Direct Segment Optimization Optimizing conformal plans without IMRT
Jennifer M. SteersJennifer M. Steers1,21,2 , Martha M. Matuszak , Martha M. Matuszak1,21,2 , Benedick A. Fraass , Benedick A. Fraass11
Departments of Radiation OncologyDepartments of Radiation Oncology11 and and
Nuclear Engineering & Radiological SciencesNuclear Engineering & Radiological Sciences22
University of Michigan, Ann Arbor, MichiganUniversity of Michigan, Ann Arbor, Michigan
Outline
• Introduction– What is DSO?– Why are in we interested in DSO?– Goal
• Methods and Materials• Results– IMRT vs. DSO comparisons
• Conclusions
• Direct segment optimization (DSO)– form of direct aperture optimization (DAO) but
based on flat fields, not on beamlet distribution
• Utilizes user-defined cost functions to optimize the following:– Beam weights– MLC positions
What is DSO?
What is DSO?
Why are we interested in DSO?
• Plans are optimized conformal plans – Can reduce delivery time over IMRT– Can result in fewer MUs when compared to
IMRT
• Plans do not require IMRT QA - plans could be started or adapted much quicker• It may make tweaking and optimizing leaf
positions in conformal plans, such as SBRT, much quicker
Goal
• Can DSO produce simpler plans comparable in quality to IMRT plans with the same beam angles and cost function?
Methods and Materials:Features of DSO
• MLCs and beam weights can be optimized separately or together
• Search strategy options– Ordered and random searches– Step sizes
• DSO offers fewer degrees of freedom per beam when compared to IMRT– User must create segments in a plan before
optimizing
IMRT vs. DSO Setup
• Planning goal: Minimize dose to OARs and normal tissues without compromising target uniformity
• Same gantry angles were used between the IMRT and DSO plans except when needed segments were added to the DSO case
• The same cost function was used to optimize both the IMRT and DSO plans
Evaluation of Comparisons
• Both plans were optimized and evaluated with the following metrics– DVHs – MU/fx and beam-on time/fx– Mean doses to structures– Max structure doses (to 0.5cc or 0.1 cc)
– D95 (for PTVs)
– 3D dose comparisons
Brain Planning Goals
• Target : 60 Gy (Min: 59, max: 61)• Normal brain: minimize dose (threshold 0,
power 2)• Optic structures: minimize dose (threshold
0, power 2)
7-Field, Non-coplanar Plan – Brain11
Normal Brain
PTV
Normal Brain
Chiasm
Maximum Doses (Gy) * = Max dose to 0.5 cc; ** = Max dose to 0.1 cc
PTV* Chiasm** R Eye** L Eye** Normal Brain*
DSO 63.5 15.1 1.0 1.0 63.4
1x1 Beamlets 69.2 17.9 1.0 1.0 62.9
Mean Doses (Gy)
PTV Chiasm R Eye L EyeNormal Brain
DSO 59.67 8.52 0.59 0.73 25.60
1x1 Beamlets 59.71 8.53 0.60 0.63 24.56
Brain11
PTV D95 (Gy)
DSO 56.1
1x1 Beamlets 56.2
Plan Comparison Brain16
Brain11
MU/Fx SegmentsEstimated beam-on
time/Fx (min)
DSO 237 7 0.4
1x1 Beamlets 845 351 6.6
1x1 beamlet plan DSO plan
IMRT vs. DSO
Brain11
1x1 beamlet plan DSO plan
IMRT vs. DSO
Brain11
Lung Planning Goals
• Using adaptive protocol (2007-123)• PTV: 85 Gy (Min: 84 Gy, max: 86 Gy)
• Esophagus: NTCP < 47% (Veff = 33%)
• Heart: NTCP < 5%• Normal Lung: NTCP < 17.2%• All normal structures: minimize dose
(threshold 0, power 2)
7-field, Non-coplanar Plan – Lung2-123
Heart Cord
Esophagus
PTV
Heart Cord
Esophagus
PTV
Maximum Doses (Gy) * = Max dose to 0.5 cc
PTV* Cord* Esophagus* Heart* Lung-P2*
DSO 90.4 44.0 83.6 88.1 89.3
1x1 Beamlets 89.4 35.4 82.8 88.9 91.5
Mean Doses (Gy)
PTV Cord Esophagus Heart Lung-P2
DSO 84.2 3.7 14.0 14.0 11.1
1x1 Beamlets 84.4 5.0 15.1 15.8 12.6
Lung2-123
PTV D95 (Gy)
DSO 78.5
1x1 Beamlets 80.7
Plan ComparisonLung2-123
Lung2-123
MU/Fx SegmentsEstimated beam-on
time/Fx (min)
DSO 492 14 0.8
1x1 Beamlets 680 300 5.3
5-field Axial plan – Liver4
Cord
Normal Liver
PTV
Cord
Normal Liver
PTV
Cord
Normal Liver
PTV
Maximum Doses (Gy) * = Max dose to 0.5 cc
PTV* Normal Liver* R Kidney* L Kidney* Cord*
DSO – 5 Segments 94.2 92.9 2.0 1.3 15.5
DSO – 7 Segments 92.7 92.9 1.8 1.4 12.8
1x1 Beamlets 92.2 95.1 1.5 1.0 11.3
Mean Doses (Gy)
PTVNormal
LiverR Kidney L Kidney Cord
DSO – 5 Segments 89.9 29.4 1.1 0.7 2.1
DSO – 7 Segments 90.1 28.3 1.1 0.7 2.0
1x1 Beamlets 90.0 25.6 0.2 0.3 2.5
Liver4
PTV D95 (Gy)
DSO 86.5
1x1 Beamlets 88.5
Plan Comparison Liver4
MU/Fx SegmentsEstimated beam-on
time/Fx (min)
DSO – 5 segments
257 5 0.43
DSO – 7 segments
267 7 0.45
1x1 Beamlets 586 303 5.24
Liver4
Liver4
1x1 beamlet plan DSO plan
IMRT vs. DSO
Conclusions
• DSO can successfully optimize several types of planning scenarios while reducing treatment time
• Optimizing with DSO can produce plans comparable in quality to IMRT with less MUs and segments
• No need for the type of QA associated with IMRT plans
• Overall plan complexity is less
Conclusions
• In the future:– DSO may be more time efficient while yielding
similar quality plans as IMRT in an adaptive re-planning scenario
– Since DSO produces flat-field plans, it may be useful for cases with inter- and intra-fraction motion
– Looking at other sites that could benefit from twiddling (prostate, SBRT, pediatric)
Questions?
