Yelin Suh and Paul Keall Stanford University - GSI

Yelin Suh and Paul Keall
Stanford University
4D Treatment Planning Workshop at GSI
December 9, 2010

Rationale for 4D Radiotherapy
4D Treatment with Dynamic MLC Tracking
4D Radiotherapy Treatment Planning
Specific Solutions
Towards General Solution
Delivery of 4D Treatment Plans

Motion in radiotherapy
“Respiration causes target motion, which is known to be one of the technical
bottlenecks in radiotherapy, especially for stereotactic radiosurgery and IMRT.”
Hypothesis
Suh et al. “Aperture maneuver with compelled breath (AMC) for moving tumors:
A feasibility study with a moving phantom,” Med Phys 31: 760-6, 2004
Instead of seeing anatomic motion as an obstacle to radiotherapy,
utilizing motion has potential to improve radiotherapy treatment.
Including motion as an additional degree of freedom in radiotherapy
treatment may yield at least as good solutions with motion tracking
as those with no motion (motion = 0 is a subset of motion > 0).

Highly constrained treatment plan
A 4D CT planning scan of
one lung cancer patient with
a small (3 cm 3 ), mobile tumor (2.1 cm)
D 5% 36-40% ↓ D 5% 29-35% ↓

3D
4D
Planning Delivery
3D CT
Leaf Sequence for
a 3D Treatment Plan
L ( MU )
4D CT
Leaf Sequence for
a 4D Treatment Plan
L ( MU, θ )
3D Plan
Real-time 3D Target
Position
( x, y, z )
Deliver in Treatment Machine
Available technology Published (Sawant et al., 2008)
Published (Suh et al.,
2009; Gui et al., 2010)
4D Plan
Real-time 3D Target
Position and Phase
( x, y, z | θ )
Deliver in Treatment Machine

4D IMRT treatment planning with dynamic MLC tracking
including MLC leaf motion constraints, which takes
respiratory motion into account and is robust to the
variations of fractional time spent in respiratory phases
within a given 4D CT planning scan
4D Radiotherapy
… is the explicit inclusion of the
temporal changes in anatomy
during the imaging, planning,
and delivery of radiotherapy …
ASTRO Meeting Panel Discussion, 2004
Requires
“Respiratory motion
WILL change
between imaging
and delivery, and
between treatment
fractions.”

4D IMRT treatment planning with dynamic MLC tracking
including MLC leaf motion constraints, which takes
respiratory motion into account and is robust to the
variations of fractional time spent in respiratory phases
within a given 4D CT planning scan
4D Radiotherapy
1. Imaging
4D CT, as the best estimate
2. Planning
Treatment-planning optimization
robust to variable motion
3. Delivery
Treatment-delivery system flexible
enough to account for variable motion
Requires
“Respiratory motion
WILL change
between imaging
and delivery, and
between treatment
fractions.”

Ideal??
Deliverable!!
Extend 3D IMRT treatment planning to 4D
with deliverable constraints

4D treatment planning that accounts for 3D target motion
by applying real-time dynamic MLC motion-tracking algorithm
Suh et al. “4D IMRT treatment planning using a dynamic MLC
motion-tracking algorithm,” Phys Med Biol 54: 3821-35, 2009
* Accounting for 1D motion along the major axis was
previously published.
Suh et al. “A deliverable 4D IMRT planning method
for dynamic MLC tumor tracking delivery,”
Int J Radiat Oncol Biol Phys 71: 1526-36, 2008
Robust to the variations of fractional time spent in phases within 4D CT
Not optimal, but deliverable with currently available technology
Clear path to clinical implementation by using the same algorithm
between planning and delivery for determining MLC leaf sequences

Integration of:
General-purpose optimization system
to optimize MLC leaf positions in L
for individual respiratory phases
Commercially available TP system
to compute individual phase doses
and a deformable-summed 4D dose,
and Composite Objective Value,
on the basis of L and H
* SNOPT, Sparse Nonlinear OPTimizer (Stanford Systems Optimization Lab)

Solving various approaches within a common framework
Solve general solution
Framework for solving
general solution
Limiting
degrees of freedom
Solve more
constrained solutions
Assess clinical benefit of the general vs. more constrained solutions
to 4D IMRT treatment planning with dynamic MLC tracking for
radiotherapy treatment of thoracic and abdominal tumors

Inhale Phase
Dashed: after 3D optimization
Solid: after 4D optimization
4D Plan
Exhale Phase

Inhale Phase
Dashed: after 3D optimization
Solid: after 4D optimization
4D Plan
Exhale Phase

Apply the solutions to 4D CT scans of a cohort of lung cancer patients
Perform treatment planning using the solutions for various
4D treatment scenarios developed, on 4D CT planning scans
of lung cancer patients to estimate the impact of 4D
treatment planning for the population of lung patients
Quantify the clinical benefit of 4D treatment planning
Compare and evaluate treatment plans generated using
various methods, in terms of clinical significance and
statistical significance on tumor dose escalation and healthy
tissue dose reduction, and impact of the proposed work on
the improvement of IMRT treatment of lung cancer

To experimentally investigate 4D treatment delivery
to account for motion, rotation, and deformation of
tumors and normal tissues
Hypothesis
Delivering 4D plans to account for the discrepancy
between motion quantified from a 4D CT planning scan
and that observed during each treatment delivery
further improves the accuracy of radiotherapy treatment,
as anatomy changes between imaging and delivery and
between treatment delivery fractions.

For lung or liver cancer patients with implanted markers:
Imaging 4D CT planning scan
Planning where motion, rotation, and deformation of
tumors and normal tissues are included
Delivery where target motion different from that in the 4D plan
is accounted for, using real-time target localization
Requirements:
(1) treatment‐delivery system flexible enough to account for
variable target motion during treatment delivery
(2) real‐time information on both 3D position and phase of
target motion from either measurement or estimation

4D
Planning Delivery
4D CT
Leaf Sequence for
a 4D Treatment Plan
L ( MU, θ )
4D Plan
Real-time 3D Target
Position and Phase
( x, y, z | θ )
Deliver in Treatment Machine
3D target position and respiratory phase information
quantified from
a 4D CT planning scan
Discrepancy
incoming from
a position monitoring
system in real time

4D delivery when
target motion during delivery is
1) same as (3 cm)
2) smaller than (1.5 cm)
3) larger than (4 cm)
motion during imaging (3 cm)
Static delivery of
each of individual phase plans
for comparison

Static delivery
Same motion during delivery as imaging
Smaller motion during delivery than imaging
Larger motion during delivery than imaging

Difference of ellipses
between 4D and static delivery
Translation/Motion Hysteresis
Center X position 2.3 mm (0.1-5.1)
Center Y position 2.1 mm (1.5-10.1)
Rotation
Orientation 1.3 o (0.2-4.0)
Deformation
Major axis length 0.6 mm (0-1.6)
Minor axis length 0.7 mm (0.3-1.7)

Using motion as a degree of freedom for dynamic MLC tracking can
improve radiotherapy treatment for 4D anatomy.
Specific and general solutions to the 4D treatment planning with
dynamic MLC tracking problem have been investigated.
4D treatment plans for dynamic MLC tracking result in leaf sequences
as a function of monitor units and respiratory phases.
4D optimization with dynamic MLC tracking can generate a better 4D
treatment plan than the sum of individually optimized phase plans.
4D treatment plans can be delivered with current delivery technology,
while accounting for variable target motion during delivery.

Acknowledgements
- NIH/NCI grants R01 CA 93626 and P01 CA 116602
- Philips Medical Systems
- Varian Medical Systems