Reduced Dose of Proton CT Compared to X-Ray CT in Tissue-Density Variation Sensitivity

1
BNL Proton Radiography and Tomography
Reduced Dose of Proton CT Compared to X-Ray CT
in Tissue-Density Variation Sensitivity T.
Satogata, T. Bacarian, S. Peggs, A.G. Ruggiero,
and F.A. Dilmanian Brookhaven National
Laboratory, Upton USA
ABSTRACT Proton therapy has advantages over
conventional X-ray therapy in that it produces
tighter dose distributions around the tumor due
to the sharp range cutoff of the proton Bragg
peak. Because the dose distribution is highly
localized, high-precision treatment planning is
also required for proton therapy. Traditional
treatment tomography, X-ray computed tomography
(XRCT), is inadequate for this planning because
the proton stopping power mostly depends on the
electron density. A more natural approach is
low-dose proton computed tomography (PCT), where
the electron density in the patient or phantom is
mapped directly by a proton beam of higher energy
than the therapy energy. We report detailed
comparisons of PCT against XRCT based on our
Monte Carlo simulations and tomographic
reconstruction techniques. The preliminary
simulations of PCT and XRCT used pencil-beam
scanning through a 20-cm water cylinder with a
2-cm off-axis water cylinder of 1 higher
density. The PCT Monte Carlo simulations
included Bethe-Bloch energy loss and straggling,
multiple Coulomb scattering inelastic nuclear
collisions were neglected. The XRCT simulation
used attenuation calculations for each ray, and
added statistical noise to the resulting
projections according to the prescribed subject
absorbed dose of 4 cGy. Both the PCT and the
XRCT simulations used bow-tie water phantoms
that reduce peripheral dose in XRCT (when placed
between the incident beam and primary phantom),
and homogenized the proton tracks and reduced
primary phantom dose in PCT (when placed behind
the primary phantom relative to the incident
beam). The XRCT simulation measured intensity
transmission the PCT simulation measured both
intensity and energy transmission. The results
demonstrated a 7-fold advantage for PCT in image
contrast-to-noise ratio for the same mean
absorbed dose of 4 cGy to the subject. These
results are in accord with recent reports of
experimental and simulated findings from the Paul
Scherer Institute (PSI), Switzerland 1,2
INTRODUCTION Radiation therapy is used to treat
some types of cancers, and facilities such as the
Loma Linda Proton Treatment Center and Paul
Scherer Institute (PSI) 1 have active clinical
proton radiation therapy programs. The primary
objective of any radiation therapy is to deliver
a high dose of radiation to the tumor, while
limiting the radiation dose to healthy tissue.
Protons deliver their radiation dose in a highly
localized region compared to broad-spectrum
dosing from photon (X-ray) therapies. However,
treating many cancers with proton therapy is
potentially suboptimal due to insufficient
knowledge of electron density distribution in the
patient, a crucial requirement for accurate
treatment planning. Here we evaluate simulations
of contrast to-noise ratio (CNR) in proton
computed tomography (PCT) compared to that in
X-ray tomography (XRCT) for the same mean subject
dose of 4 cGy. The rationale for the present
work is the large slope of the dose versus depth
curve (solid red in Fig. 1 for monochromatic
protons) at the distal edge of the Bragg peak,
producing large image sensitivity to density
variations. Experimental results from PSI show
that planar proton radiography on simple phantoms
produces an order of magnitude lower radiation
dose than planar x-ray radiography for the same
CNR. X-Ray and Proton Dose vs Depth in
Water Figure 1 Normalized dose vs
penetration depth in water for 2 MeV X-ray
photons and 185 MeV protons. Note the reduced
dose and sharp Bragg edge of the proton dose
distribution. Low-energy photons (60 keV used in
present work) produce an even larger surface dose
than that pictured. In PCT, the proton beam
energy can be set so the falling edge of the
Bragg peak lies on a distal detector plane. Small
variations in intercepted tissue density then
cause large changes in detector dose due to the
large slope at the Bragg edge this advantage can
either be used to reduce the dose for a given
CNR, or improve CNR over that provided by XRCT
for a given subject dose. SIMULATION METHODS The
phantom The phantom for both PCT and XRCT was a
water cylinder (density of 1.00 g/ml) of 20 cm
diameter which included a paraxial water cylinder
of 2 cm diameter with a density of 1.01 g/ml
positioned 7 cm off axis (Fig. 2). A bow-tie
water
PCT Simulation Geometry Figure 2 The PCT
simulation geometry. A 20 cm diameter water
cylinder (r1) contains a slightly denser 2 cm
cylindrical tumor. The water-density bow-tie is
used to provide a near-constant penetration
depth, maintaing the Bragg edge on the detector
and minimizing dose to the phantom. phantom is
positioned either in front of (for XRCT) or
behind (for PCT, see Fig. 2) the phantom to serve
the following purposes a) to equalize the
radiation absorbed dose throughout the primary
phantom, b) to equalize the signal in the
detector at different positions of the beam
across the phantom, and c) most importantly for
PCT, to produce a square subject from the
combination of the bow-tie phantom and the
phantom, so that the position of the Bragg peak
will be always at the exit end of the bow-tie
phantom at different beam positions when using a
constant incident beam energy of 167 MeV. The
distal bow tie position for PCT also serves to
reduce the total mean dose to the primary
phantom. General imaging parameters. Both
PCT and XRCT beams were pencil beams 1 mm wide
(laterally) and 2 mm high (axially). Computer
tomography projections were acquired in a
translational/rotational geometry, with a 1-mm
translation step and 2o steps over 360o.
Tomographic images were reconstructed using a
standard filtered back-projection method. PCT
simulation CT projections for PCT were
generated using in-house Monte Carlo proton
simulations that included Bethe-Bloch energy
loss, energy straggling, and multiple Coulomb
scattering. No inelastic-scattering calculations
were included. The incident beam energy was tuned
to position the center of the Bragg peak at the
exit end of the bow-tie phantom, i.e., half the
Bragg peak counts reached the detector. The
measured parameters were proportional energy and
intensity loss projections, as shown in Figures
3a and 3b. Each scan step assumed a separate
beam pulse, i.e., it was registered independently
in the detector. No trajectory reconstruction
was used to produce high spatial resolution, but
1mm is achievable according to the results
described in 1. XRCT simulation The XRCT
simulations used a monochromatic incident 60 keV
photon beam. The simulations were not Monte
Carlo, but used deterministic attenuation
calculations to produce CT projections. Synthetic
random noise, consistent with the desired mean
absorbed target dose of 4cGy, was then added to
the individual projections.
Sample Proton Energy/Intensity Transmission
Figures 3a and 3b Proportional proton
energy and intensity transmission projections
through a single angle scan in simulated PCT,
using the geometry of Fig. 2 and an incident 167
MeV p beam of 200 protons per 1mm transverse scan
pixel. RESULTS AND DISCUSSION PCT and XRCT
results are shown in Figures 4a and 4b
respectively. The mean subject dose to the
primary phantom in the PCT simulations were about
4 cGy. In the XRCT results, the image noise was
adjusted to be equivalent of that obtained in 4
cGy. Analysis of thse measurements showed a
7-fold larger CNR for the PCT image for the same
mean absorbed dose. The results indicate the
potential of PCT in medical imaging. This method
can be used not only in producing accurate
electron-density images for treatment planning in
proton therapy, but also in imaging brains and
other targets in which high image contrast is
required within the limitations of the mean
absorbed radiation dose to the subject. PCT and
XRCT Reconstructed Images Figures 4a and 4b
PCT and XRCT image reconstructions from filtered
backprojection. Image reconstruction for the PCT
case was performed with energy transmission
profiles as shown in Fig. 3a. ACKNOWLEDGMENTS This
research was funded by the United States
Department of Energy. We thank Adam Rusek,
Joanne Beebe-Wang, Uwe Schneider, and Eros
Pedroni for invaluable discussions. REFERENCES 1
Uwe Schneider and Eros Pedroni, Proton
Radiography as a Tool for Quality Control in
Proton Therapy, Med Phys Vol 22 Issue 4, p. 353
(April 1995) 2 Uwe Schneider and Alexander
Tourovsky, Range-Uncertainty Imaging for
Obtaining Dose Perturbations in Proton Therapy,
IEEE Trans Nuc Sci Vol 45 No 5, p. 2309 (October
1998)