass (g) Esophagus 13.751 14.3698 1.045 ICRP (2002) Alimentary System Eyes 1.936 2.0715 1.07 ICRU (1989) Eye Lens Heart 1.782 1.8354 1.03 ICRP (2002) Heart Kidneys 4.02 4.2210 1.05 ICRP (2002) Kidney Liver 7.145 Muscle & Soft 554.325 Tissue Pyloric Caeca 7.884 7.5737 1.06 ICRP (2002) Spleen 582.041 1.05 ICRU (1989) Muscle, Skeletal 8.238 1.045 ICRP (2002) Alimentary System Rectum 11.75 12.278 1.045 ICRP (2002) Alimentary System Skeleton 12.502 24.003 1.92 ICRU (1989) Bone, Cortical Spleen 0.512 0.54272 1.06 ICRP (2002) Swim Bladder 33.672 0.040440 0.001201 Saunders (1953) 0.372 0.38688 1.04 ICRU (1989) Spleen Lake Trout Swim Bladder (0-­‐16.5 Feet) Testis Testes Table 4 Absorbed fraction (AF) of electron energy for source in liver as a function of
energy. AFs whose coefficient of variance (COV) exceeded 10% appear as a dashed line
(-); those AFs with COVs between 5% and 10% are underlined as a cautionary note
Energy (MeV) 1.000 1.500 0.200 0.400 0.700 2.000 4.000 Muscle/Soft Tissue 9.62E-­‐03 2.66E-­‐02 4.88E-­‐02 6.43E-­‐02 8.51E-­‐02 1.03E-­‐01 1.74E-­‐01 Swim Bladder Skeleton Eyes Heart Liver Brain -­‐ 1.40E-­‐04 -­‐ 1.12E-­‐03 9.84E-­‐01 -­‐ -­‐ 2.97E-­‐04 -­‐ 3.13E-­‐03 9.56E-­‐01 -­‐ -­‐ 6.61E-­‐04 -­‐ 6.65E-­‐03 9.18E-­‐01 -­‐ -­‐ 1.18E-­‐03 -­‐ 1.03E-­‐02 8.89E-­‐01 -­‐ -­‐ 2.40E-­‐03 -­‐ 1.59E-­‐02 8.48E-­‐01 -­‐ 1.53E-­‐06 3.91E-­‐03 -­‐ 2.09E-­‐02 8.13E-­‐01 -­‐ 3.86E-­‐06 1.06E-­‐02 6.13E-­‐06 3.40E-­‐02 6.90E-­‐01 -­‐ Esophagus Rectum Spleen Testes Pyloric Caeca Kidney Total Body 1.90E-­‐03 1.47E-­‐03 -­‐ -­‐ 1.15E-­‐03 2.81E-­‐04 1.00E+00 5.31E-­‐03 4.13E-­‐03 -­‐ -­‐ 3.38E-­‐03 8.06E-­‐04 1.00E+00 1.02E-­‐02 7.41E-­‐03 -­‐ -­‐ 6.46E-­‐03 1.38E-­‐03 9.99E-­‐01 1.41E-­‐02 8.95E-­‐03 -­‐ -­‐ 9.22E-­‐03 1.62E-­‐03 9.99E-­‐01 1.97E-­‐02 1.04E-­‐02 -­‐ -­‐ 1.37E-­‐02 2.00E-­‐03 9.98E-­‐01 2.46E-­‐02 1.11E-­‐02 -­‐ -­‐ 1.77E-­‐02 2.14E-­‐03 9.97E-­‐01 4.08E-­‐02 1.21E-­‐02 3.81E-­‐06 -­‐ 2.90E-­‐02 1.97E-­‐03 9.92E-­‐01 Target 20